Digitranscope the Governance of Digitally-Transformed Society

DigiTranScope The governance of digitally-transformed society

Joint Research Centre 2 3

Table of Contents

Abstract...... 6 Acknowledgements...... 7 Executive summary...... 9 Digitranscope: An Introduction...... 13 About Digitranscope and the Centre for Advanced Studies...... 13 About Digitranscope...... 13 The Shifting political landscape...... 15 Structure of the project and of the book...... 19 For the benefit of all!? Some notes on the role of government in the digital transformation...... 23 Introduction...... 23 Three dominant strategies...... 25 A European way?...... 28 Action perspectives...... 32 Conclusion and closing remarks...... 36 Data governance models emerging from the practices of social actors...... 41 Introduction...... 41 Data governance: A social science-informed definition...... 42 Research Strategy...... 44 Emerging data governance models...... 45 Discussion...... 51 Seeing Platform and Data Co-operatives Through European Pandemic Citizenship...... 59 Introduction: European Pandemic Citizenship at Stake...... 59 Rationale: Co-operatives as a Collective Resilient Response to the Pandemic Crisis...... 62 Discussion: Co-operatives and Pandemic Citizenship...... 64 4 Table of Contents Table of Contents 5

Taxonomy: Shedding Light on Platform and Data Co-operatives ...... 66 Digitranscope Experiments: Digital Twins and Smart Final Remarks ...... 70 Cities, case studies of Amsterdam and Duisburg ...... 135 Citizen-generated data for policy: Introduction ...... 135 A review of EU projects ...... 77 Amsterdam case study ...... 136 Introduction ...... 77 Duisburg case study ...... 140 The key-features of European CGD projects ...... 81 Findings ...... 143 Findings from interviews ...... 84 Involving children in the renewable energy transition: Discussion ...... 86 exploring the potential of the digital transformation ...... 149 Conclusion ...... 90 Introduction ...... 149 Commercial sector data for the public interest? The Warsaw Experiment ...... 152 A qualitative research on data sharing practices in EU cities ...... 95 The Amsterdam Experiment ...... 155 Introduction ...... 95 Using gaming in digital twin setting: opportunities and limitations ...... 159 How local administrations access to private sector data ...... 95 Conclusions ...... 167 Methodology ...... 97 Results ...... 99 List of boxes ...... 174 Discussion ...... 106 List of figures ...... 174 Probabilistic Synthetic Population modelling List of tables ...... 175 for EU policy support ...... 111 About the Authors ...... 176 Introduction ...... 111 Review of Methods ...... 112 Applications ...... 114 Conclusions ...... 119 Semantic Text Analysis Tool (SeTA) ...... 125 Introduction: What is SeTA and why it is important ...... 125 The technology ...... 127 A different approach to summarising text ...... 131 Stepping into the future ...... 132 Conclusions ...... 133 6 7

Abstract Acknowledgements

This volume presents the key outcomes and research findings of the Dig- The authors are particularly grateful to Mr. Vladimir Sucha, former Director itranscope research project of the European Commission Joint Research Centre. General of the European Commission Joint Research Centre for believing in The project set out to explore during the period 2017-2020 the challenges and the scope of the project and his encouragement in setting it up. We are also opportunities that the digital transformation is posing to the governance of grateful to Jutta Thielen-Pozo, head of the Scientific Development Unit, and all society. We focused our attention on the governance of data as a key aspect to her wonderful colleagues for her leadership and support, and to Alessandro understand and shape the governance of society. Data is a key resource in the Annoni, former head of the Digital Economy Unit for his help in connecting to digital economy, and control over the way it is generated, collected, aggregated, policy colleagues and transitioning the activities of Digitranscope into his Unit and value is extracted and distributed in society is crucial. We have explored to develop further in the years to come. the increasing awareness about the strategic importance of data and emerging A special thanks goes to Prof. Michael Blakemore for his important contribution governance models to distribute the value generated more equitably in soci- in the formative year of the project, and Prof. Peter Lloyd for his constant curi- ety. These findings have contributed to the new policy orientation in Europe osity, engagement, and insights. on technological and data sovereignty and the sharing of data for the public Finally, the authors would like to thank all the participants to the Digitranscope interest. The digital transformation, the rise of artificial intelligence and the workshops for their precious contributions to the project. Internet of Things offer also new opportunities for new forms of policy design, implementation, and assessment providing more personalised support to those Authors: Massimo Craglia, European Commission Joint Research Centre who need it and being more participative throughout the policy cycle. The use of Henk Scholten, Vrije Universiteit Amsterdam and GEODAN digital twins, gaming, simulation, and synthetic data are just at their beginning Marina Micheli, European Commission Joint Research Centre but promise to change radically the relationships among all the stakeholders in Jiri Hradec, European Commission Joint Research Centre governance of our society. Igor Calzada, European Commission Joint Research Centre and University of Oxford Steven Luitjens, Ministry of the Interior and Kingdom Relations, Government of the Netherlands Marisa Ponti, University of Gothenburg Jaap Boter, Vrije Universiteit Amsterdam

Contributors: Anna Berti Suman, Dutch Research Council and Joint Research Centre Margherita Di Leo, Arcadia SIT srl. Corentin Kusters, GEODAN Magdalena Grus, Vrije Universiteit Amsterdam Sanne Hettinga, Vrije Universiteit Amsterdam 8 9

Executive summary

This volume presents the key outcomes and research findings of the Dig- itranscope research project of the European Commission Joint Research Centre. The project set out to explore during the period 2017-2020 the challenges and opportunities that the digital transformation is posing to the governance of society.

Policy context The development of the project coincided with an increasing recognition of the importance of Artificial Intelligence (AI) to master the increasing volumes of big data available on a daily basis. The control of AI and of the data underpinning its development are strategic for the future development of society, and the focus of an increasing geopolitical competition. The European Union has identified technological and data sovereignty as key priorities for Europe and developed several policy initiatives to strengthen its regulatory framework and increase its preparedness to address the dual digital and green deal transformations.

Key conclusions There are many dimensions to address the governance of a digitally-transformed society and the project focussed on the governance of data as a critical aspect. Data is a key resource in the digital economy, and control over the way it is generated, collected, aggregated, and value is extracted and distributed in society is crucial. We have explored the increasing awareness about the strategic importance of data and emerging governance models to distribute the value generated more equitably in society. These findings have contributed to the new policy orientation in Europe on technological and data sovereignty and social inclusion.

At the same time, the digital transformation, and the rise of artificial intelligence and the Internet of Things, offer also new opportunities for new forms 10 Executive summary Executive summary 11

of policy design, implementation, and society. Data cooperatives are dem- adopting different operational modes izen-centred policies, targeted to the citizens of today and tomorrow to assessment providing more personal- ocratic and collective forms of data and strategies, a practice that at the those who need intervention most have a say in shaping their environ- ised support to those who need it and governance in which data subjects present time is still challenging and without the use of personal data. We ment. We conclude that technological being more participative throughout voluntarily pool their data together precarious for most cities. The vast found that the opportunities are very change is much faster than the ability the policy cycle. The use of digital to create a common pool for mutual majority of use cases examined for significant, and for this reason many of governments to react. Therefore, it twins, gaming, simulation, and syn- benefits. We examined how they re- the chapters in this section consits in governments and statistical agen- is necessary to anticipate and shape thetic data are just at their beginning late to the notion of platform coop- niche initiatives or pilot projects. The cies are becoming interested in this the future direction of development but promise to change radically the erativsm and explained why they are scaling up of the relative data gov- methodology. The concept of digital through foresight studies, qualita- relationships among all the stake- gaining relevance in current forms of ernance models in the future depends twins has been known and applied for tive research, and experimenting with holders in governance of our society. European pandemic citizenship. We on the ad-hoc policy measures that many years in manufacturing, creat- new technology and methods, rather researched also EU projects based on will be established to support them. ing a digital replica of an artefact for than trying to fix the present that too Main findings citizen-generated data (CGD), intend- The chapters provide conceptual tools testing and assessment before going quickly becomes the past. Govern- With respect to the governance of ed as data that people or their organ- for a thoughtful discussion on the into production. The increased avail- ments play a key role, but it is ulti- digital data, we examined data shar- isations produce to directly monitor, approaches for accessing and shar- ability of data, and processing pow- mately up to all of us to shape our ing and control as a socio-technical demand or drive change on issues ing data that foster a more equitable er at declining costs, makes it now futures. practice. We analyzed four emerging that affect them. The growth of cit- digitally transformed society. possible to develop digital twins for models of data governance - data izen-generated data give the public entire cities and nations. We discuss sharing pools, data cooperatives, sector more opportunities for ad- With respect to the governance with the use of digital twins in Amster- public data trusts, and personal data dressing critical social and economic the digital transformation we have dam and Duisburg to address local sovereignty - and inquired to what issues, at the same time offering new explored the use of synthetic pop- problems and found them effective extent they support different, more avenues for active citizenships and ulations, digital twins, and gaming tools to communicate with all the balanced, power-relations between reshaping the relationships between environments. The development of stakeholders involved from govern- actors compared to the dominant one citizens and local governments. Fi- synthetic populations through AI and ment officials to business and the of datification (in which few dominant nally, also local governments could machine learning methods results in public. We tested also the combined corporate actors get most of the val- directly help to redistribute the val- an artificial set of individuals, fami- use of digital twins and gaming en- ue). Data cooperatives and civic data ue of data actors across society. We lies, and households with the same vironments to engage school children trusts, in particular, are established to explored how European municipali- characteristics and behaviour of the in the energy transition and urban redistribute the value generated from ties are getting access to commer- true population. This allows the de- planning, and found this combination personal data more equitably across cial sector data of public interest sign, modelling and testing of cit- as having many opportunities to get 12 13

Massimo Craglia and Henk Scholten Digitranscope: An Introduction

About Digitranscope and the Centre for Advanced Studies Digitranscope is a three-year research project of the Centre for Advanced Stud- ies at the European Commission Joint Research Centre. The Centre for Advanced Studies (CAS) of the European Commission’s Joint Research Centre (JRC) was created in 2016 to improve the connections between science and policy and help inform better and influence the regulatory frameworks needed to address the new and emerging societal challenges confronting the European Union and our societies as a whole. The CAS aims at creating the conditions necessary for innovative and interdisciplinary research, as well as offering a creative and generative space in which ideas and knowledge in emerging thematic fields across different scientific and technological disciplines can thrive and flourish. As such, the CAS has become an incubator for formal inquiry, stimulating ideas and activities and providing the JRC with new insights, data projections and solutions for the increasingly complex medium and long-term challenges facing the EU, especially in the fields of artificial intelligence, demography, big data and digital transformation.

About Digitranscope Digitranscope originated from the JRC Strategy 2030. The strategy identified ten strategic topics on which the JRC should concentrate to anticipate future policy requests. One of these topics was Data and Digital Transformation to address which the JRC set up a transversal project on Artificial Intelligence and Digital Transformation led by the Digital Economy Unit, as well as a research project, more exploratory in nature, on digital transformation to be implemented in the CAS. The project originally proposed to address two key issues: i) how the information glut triggered by the digital transformation reverses the cognitive balance between humans and machines, and ii) the impact of digital information technology on the rules and institutions that guide modern societies. The proposal led to the establishment in 2017 of two projects: Humaint on 14 Digitranscope: An Introduction Digitranscope: An Introduction 15

At the early“ stages of the project, we recognised human behaviour and machine intel- At the early stages of the project, that the governance of The Shifting political landscape the technology it is necessary also to mercial sector data, and a Regulation ligence and Digitranscope, on digital we recognised that the governance digitally transformed On Data Governance: govern better European data. Techno- on the governance of data (Data Gov- transformation and the governance of digitally transformed societies The most significant event that oc- logical and data sovereignty became ernance Act2) launched in December societies revolves to a of human society. revolves to a large extent around curred during the lifetime of the pro- key objectives of the new Commission 2020. The latter extends the catego- . the governance of data: those who large extent around the ject was the emergence of Artificial which took office in November 2019. ries of public sector data available for The objectives of Digitranscope are: control the production, integration, governance of data: those Intelligence (AI) as a key geopolitical reuse, creates the framework for the ★ to explore the changing flows, use and dissemination of data have battleground, particularly between As a result of this increased atten- sharing of business data, and facili- who control the produc- ownership, quality and implica- formidable levers of power in to- the US and China. This brought AI also tion to data, the Commission stepped tates the reuse of personal data via tions of digitised data and infor- day’s digitised society. What we did tion, integration, use and at the forefront of the European polit- up its work on Business-to-business data intermediaries or on altruistic mation; not realise at the start of the project dissemination of data ical attention with an initial strategy (B2B) and Business-to-government grounds. ★ to identify the key policy chal- is that what we thought was going have formidable levers of on AI adopted by the Commission in (B2G) data sharing that had start- From data sharing to data analytics: lenges relating to massive inter- to be an exploratory project looking April 2018 (EC. 2018a), followed by ed in 2017 with the Communication It is important to note that the evo- power in today’s digitised connection (the Internet of Things 5-10 years ahead turned instead a Coordinated Plan with the Member ‘Building a European Data Economy’ lution of technology, and the les- - IOT) and the associated opportu- into one providing already direct in- society. States in December 2018 (EC, 2018b), (EC, 2017) and then the COM Towards sons learned from the big commer- nities and risks; put to policy as policy priorities shift- the establishment on a High-Level a Common Data Space (EC, 2018c). cial platforms, has brought also a ★ to determine what skills are needed much faster than we anticipated. ” Expert Group1 to advise on the de- These Communications had provided change not just in what is shared but ed to live fulfilling and healthy We highlight some of these shifts in velopment of ethical guidelines for AI a set of guidelines but now it seemed how data is shared. In the past 20 lives in a digitally transformed the next section. and priority areas for investment in appropriate to develop them further years data publication for reuse, via society, and to explore how to of- 2019, and an AI White Paper in Feb- into a regulatory framework. The catalogues and portals, was seen fer all citizens the opportunity to ruary 2020 (EC, 2020a) setting the Commission therefore, organised sev- as the end of the process of data develop these skills, and; framework for a consultation on a eral workshops and studies on data collection, analysis and use by the ★ to explore innovative forms of risk-based regulatory framework for governance which contributed to the (public sector) data custodian. It was governance for Europe leverag- AI. Why is this important? Because it European Strategy for Data published often also perceived as a burden being the characteristics of digital immediately became clear that data (EC, 2020b) in February 2020 estab- cause the organisation publishing transformation. is the absolutely key asset underpin- lishing several European common the data was not a direct beneficiary ning the development of AI, and that data spaces in different thematic of the value subsequently generated to govern the future development of domains (e.g. environment, health, agriculture, automotive, finance, etc.) 2 https://ec.europa.eu/digital-single-market/en/news/ 1 https://ec.europa.eu/digital-single-market/en/high- proposal-regulation-european-data-governance- level-expert-group-artificial-intelligence including both public sector and com- data-governance-act 16 Digitranscope: An Introduction Digitranscope: An Introduction 17

We have seen“ a rapid development of digital by third parties or accrued by society From data analytics to digital twins: twins also for urban With the much-increased availability as a whole in terms of greater trans- There are many definitions of digital areas, and in of high-resolution data from space parency and accountability. Observ- twins but a generic one is provided by and airborne instruments, sensor net- environmental ing the big data platforms at work El Saddik (2018) as follows: A digital works, public administrations, and the it became noticeable that to them twin is a digital replica of a living or applications. general public, we have seen a rapid data publication was the beginning non-living physical entity. By bridg- development of digital twins also for of the value-creation chain, not the ing the physical and the virtual world, ” urban areas, and in environmental end! In fact, social media platforms data is transmitted seamlessly allow- applications. A survey of digital twins and search engines, do not even cre- ing the virtual entity to exist simulta- in the environmental domain by Nati- ate the original data, they let the neously with the physical entity. vi, Craglia and Delipetrev (2020) indi- users do so. They then integrate Digital twins have been used in in- cates that indeed urban management the users’ data, add value through dustry for several decades, largely as and earth system modelling are the analytics, repackage into products an extension of computer-aided de- two more prominent areas. European or services, and sell to third parties sign. They allow simulation and test- policy supports these developments thus monetizing the added value ing of artifacts before moving into with new initiatives to develop a created. This shift in the datafication production. With the development of smart cities’ ecosystem3 and Desti- paradigm (Mayer-Schoenberger and Industry 4.0 and the vast increase in nation Earth4 aiming at developing Cukier, 2014; Ericsson, 2014) has led sensors networks and computer pro- a very high-precision digital model to an increasing call for public au- cessing digital twins they have seen a of the Earth to monitor and simulate thorities to also shift from publishing significant growth in every sector, as natural and human activity, and sup- datasets in portals to open access show in this classification by ISO. port European environmental policies. via machine-to-machine Application With the deployment of 5G networks Protocol Interfaces (APIs) interfaces and the diffusion of the Internet of and add value to the data they pub- Things we are likely to see a step lish by adding the intelligence via change in the diffusion of the digital analytics on who uses the data for Table 1.1. Industries and applications of Digital Twins (Source; Nativi, Craglia, Delipetrev, 2020) twin paradigm, which represent the what purpose (Vaccari et al. 2020). 3 https://ec.europa.eu/digital-single-market/en/smartcities-smart-living 4 https://ec.europa.eu/digital-single-market/en/destination-earth-destine 18 Digitranscope: An Introduction Digitranscope: An Introduction 19

Digitranscope“ project on two main tracks. evolution of the big data analytics analytics into a single (virtual) plat- tegrates simulation into the data The governance of digital On new forms of policy design: Structure of the project and of discussed earlier. This evolution is de- form. The development of the IoT and processing and analytics platforms data including the role of Significant policy shifts have also the book picted in Fig 1 below. As shown, the edge computing brings analytics and strengthening the move towards dy- emerged in this area during the last Against the background of the rapid- government in emerging shift from traditional data processing processing already at the level of namic and interactive environments few years and become increasingly ly-evolving technological and policy to big data already required a closer the sensor collecting the data while based on data streams and feed- models of data mainstream. Notably, the increasing landscape highlighted above, we de- integration of data processing and the development of digital twins in- backs-loops (see Nativi et al. 2020) governance. use of big data analytics to profile and cided to structure the Digitranscope nudge voters following the example project on two main tracks. New forms of governance of the commercial sector recognised ★ The first track investigated issues with digital data including not only the power of data but also around the governance of digital experimenting with the the emotional side of decision mak- data including the role of govern- use of publicly available ing. The mantra of evidence-based ment in emerging models of data decision-making that was all the rage governance; citizen-generated data for profiling and the in the 1990s has come under increas- data for public policy; citizenship design of policies targeting scrutiny together with the scien- and data cooperatives, and the ed to specific needs. tific method when applied to social perspectives of city governments and political phenomena (Funkowitz in accessing and using data held et al. 2016). We have seen therefore by the commercial sector. ” a greater acknowledgement of the ★ The second track investigated multi-faceted dimensions of rational- new forms of governance with ity, decision-making, and post-normal digital data including experimen science. Communication, participation, ting with the use of publicly avail- and the use of narratives have gained able data for profiling and the currency exploiting also the new op- design of policies targeted to portunities of the digital transition, specific needs and groups, and from the booming of citizen-generat- used this in the context of energy Figure 1.1: Evolutionary landscape in data analytics (Source: Nativi S. and Craglia M., 2020) ed content for science and policy to transition and the risk of infec- the development of digital twins for tion in the COVID-19 crisis. We policy simulation, co-creation, and have organized experiments to communication. involve children in energy transi- 20 Digitranscope: An Introduction Digitranscope: An Introduction 21

tion through digital twins in con- policy modelling by Jiri Hradec and References European Economic and Social Com- Mayer-Schönberger, V., & Cukier, K. trolled gaming environments. We colleagues, and the use of semantic El Saddik A. Digital Twins: The Conver- mittee and the Committee of the Re- Learning with Big Data; The Future used the emerging lessons from text analysis for policy assessment gence of Multimedia Technologies IEEE gions. Towards a Common Data Space. of Education. Eamon Dolan/Houghton the deployment of the Internet also by Jiri Hradec, followed by the Multimedia 25(2), 2018. https://ieeex- COM/2018/232 final. Publications Office: Mifflin Harcourt. 2014 of Things and digital twins for reports of two experiments carried plore.ieee.org/document/8424832 Luxembourg. https://ec.europa.eu/trans- Nativi S. Craglia M. & Delipetrev B. “smart” cities to develop a City Op- out in the project, one on the case Ericsson. The impact of datafication parency/regdoc/rep/1/2018/EN/COM- Destination Earth: Survey on “Digital erating System and used AI meth- studies of Amsterdam’s and Duis- on strategic landscapes. Ericsson report, 2018-232-F1-EN-MAIN-PART-1.PDF Twins” technologies and activities, in ods to extract knowledge from burg’s digital twins by Coren Kusters 2014. European Commission. White Paper the Green Deal area. Publications Office policy documents and apply it in and Henk Scholten, and the other on European Commission 2017. Com- on Artificial Intelligence: A Europe- of the European Union. 2020. https:// the context of impact assessment. the use of gaming to Involve children munication from the Commission to the an Approach to Excellence and Trust. publications.jrc.ec.europa.eu/repository/ in the renewable energy transition by European Parliament, the Council, the COM(2020)65final. 2020a. https:// handle/JRC122457 The book is organised following these Jaap Boter and colleagues. We con- European Economic and Social Commit- ec.europa.eu/info/sites/info/files/com- Nativi S. and Craglia M. Destination two parts. The first part includes four clude the book with a chapter reflect- tee and the Committee of the Regions. mission-white-paper-artificial-intelli- Earth. Presentation at the Kick-Off chapters addressing the data gov- ing on the key messages and lessons Building A European Data Economy. gence-feb2020_en.pdf Meeting, March 6th 2020. ernance models emerging from the learned. COM/2017/09 final. Publications Office: European Commission. Communi- Nativi S. Kotsev A., Scudo P. et al. IoT practices of social actors by Marina Luxembourg cation from the Commission to the 2.0 and the INTERNET of TRANSFORMA- Micheli and colleagues, a deep dive Before diving into these two worlds European Commission. Communi- European Parliament, the Council, TION (Web of Things and Digital Twins): into one of these models, that of data of the policy of, and with, the digi- cation: Artificial Intelligence for Europe. the European Economic and Social A multi-facets analysis. Publications and platform cooperatives by Igor tal transformation Steven Luitjens, 2018a. https://ec.europa.eu/digital-sin- Committee and the Committee of the Office of the European Union. 2020. Calzada, followed by a review of EU a senior programme manager in gle-market/en/news/communication-ar- Regions. European Strategy for Data Vaccari L. Posada M., Boyd M., et al. projects on citizen-generated data for the Dutch government in charge of tificial-intelligence-europe COM/2020/66 final https://eur-lex. 2020. Application Programming In- policy by Marisa Ponti, and an anal- various digital transformation pro- European Commission. Coordinated europa.eu/legal-content/EN/TXT/?- terfaces in Governments: Why, what ysis of the practices of data sharing grammes, helps us to put all the Plan on Artificial Intelligence. 2018b. qid=1593073685620&uri=CELEX- and how, EUR 30227 EN, Publications between the commercial sector and contributions that follow into context https://ec.europa.eu/digital-single-mar- %3A52020DC0066 Office of the European Union, Lux- local government in twelve cities by with a personal reflection on the role ket/en/news/coordinated-plan-artifi- Funtowicz S, Ravetz J. Science for embourg. https://ec.europa.eu/jrc/en/ Marina Micheli. of government in the current digital cial-intelligence the post-normal age. Futures 31: 739– publication/eur-scientific-and-tech- transformation. European Commission 2018c. Com- 755, 1993. http://www.uu.nl/wetfilos/ nical-research-reports/applica- The second part explores the use of munication from the Commission to the wetfil10/sprekers/Funtowicz_Ravetz_Fu- tion-programming-interfaces-govern- probabilistic synthetic populations for European Parliament, the Council, the tures_1993.pdf. ments-why-what-and-how 22 For the benefit of all? 23

Steven Luitjens1 For the benefit of all!? Some notes on the role of government in the digital transformation

Introduction

In the autumn of 2014 I was invited to reflect on the findings of two U.S. con- sultants during their study tour to Silicon Valley exploring the latest trends. At the time I was head of the agency Logius, responsible for the generic digital infrastructure of the Dutch government for service delivery to citizens and companies. I was surprised to see that my strategy was very much in line with the common approach in Silicon Valley: launching new open standard software solutions, industrializing and commoditizing them and making them the de fac- to standard for all to use before anyone else does it (Wardley, 2014). My reflection resulted in adding an extra chapter to the report, sketching what Logius was doing especially in the field of the generic authentication solution DigiD. It was the only chapter that was not about private initiatives and this was probably the justification for opening the chapter with an astonishing question for me: “What role, if any, should the government have in developing, overseeing and when necessary regulating the explosive development of the Matrix?” (Mo- schella, and Mead, 2015).

In January 2019 a quite similar study tour to Silicon Valley was made by a group of Dutch civil servants. Software-as-a-service was no longer the issue, it was all about artificial intelligence and big data. One of the most intriguing findings was that by now enormous amounts of data -including location data on individuals, companies and strategic resources- were commercially for sale on an already quite mature market for everyone who is interested. When they asked whether the companies collecting and selling these data feel and indeed

1 The author wishes to express his gratitude to emeritus professor Peter E. Lloyd for his comments on the draft version of this chapter. 24 For the benefit of all? For the benefit of all? 25

Government“ is certainly maximize“ not the one making financial profit have any responsibility for what is things happen in the mation society that they traditionally Three dominant strategies maximize like the deindividuation theory and 4 being done with it, the reaction was datafication age, but at had in industrial society . This role is Continued digitization is now systemi- political control the bystander effect. Central in da- always the same: “That’s a very good not just fixing market failure, as often cally disrupting socio-economic struc- tafication is the need to collect and best the one watching maximize question!”. The companies made it suggested by the private sector. Par- tures, revenue models and the power combine, in one way or the other, as absolutely clear that they take no re- things happen and mostly ticularly in the datafication era, much balance in societies worldwide. Es- confusion many personalized data as possible sponsibility whatsoever for the usage. just the one wondering is at stake that requires government pecially since the datafication phase in order to make all sorts of assump- The established data economy in Sil- attention and intervention. The COV- and the emergence of the Internet tions and the best possible predic- what’s happened. ” icon Valley appeared to be a simple, ID-19 crisis, for one, has proven not of Things, we are confronted with in history for government-con- tions. Gathering ever more data is unregulated and unsupervised mar- only to experts that our information hyper-complexity and hyper-connec- trolled platforms to gain unprec- essential for data platforms. In fact, ket of demand and supply2. ” position now fully depends on the Big tivity. The main feature of the current edented state power over society they are really insatiable hungry. Tech firms, although the algorithms situation is, that a small number of - the dominant paradigm in China; The popular phrasing of data being they use are a black box and the data platforms own unbelievable amounts ★ to maximize confusion: datafica- When we take the first strategy, fol- the new oil is at least true in the understanding it and without govern- they provide is data we ourselves ap- of data on almost everyone and tion as the easiest way in history lowing the money -seeing how the sense that the current datafication ment effectively acting on it. In the parently have given them. So, rhetori- everything, anytime and anywhere. for everybody who has an inter- major players make their profits- is phase3 in the ongoing digitization has even more popular anonymous tri- cally speaking, are we sure that we’re These platforms are revolutionizing est in creating massive insecurity, the most direct way to understand resulted in the lightning-fast rise of a partite division of organization types, in the right game to begin with? the future of societies in ways that uncertainty, distrust and outright what’s happening. In recent years, whole new generation of companies government is certainly not the one are not all beneficial to say the least. chaos and to destabilize society the evidence is piling up that at least that are tirelessly inventive in discov- making things happen in the datafi- wherever and whenever they like the US Big Tech firms are really only ering new applications for their tech- cation age, but at best the one watch- In the Digitranscope project we iden- - the dominant paradigm for all in IT for the money and that they go nology almost every day, thus broad- ing things happen and mostly just the tified three dominant strategies in sorts of individuals and groups, very far indeed to maximize profits ening their scope and impact to all one wondering what’s happened. datafication: including state actors. for their shareholders and nobody kinds of different sectors that don’t ★ to maximize financial profits: da- else. They do this with all their might, know what hits them and how to re- There is growing consensus, at least tafication as the easiest way in All three strategies use the same showing great techno-optimism and act. Datafication rewrites the rules in the EU and individual Member history for privately-owned plat- technology, combined with insights using a very instrumental image of of almost every game in the book, States, that governments urgently forms to make unprecedented from psychology and sociology. It citizens as simply docile consum- without society fully recognizing and have to play the full role in the infor- amounts of money - the domi- is behavioural economics at work in ers, that they can manipulate to nant paradigm in the US; the fullest sense of the word (Aldred, do whatever they want them to do 2 https://www.geonovum.nl/uploads/documents/Geon- 4 See for example https://www.rathenau.nl/en/ ovum_Silicon_Valley_Studiereis_Rapport_eindver- digitale-samenleving/urgent-upgrade; and https:// ★ to maximize political control: 2019), skilfully making use of con- (Aldred, 2019; Zuboff, 2019). Their sie%20%28080519%29.pdf www.rathenau.nl/en/digitale-samenleving/directed- 3 See also https://en.wikipedia.org/wiki/Datafication digitalisation datafication as the easiest way cepts from crowd or mob psychology products and services are addic- 26 For the benefit of all? For the benefit of all? 27

tive-by-design5 and the sales phi- the second strategy is aimed at cre- years daily hacking attempts and has, in addition to its disruptive ef- cautious7. The third strategy is the losophy is largely the drug dealer ating a perfect surveillance society successful cyberattacks to disrupt fects and amongst other things, se- most imminent and acutely danger- model6. Furthermore, they do not run by government. For most Euro- social and economic structures. rious consequences for feeding ine- ous. With so many lies, half-truths shy away from actively discrediting peans this seems unthinkable. At And we know about the existence qualities, for undermining solidarity, and deliberately false messages, and even sabotaging efforts from the same time, we see a fast-grow- of the dark web for all sorts of for the future of work (although it we lose our sense of morality. This governments to regulate them. As ing number of cameras in our own criminal trades, or the market of is still very much in debate how this is not just a philosophical point; it is far as at least some are concerned, streets and other, less visible moni- DDoS attacks-as-a-service for sale will precipitate in our daily lives ex- practical. People have died because the very existence of government toring devices in all EU countries. So by anyone who pays a reasonable actly) and for exercising our funda- they don’t know what to believe will be redundant in a mature infor- is it a discussion of scale on which price. In addition to these increas- mental human rights (like the right with COVID-19 news. Experts like mation society (Zuboff, 2019). Ap- we permit surveillance or is there ingly grim threats, the datafication to liberty, the right to privacy and Francesca Bria and her group stand parently they have alternative facts more to our dislike of what happens age has brought us some new chal- the right to self-determination). The out by acknowledging this already when it comes to the vital role gov- outside the EU zone? The Chinese lenges that have become very seri- days are long gone, that digitization for some time, but the overwhelm- ernments have played and still play government is, compared to us, at ous, such as the enormous spread was the promising fresh way of in- ing power of the data monopolists in risky financing basic research and least quite transparent to its citizens of fake news, the rise of influenc- novative techies to improve service makes it a sideshow. In view of this in constantly developing the basic about what they’re doing. ing elections, of cyberbullying, or delivery and realize cost cuttings. it is really high time, and hopeful- infrastructure they all use and bene- of using social media for online The real focus of especially the ma- ly not too late, for the EU and their fit from (Mazzucato, 2018a). At first, we did not distinguish the shaming, criminalizing and cancel- jor innovators and frontrunners has individual member states to jointly third strategy as a separate one. ling people [3]. Fostering conflict dramatically shifted in the datafica- take a clear stand and to start act- The second strategy is essentially But looking at its clearly different and accepting inequality as a given tion age, and as a result we are now ing more energetically, coherently the same when comparing the un- objectives, its impact and the pro- is a prescription for social break- trapped in a completely different and boldly. What is our strategy? derlying principles and insights, but fessional and industrialized way in down. The current technologies situation. Analysing them in terms What is our European way? the objectives differ. Where the first which it is deployed by now, we de- easily accelerate this. Going wider, of objectives, underlying principles strategy is aimed at creating a per- cided it should definitely be added it is an opening to social destabili- and effects from the perspective fect surveillance society run by a few to the other two. The third strategy zation and populist-nativist move- of our European public norms and private companies (Zuboff, 2019), is the most sinister one, with many ments and thus a serious danger to values, each strategy is essentially faces that constantly keep chang- democracy. abject and reason for serious con- 5 https://www.theguardian.com/technology/2018/ 7 See for example the warnings of Stephen Hawking jan/08/apple-investors-iphone-addiction-children. ing. It has become an arms race cerns. You don’t have to be a sea- https://www.theguardian.com/science/2016/oct/19/ 6 Earning revenues in three steps: first you make life stephen-hawking-ai-best-or-worst-thing-for-huma- easy for people, let them try and give it for free, with quite a colourful collection of The conclusion about these three soned pessimist, a left-wing activist nity-cambridge and the joint concerns of the Pope, second you make them addicted, and third you start IBM and Microsoft on AI on https://www.bbc.com/ taking their money (or data!). weapons. We have seen already for strategies seems crystal clear. Each or technophobic to be at least a bit news/technology-51673296. 28 For the benefit of all? For the benefit of all? 29

A European way ID-19 crisis has not resulted in any p. 263 and 265). As shown in other munity- that, together with the state has shown that horizontal network ient society of self-reliant citi- When devising an alternative strate- setback or at least a serious slowing chapters, we have collected several and the market, supports prosper- extensions can be done at zero cost zens and competitive companies, gy, there are in essence three ques- down in revenue growth for Big Tech; interesting cases and done some ous and resilient societies. Involving (that’s how they make their super based on solidarity and fuelled by tions. The first is, of course, what on the contrary. It is the niche-play- fascinating experiments ourselves, the local level and the community profits). With the right leadership, a strong European public-private kind of society we want to live in? ers with less generic platforms that using the same technology as in the seems critical. There are wonderful horizontal social networks -com- R&D agenda; The second is, how to keep this so- are hit8, strengthening Big Tech even dominant strategies but at the same examples of creative things when munities of interest in the interest ★ Economic and social development ciety prosperous in the datafication further. Many suggest that this is time really challenging the idea that choosing this perspective. It might of communities- can maybe also be being equally important and go- age and beyond, while at the same in fact the whole point of the game it is inevitable that just a happy few turn out that communities can es- extended at zero cost. ing hand-in-hand, with room for time defending this prosperity when we’re in and thus something of a law benefit. Just look at the promising pecially be helpful to have a sharp customization that relates to in peril? And the third is, what we are of nature so to speak. But is this re- steps we have taken with big data eye for situational differences when In the DigiTranScope project we con- situational differences, priorities going to do? When trying to answer ally true? analyses on absolutely non-person- implementing solutions, instead of cluded that, all in all, a proactive Eu- and preferences; these questions, this chapter focus- alized open data about ‘synthetic imposing the same standardized ropean way to move forward in the ★ Embracing the benefits of the es on the role of government. But A European way should not start people’, enabling municipalities to developments everywhere, regard- datafication age on its own terms newest technology while at the let’s start somewhat more in gen- from the recurring simplification implement neighbourhood-specific less of the real issues and interests can best be built on the following same time keeping our eyes open eral. that value creation is only about policy interventions (see Chapter 9) locally at stake. During one of the foundations: for the downsides and acting making money. In the Digitranscope Digitranscope workshops there was ★ Government focusing on public firmly against them if necessary. The technological developments are project we wanted to go beyond Characteristic of most cases and ex- the interesting observation that this value driven innovation, for the The European way should certainly astonishing and have already result- that. We have been exploring and periments we studied, is that active is one of the points Big Tech again benefit of all and grounded in the not be reactive or even defensive ed in huge and useful progress in discussing how “a better alignment digital citizenship is stimulated and and again misses and thus one of European Convention of Human and protectionist, based on fear of various fields to seize opportunities between risks and rewards, across that both economic and social re- the reasons why their solutions are Rights as its moral compass (CoE, the evil associated with the datafi- for much requested improvements public and private actors, can turn turns occur -without one pushing the not equally popular in the cultures 1950); cation era when looking at where and to realize solutions for diffi- smart, innovation-led growth into other away- in settings where civil of the U.S., China and Europe10. The ★ Opting for open innovation 2.011, current strategies are headed. The cult-to-solve problems that we really inclusive growth”, and how to ensure society is seriously taken on board9. old problem was the challenge of co-created between public part- European response should be for- need to deal with in our society. At that “public value […] is not created In that sense there is definitely a scaling up small initiatives. IT can ners, private partners, communi- ward-looking with the overall ob- the same time it is clear, that in all exclusively inside or outside a pri- relation to the plea for reenergizing be useful here if it is used wisely ties and science; jective of creating and maintaining strategies up till now, the benefits vate-sector market, but rather by a the role of what has been called the by all cooperating parties. Big Tech ★ Creating a prosperous and resil- a proper balance between opportu- are very unevenly distributed. Da- whole society” (Mazzucato, 2018b, Third Pillar (Rajan, 2019) -the com- nities and threats, recognizing what 10 https://ec.europa.eu/jrc/communities/sites/jrccties/ tafication has so far been a “winner files/report_workshop_Digitranscope_connectingpol- 11 This means going beyond ‘traditional‘ open innova- is appropriately expressed as the 8 For example Booking.com. See https://www.ft.com/ 9 Not an entirely new idea as such, but the scale and icydevelopersanddecisionmakerswithDigitranscope- tion as originally promoted by Chesbrough, that was takes all” economy. Even the COV- content/64716675-b461-4fd3-ae0f-836973c68f12. the technology certainly are. experiments.pdf. mostly confined to private innovation. Janus face of technology (Moschella, 30 For the benefit of all? For the benefit of all? 31

Policy-making“ is not Big Tech, but also “the major players in just a matter of setting international finance or in auditing for 2011). Technology innovations are strict and for them beneficial con- goals and monitoring example- have huge interests in preserving the world here to stay and will inevitably go ditions they set, when experiment- their achievement; as it is today, and that they are very determined even further. Government should ac- ing with the use of apps to support cept this as a fact and encourage an crowd control13. COVID-19 has defi- it requires permanent and creative in finding ways to ensure that. investigative, exploratory and exper- nitely proven that the European way alertness of the ongoing This means they will resist structural imental attitude (‘permanent beta’) should stem from the conviction that changes, in order not to changes with all their might: towards what is happening, knowing it is really urgent for Europe to aban- miss the point. the system fights back! that it is an adventure where mis- don its innocence and naivety. This takes are made and where there is does not mean that we must make no “first time right”12 a complete break with our approach ” ” . of the last years. This certainly goes The COVID-19 crisis has been the for legislation. The GDPR has really as argued, on the one hand a new cisions have to be made fast with ultimate wake-up call -and in that been an excellent impetus to reset evolutionary stage in the ongoing far-reaching consequences (Harari, sense a gift- for everyone to fully un- the way Big Tech should handle data, digitization. But on the other, it is a 2020) in the knowledge that pow- derstand what hyper-complexity and and understandably set an example creeping revolution when we take erful organizations and institutions hyper-connectivity means in practice, for the state of California, among the trouble of seriously studying -Big Tech, but also the major players and the perfect demonstration of others. And let us also continue the underlying strategies. And these in international finance or in auditing how far Big Tech’s influence extends, our multi-stakeholder dialogues on studies should be done incessantly. for example- have huge interests in how heavily we depend on it, and open standards and frameworks, or Policy-making, particularly in this preserving the world as it is today, how this dictates our actions. De- on transparency in algorithms and area, is by nature not just a matter and that they are very determined spite our own efforts to register data, on ethical guidelines. But in view of of setting goals and monitoring their and creative in finding ways to en- only Google data and algorithms can all the fast changes, there should achievement; it requires permanent sure that. This means they will re- help us to monitor effectively the certainly be room for frequent ad- alertness of the ongoing changes, in sist structural changes with all their spread of the virus. And we can’t get ditions and fresh accents in the pol- order not to miss the point 14. There might: the system fights back! around Google and Apple with the icies and for stronger leadership in are no quick fixes; major difficult de- enforcing directives. Datafication is,

12 See Make it happen!, a report of the Dutch 14 I am somewhat paraphrasing Charles Leadbeater Information Society and Government Study Group, here, who once answered during a conference on established by the Dutch government. https:// 13 https://www.forbes.com/sites/zakdoff- how governments perform in digitization: “I’m sure www.digitaleoverheid.nl/wp-content/uploads/si- man/2020/06/19/how-apple-and-google-created- they’re hitting all their targets, but they’re missing tes/8/2017/09/Make-it-Happen.pdf this-contact-tracing-disaster/ the point”. 32 For the benefit of all? For the benefit of all? 33

Action perspectives the various national and transnation- Secondly, there is also almost no- of Estonian government officials be- population and showing a significant People won’t adjust when they miss The third question we asked our- al layers, as well as between govern- where any way for an ordinary per- ing convicted for viewing personal relation to social inequality in the the appropriate tools, designed ac- selves is what a European way means ments, industry, civil society and sci- son to correct personal data or have data without a plausible reason. more general sense (van Dijk, 2020). cording to their capabilities17. But in terms of what it is that we actually ence. Apart from identifying several them corrected, or to explicitly either Other countries, like for example When the objective of a European it takes more than just fixing this should do. There is an abundance of tips and tricks, we searched for real refuse or give consent for sharing the Netherlands, are discussing and way is -as proposed- first and fore- problem to have an inclusive infor- reports on the current developments. game-changers to consider as pillars personal data. experimenting to take the same most digitization and datafication mation society. When people are But at the same time, there is a re- for the European way to fundamen- step. for the benefit of all, this trend must not involved in society in gener- markable shortage of proposals for tally shift the course of the current Despite the GDPR, a huge amount be broken. Once again, this is not a al, when they feel left out, have no concrete action and, looking at the system. We are working out three. of personal data is still continuously Of course, the real game-changer call to action for just governments control over their lives, and watch role governments should play, pol- collected and used online by many will be to give people this right not to deliver accessible services as al- everything that happens full of fear iticians and top officials in the civil Data sovereignty private as well as public parties in just for governmental data collect- ready agreed (EC, 2016) but also for and distrust, they will also not be service show a serious shyness and The probably most talked about the EU without people giving con- ing and usage. It should certainly companies. And it is not only about part of the digital changes that are embarrassment to act. game-changer at the moment is to sent or even knowing that it is hap- also include private companies. In accessibility or using comprehensi- flooding them. As long as that is the put real control over personal data pening. Of course, many are glad not most cases, this is for now not in ble words, but also about an active case, socio-economic innovations In the Digitranscope project, we into practice. Everyone’s right to to have to give their data every other scope. But it most certainly must be. multichannel approach because for that really last won’t materialize. have therefore conducted not just control his or her personal data is second for endless reasons and hap- some e-mail is so old-fashioned by Realizing a European way as we pro- desk research on the challenges that one of the principles of the GDPR. py for personalized advertisements Inclusive growth now that they don’t use it anymore, pose will then fail. governments face and on how these But in general people have no clue that suit them. But the way in which A second game-changer to consider whereas others have still to learn relate to new governance models how to exercise this right and even things are happening at the moment is to create a breakthrough to drive the basics. It’s about informing peo- An effective game-changer for in- more in general15. In addition, we if they do have a clue, it is almost is serious food for thought. It is not true inclusive growth. This goes far ple about their obligations as well as clusive growth is not easy to find also organized intensive multidisci- impossible to actually be in control. only structurally against the law. It beyond allowing everybody to fully their rights as digital citizens, about and deserves further investigation. plinary workshops to (a) collect and First, transparency is still at best fuels uncertainty and mistrust of participate in the information soci- being transparent and building up However, one thing is clear. Inclu- compare experiences with the digi- fragmented when it comes to proac- being followed and monitored every ety. their trust on what is being done sion is not just an ideological issue, tal transformation of governments tively letting people know who has step of your life. with their data and what they may it is a matter of common sense. If themselves and (b) explore ways to personal data on them, who uses Despite the popular idea that it is just expect at what moment, et cetera. you don’t strive for an inclusive in- create an effective interplay between them and how, and with whom they Estonia has been the first country the elderly that can’t keep up with And last but not least, it is not just formation society because it’s the are shared and for what reasons16. to empower its citizens to really ex- the changes, the fact is that there a matter of pious promises, but also right thing to do, then do it because 15 For an example of EU studies on this; see https:// ercise their right to control personal is abundant proof that the digital di- of really tackling organizations that ec.europa.eu/jrc/en/publication/future-govern- 16 We consciously stay away from the term ‘ownership’ ment-2030. of data. data and there are indeed examples vide is growing in all sections of the don’t comply. 17 https://news.un.org/en/story/2019/06/1040131. 34 For the benefit of all? For the benefit of all? 35

we can’t afford the dropout, which islation) (Kimman, 2011). When it the world. The Danish contain- not part of industry self-regula- as the “iron law of the market” according to some estimates has comes to datafication, we seem very er giant Maersk was one of the tion schemes, nor government and that inequality, monopolistic now reached 20 to 25 percent of the liberal in addressing each other and first victims, but it soon turned enforcement tools? control or ownership of proper- population in the Netherlands, for very reluctant in actually enforcing out that many more companies ★ The second is the enforcement ty are political choices (Piketty, example18. codes of conduct, contracts or even were affected. American pharma- of the GDPR. If, as argued, the 2019). laws. The EU and its member states ceutical company Merck, Russian competition in datafication for The third game-changer would be Stricter and more comprehensive mainly limit themselves to demand- oil giant Rosneft, British advertis- Big Tech is essentially about who to seriously start acting in this line compliance and law enforcement ing fair tax payments from the Big ing company WPP, Spanish food holds the largest amount of per- of thinking. This does not automat- The third game-changer we propose Tech companies and imposing fines company Mondelez: computer sonalized data, their operation ically mean more regulations and/ is stricter and more extensive com- for breaking the rules, even if this is systems were going down every- fundamentally collides with the or administrative burden. The Dutch pliance and law enforcement, includ- done systematically and not inciden- where. In Ukraine, the airport heart of the GDPR. For private Council of State (the supreme ad- ing more systematic supervision on tally. near Kiev was hit, as were sev- data platforms, working within visory body on legislation and the how the datafication market as a eral banks. This was not the first the GDPR means the end of their highest general administrative whole is developing. We think this During the Digitranscope project, or last time that major chain ef- very lucrative revenue model court) has convincingly argued that is the most seriously missing link at we touched on compliance and law fects and even global disruptions and ultimately the end of their -in view of the hyper-complexity and the moment and an indispensable enforcement many times. We give have occurred as companies are business. So what are we trying the hyper-connectivity we live in addition to current legislation and here two situations to reflect on the negligent in their cybersecurity to achieve by imposing fines for and the speed of change- it is best the other two game-changers if we challenges we see with the existing and do not consider simple things violations? Again, revocation of to downsize laws to the core values want them to be effective and make approach: like timely patches necessary. operating licenses comes to mind at stake instead of losing ourselves a real difference. ★ The first concerns addressing cy- How should we respond to these as probably a more effective ap- in detailed rules that cloud what we bersecurity weaknesses that are kinds of situations? It seems that proach. Or, if their market power essentially want to regulate19. This All modern organizations, whether structural and have significant we limit ourselves to a firm con- and their power in general are in- helps us much better to identify sys- they like it or not, are also moral implications. As an example, think versation or perhaps a fine. Ap- deed so overwhelming, why not at tem failures and to intervene firmly actors. Analytically, their behaviour back on the worldwide ransom- parently, serious measures are least consider forcing the split of if necessary. can be regulated on three levels (by ware attack in June 2017 that not being considered, even if the these companies, just as we did self-discipline, by self-regulation in struck several major companies, consequences are enormous and a few years ago when banks and sectors or professions, and by leg- simply because they hadn’t up- critical processes are disrupted insurance companies merged on dated their software in time. The for days. What about revoking a large scale? In this sense, we 18 https://www.government.nl/documents/re- 19 See (in Dutch) https://www.raadvanstate. ports/2018/06/01/dutch-digitalisation-strategy. attack spread like wildfire around operating permits? Why is this agree that there is no such thing nl/@112661/w04-18-0230/ 36 For the benefit of all? For the benefit of all? 37

Conclusion and closing remarks we now know it. That role is to proac- the interplay between the layers of led to an interesting reappraisal in References Mazzucato M. The entrepreneurial There is a Dutch saying that in lit- tively protect public interests based government must be determined on society of the role of both govern- Aldred J. License to be bad: how state: debunking public vs private eral translation says „gentle healers on the type of society we want. It a situation-by-situation basis, and ment and science. This also applies economics corrupted us, Allen Lane. sector myths, Rev. Ed., with a new make stinky wounds“ which means is the natural role of the State to politically this seems quite possible. to digitization and datafication. The 2019 foreword, Penguin Books 2018a. that it is better to treat a problem condition the game by determining EC and the other layers of govern- Cocking D. and van den Hoven J. Mazzucato M.. The value of thoroughly, even if the treatment is the rules, explicitly based on human Of course, after this exploratory ment in the EU would do well to use Evil online, Wiley Blackwell. 2018 everything: making and taking in the painful, otherwise it can get worse. rights and fundamental freedoms, Digitranscope study, the work is not this momentum together to really Council of Europe. European global economy, Allen Lane. 2018b. Despite all the miracles that we and by systemically enforcing the over yet. This should also include implement the European way as Convention on Human Rights. 1950. Moschella D. The next generation of clearly like and benefit from (Ward- laws about playing the game. topics that we have not focused on recommended. At the same time, https://www.echr.coe.int/Pages/home. digital game-changers, Leading Edge ley, 2014), the datafication has so far. Looking at what we touched the mistrust we now feel about Big aspx?p=basictexts Forum. 2011. progressed so far that this saying The systemic changes we’ve seen in this chapter, it would make sense Tech, but also about other private Cukier K. and Mayer-Schoenberg- Moschella D. and Mead K. Lever- comes to mind. We are well on the over especially the last decade are to explore the role of government sectors, does not necessarily lead to er V. The Rise of Big Data: how it’s aging ‘The Matrix’- digital ecosystem way to completely losing the human not easy to adjust. As mentioned, to make Europe less dependent on a renewed confidence of society in changing the way we think about dynamics and strategies, Leading touch in datafication, and in that major interests are at stake. The the US and East Asia in this ongoing government. In the EU, for example, the world, Foreign Affairs, May/June Edge Forum. 2015. sense things are really getting out of proposed game-changers are inter- competition. Another issue would governments also use AI in incom- 2013. https://www.foreignaffairs.com/ Piketty T. Capital and ideology, Har- hand. If the question at the begin- ventions that will only be effective be how to improve the internal or- prehensible, hidden and rigorous articles/2013-04-03/rise-big-data vard University Press. 2019. ning of the chapter was “And what, if deployed at EU level. There, the ganization of public administra- ways that rightly scare people. A European Commission. Directive Rajan R. The third pillar: how mar- if any, is the role of government in guiding principles must be formu- tion to act in a more coherent and successful European path depends (EU) 2016/2102 of the European kets and the state leave the commu- all of this?”, the answer seems now lated and maintained. This does consistent manner. The silos as we on the clear profile, transparency Parliament and of the Council of 26 nity behind, William Collins. 2020. clear enough. Big Tech and the mar- not mean that the implementation know them all work from their own and unambiguous intentions of the October 2016 on the accessibility of van Dijk J. The digital divide, Polity ket it has created will not change by cannot start with a number of front perspective. One moment, Big Tech government. It needs a government the websites and mobile applications Press. 2020. itself because it is much too profita- runners or that everything has to be companies are asked to invest and that is not part of the problem, but of public sector bodies. Publications Wardley S. Of wonders and disrup- ble, and the community cannot force implemented in exactly the same establish their data centers in the undeniably just part of the solution. Office of the European Union. 2016. tion, Leading Edge Forum. 2014. the market to do so. So it is up to the way and at exactly the same time. EU; the next moment their integrity Harari Y. N. The world after corona- Zuboff S. The age of surveillance government to do what the govern- Digitization and datafication are is questioned. So what do we really virus, 2020. https://www.ft.com/con- capitalism: the fight for a human ment exists for. That role is not just not as hot political topics as climate want from them? tent/19d90308-6858-11ea-a3c9- future at the new frontier of power, facilitating and supportive by repair- change, health care or social security, 1fe6fedcca75. Profile Books. 2019. ing social inequalities as inevitable on which political parties may differ A recent observation is that the COV- Kimman E. Understanding compli- collateral damage of datafication as in major ways. For the benefit of all, ID-19 crisis, at least in the EU, has ance, Rozenberg Publishers 2011. 38 39

PART A: on the Governance of Digital Data 40 Data governance models emerging from the practices of social actors 41

Marina Micheli, Marisa Ponti, Massimo Craglia, Anna Berti Suman Data governance models emerging from the practices of social actors1

Introduction This chapter illustrates the results of our research on the emerging data governance models. The study has examined current discourses and practices on the governance of data. In particular, it scrutinised approaches for accessing, controlling, sharing and using data in today’s platform economy and derives four emerging models of data governance. These models could be understood as inventive practices that problematize current arrangements and reassemble them in accordance to the interests of the actors involved.

1 This chapter draws on the following article: Micheli M., Ponti M., Craglia M., & Berti Suman A. (2020). Emerging models of data governance in the age of datafication. Big Data & Society, 7(2), https://doi. org/10.1177/2053951720948087 42 Data governance models emerging from the practices of social actors Data governance models emerging from the practices of social actors 43

With this research we wish“ to increase knowledge the power“ relations about the practices for data governance that are between all the actors currently developed by various societal actors - Data governance: A social non-state actors (including citizens) 2014) and data politics (Ruppert et affected by, or having an beyond ‘big tech’ - emphasizing these actors’ power science-informed definition interact. Yet this is easier in theory al., 2017). Informed by these schol- effect on, the way data The term governance has been than in practice as power disparities arships and concepts, we understand to control how such personal data is accessed and is accessed, controlled, extensively used in the last two among actors continue to exist and data governance as the power rela- used to produce different kinds of value. decades but its meaning is still matter. Market actors often benefit tions between all the actors affected shared and used, the ambiguous (Colebatch, 2014). Our from these more fluid allocations of by, or having an effect on, the way various socio-technical understanding is informed by exist- power and responsibilities (DeNardis, data is accessed, controlled, shared ” arrangements set in place ing debates in the political science 2019; Srnicek, 2017), at the detri- and used, the various socio-tech- and risk scholarship (Kooiman, 2003; ment of less (economically) power- nical arrangements set in place to to generate value from Colebatch, 2014) where, for example, ful actors such as citizens, commu- generate value from data, and how data, and how such value In the last years, following scandals (albeit increasingly divergent by ge- governance has been framed as “the nities and civil society organizations such value is redistributed between is redistributed between like Cambridge Analytica and new opolitical context), see the hegem- multitude of actors and processes (Heeks and Shekhar, 2019). actors. Such social science-informed actors. regulations for the protection of onic position of a few technology that lead to collective binding deci- definition allows moving beyond data like the GDPR, there is mount- corporations that have established sions” (Van Asselt and Renn 2011: Based on this understanding of gov- concerns of technical feasibility, ef- ing attention concerning how data de-facto quasi-data monopolies. In 431). Governance broadly refers to ernance, we examine in this chapter ficiency discourses and ‘solution- ” collected by big tech corporations terms of data governance, this is the web of actors involved, with dif- ways in which personal data collect- ist’ thinking. Instead, it points to and business entities might be ac- reflected in an asymmetry of power ferent roles, in the process of gov- ed through datafication processes is the actual goals for which data is cessed, controlled, and used by other between those corporations, which erning a system. The term stresses and could be governed. This contri- managed, emphasising who bene- societal actors. Scholars, practition- hold most of the decision-making a discontinuity from so-called “com- bution, however, adopts a social sci- fits from it, the power un(balanc- ers and policy makers have been power over data access and use, mand-and-control” by the State, and ence-informed perspective of data es) among stakeholders (or lack of exploring the possibilities of agency and other stakeholders. With this acknowledges that a broader set governance that complements other thereof), the kind of value produced, for ‘data subjects’, as well as the ‘al- research we wish to increase knowl- of actors and institutions are (also) framings, such as those of platform and the mechanisms (including the ternative data regimes’ that could edge about the practices for data involved in managing societies (pri- governance or privacy and data pro- underlying principles and system allow public bodies to use such data governance that are currently devel- vate sector, civil society and other tection law. Our perspective on data of thoughts) that sustain these ap- for their public interest mission. Yet, oped by various societal actors - be- non-government entities) (Kooiman, governance draws in particular from proaches. Based on this conceptu- the current circumstances, which are yond ‘big tech’ - emphasizing these 2003). As Wolf (2002) puts it, the science and technology studies and alisation, the chapter examines four the result of a tradition of ‘corporate actors’ power to control how such governance phenomenon takes critical data studies, which informed emerging data governance models. self-regulation’ in the digital domain personal data is accessed and used place within horizontally-organized our work through concepts of data and an overall laissez-faire approach to produce different kinds of value. structures where both state and infrastructure (Kitchin and Laurialt, 44 Data governance models emerging from the practices of social actors Data governance models emerging from the practices of social actors 45

DSPs are“ horizontal joint initiatives among Research Strategy consolidated. The review covered documents added during the review Emerging data governance data holders to aggre- for data sharing, how data can be We delved into grey and academic lit- documents, publications, news and process. Most of these are recent, as models gate data from differ- handled, and for which purposes. erature, as well as news articles and websites in English that addressed 74% have been published from 2017 This section describes the data gov- These contracts could be ‘repeata- websites of recent projects and initi- emerging practices for the govern- onwards. ernance models identified following ent sources to create ble frameworks of terms and mech- atives to look for emerging practices ance of data with a focus on the the five dimensions described above. more value through anisms to facilitate the sharing of of data governance. The collection European context. On the whole, it To guide our analysis and description These models should be understood their combination. data’ between entities, which are es- of resources started in preparation included 72 academic articles, 16 of the emerging models of data gov- as ideal types. They are abstract pecially useful for organisations that of a workshop held in October 2018 book chapters, 63 reports and pol- ernance we used the analytical di- constructs. They are not intended do not have the knowhow and legal on data governance with seventeen icy documents, and 22 websites of mensions, drawing in particular from as an exhaustive description of the economic benefits” for all the par- support to leverage data (Hall and invited experts from academia, pub- projects/initiatives. The resources Winter and Davidson (2018) and state of the art, but as a contribu- ties involved (Carballa Smichowski, Pesenti, 2017; Hardingens and Wells, lic sector, policymaking, research were collected in the time span from Abraham, Schneider and vom Brocke tion in synthetizing emerging data 2019). DSPs are horizontal joint in- 2018). Although these frameworks and consultancy firms (Micheli et al., October 2018 to July 2019, with 9 (2019) (Table 3.1). governance models. The four mod- itiatives among data holders to ag- have been referred to as data trusts, 2018). We adopted a flexible search els described are labelled: data shar- gregate data from different sources there is not a full consensus whether strategy and used a snowballing ap- Dimension Definition ing pools (DSPs), data cooperatives to create more value through their they could be assimilated to actual proach including progressively new Stakeholders The individuals, institutions, organisations or groups (DCs), public data trusts (PDTs), per- combination (Mattioli, 2017; Shka- legal trust structures or, instead, to sources according to their relevance who are affected by, or have an effect on, the way sonal data sovereignty (PDS). batur, 2019). Their overall rationality a ‘marketing tool’ that facilitates the to the theme of interest. The initial data is governed and the value created. Data sharing pools is attuned with dominant discursive sharing of data (Delacroix and Law- sources considered were identified Governance goals The objectives held by actors who influence how data Different actors join a DSP to ‘an- regimes of Big Data (Kitchin, 2014) rence, 2019: 242). The assumption for the preparation of the workshop is governed. alyse each other’s data, and help and lies in the assumption that ‘the of such contracts is that all parties and from the inputs provided by Value from data The resources expected to be generated from the fill knowledge gaps while minimiz- greatest advantages of data sharing benefit since the DSP enables them the workshop participants. A sub- use of data and how these are distributed among ing duplicative efforts’ (Shkabatur, may be in the combination of data to obtain easily data that would sequent step was to review related actors and across society. 2019: 30). By creating these partner- from multiple sources, compared otherwise be inaccessible. There is work, which addressed similar issues Governance mech- The different instruments adopted to achieve specif- ships, they ease the economic need or ’mashed up’ in innovative ways’ reciprocity between partner organ- or was directly linked to the sources anisms ic governance goals, including the underlying prin- for exclusive rights and obtain limit- (Mayer- Schonberger and Cukier, isations, but only data holders are examined. Simultaneously, we kept ciples. ed co-ownership stakes in the result- 2013 cited in Mattioli, 2017: 184). involved, as data subjects tend to be track of new publications and on-going data pool. Data is treated and excluded from the relation and are Reciprocity The power relation between stakeholders for data ing projects or initiatives. The review exchanged as a market commodity A key mechanism for DSPs is the at best depicted as passively bene- access and use. strategy proceeded iteratively, un- with the aim of producing data-driv- contract, a legal and policy frame- fiting from it. A practical limitation til the typology of the models was Table 3.1: Analytical dimensions en innovation, new services, and work, that defines the modalities for data sharing pools consists in 46 Data governance models emerging from the practices of social actors Data governance models emerging from the practices of social actors 47

DCs enable “a de-centralised data governance the transaction costs, such as data approach in which data established in UK and France in the social change and addressing soci- Public Data Trusts preparation, ensuring privacy and in- subjects ‘voluntarily pool 19th century, and from the more re- etal issues, for instance by foster- PDTs refer to a model of data gov- teroperability challenges, which put cent platform cooperativism (Scholz, ing equality, digital rights, environ- ernance in which a public actor ac- their data together, to small businesses and under-funded 2016). The cooperative movement mental causes or medical research cesses, aggregates and uses data entities at a disadvantage (GovLab, create a common pool for promotes fairer conditions of value (Carballa Smichowski, 2019; Sand- about its citizens, including data 2018). A further limitation is that mutual benefits’. production, in a non-monopolistic oval, 2020). Many DCs are ‘com- held by commercial entities, with often there is one dominant partner and transparent setting, alternative mons-based’ and open, blurring the which it establishes a relationship of (Carballa Smichowski, 2019). There- to the dominant capitalist model distinction between the notion of trust (Delacroix and Lawrence, 2019; fore, although involving potentially submitting to the” extractive logic of (Pazaitis et al., 2017). Analogously, data commons and DCs (‘open co- Hall and Pesenti, 2017; Mulgan and many actors beyond big tech plat- digital capitalism (Borkin, 2019; Ho DCs address the power unbalances operativism’) as data is shared with Straub, 2019). Several stakeholders forms, the relations are not neces- and Chuangt, 2019). Therefore, data of the current data economy and an open license and made public might be involved in this model, in- sary as horizontals (and sustainable) subjects are key stakeholders within are an explicit attempt to rebalance (Carballa Smichowski, 2019; Ho and cluding city administrators, manag- as claimed. DCs. By establishing a relationship the relationship between data sub- Chuangt, 2019; Pazaitis et al., 2017; ers of public institutions, platform of trust with the cooperative that jects, data platforms and third-par- Sandoval, 2020). companies, trusted data interme- Data cooperatives manages data on their behalf, they ty data users. Enabling mechanisms diaries, research institutions, start- DCs distribute data access/rights preserve democratic control over for DCs are ‘bottom-up data trusts’ Examples of DCs operating with ups, and SMEs. Public administra- among actors like DSPs, but dif- their data and might demand an (Delacroix and Lawrence, 2019): health data are MIDATA.coop and tions may also invite third-parties ferently from those, provide higher equitable share in the benefits pro- agreements and contracts that pro- Salus Coop that let citizens donate to access their data sources and involvement of data subjects and duced (Borkin, 2019; Delacroix and vide the means for citizens to be their personal health information for develop data-driven services or to are guided by different goals. DCs Lawrence, 2019). This model is char- informed, express their preferences scientific research. Although there offer guidance on data sharing (Hall enable a de-centralised data gov- acterised by high reciprocity since and concretely decide how to share is a growing interest in DCs for eth- and Pesenti, 2017; Morozov and Bria, ernance approach in which data sub- ‘all parties are stakeholders and are their data and for which purpose. ical approaches to data sharing and 2018). A key goal of PDTs is to inte- jects ‘voluntarily pool their data to- equally affected and bound by the use (e.g. Ilves and Osimo, 2019), at grate data from multiple sources to gether, to create a common pool for governing rules they discuss, nego- DCs need to generate sufficient in- the moment this model struggles to inform policy-making, promote in- mutual benefits’ (Ho and Chuangt, tiate and then agree upon’ (Ho and come for their maintenance and scale up and to compete against big novation and address societal chal- 2019: 204). Participants of DCs Chuangt, 2019: 203). development, but are not based on tech that are advantaged by their lenges, while adopting a responsible share data while retaining control profit-maximising objectives. They monopolistic position, their critical approach to the use of personal data over it, having a say on how it is The underlying principles of DCs stem often aim to create public value mass of users, and greater financial (Bass et al., 2018; Morozov and Bria, managed and put to value, and not from the co-operative movement, across society, including promoting resources (Sandoval, 2020). 2018). 48 Data governance models emerging from the practices of social actors Data governance models emerging from the practices of social actors 49

In PDTs,“ ersonal data public actors assume sovereignty for the role of trustees securely managing data, preserving to data of public interest to public empowerment economic growth, that guarantee citizens’ citizens’ privacy, and maximising actors under conditions specified in ublic data trusts the public value of data (Mulgan the law (Shkabatur, 2019). This was for public interests private profit and data is handled knowledge ata cooperatives and Straub, 2019). These entities considered by the EC (2020c), which and more efficient for public interest ethically, privately will be independent and unrelated then appointed a High-Level Expert public service scientific research delivery and securely. to for-profit firms and big tech cor- Group on Business-to- Government and empowerment porations, and guarantee that data data sharing. The issue has also Data sharing pools is managed without abuses through been discussed at national level in ” for private profit In PDTs, public actors assume the strong accountability and standards. Europe. For instance, French Member and economic growth role of trustees that guarantee cit- Even if citizens are mostly seen as of Parliament Belot proposed creat- izens’ data is handled ethically, pri- recipients who benefit from services ing the legal concept of ‘territorial vately and securely. Thus they imply and policies developed through PDTs, interest data’ to give local govern- the establishment of a relationship they might be explicitly involved ments the power to demand access of trust between citizens and public through ‘trust building’ governance to data (Carballa Smichowski, 2019). bodies: citizens must be reassured mechanisms such as living labs, ublic bodies that public actors are capable to public consultations and civic socie- keep their personal information ty initiatives.

Civic organizations safe and secure and that they will An underlying assumption of PDTs ata subects use data for the public interest is that all data with a public interest (Collinge, 2018). To earn trust from component is part of a nation infra-

usiness entities citizens, public bodies might engage structure, therefore the information ata subects usiness entities in citizens’ consultations and living it affords should be ‘socialised’ to ublic bodies labs, or require the intervention of produce value for citizens and socie- external independent organisations ty as a whole (Cardullo, 2019; Moro- that act as trusted intermediaries zov and Bria, 2018). Whilst at pres- (Collinge, 2018; EC, 2020c; Mulgan ent PDTs are largely limited to small and Straub, 2019). These trusted pilot projects, a key enabler would intermediaries are new institutions be a legal framework mandating Figure 3.1 - Vignette of the data governance models that are allegedly held to account for private companies to grant access examined in the chapter. 50 Data governance models emerging from the practices of social actors Data governance models emerging from the practices of social actors 51

PDSs“ are expected to produce Personal data sovereignty value in the form of Among the main mechanisms ena- public interest, but at the same time Discussion With respect to the kind of value The PDS model is characterised by data subjects’ bling PDS are personal data spaces, foster economic growth through In this chapter we contribute to the pursued, DSPs focus on producing data subjects having greater con- like Digi.me, Citizen-me or Meeco, an ecosystem of new commercial policy debate on data governance economic value, while other forms self-determination, trol on their data, both in terms of which consist of ‘intermediary ser- services supporting them. A limit of using a socio-technical perspective of value gradually “chime in” in the privacy management and data port- knowledge, and vices’ allowing users to store their this model lies in its dependence on to describe four emerging models remaining models, such as social ability compared to the current dom- public interest, personal data, collecting data dis- personal data spaces as these are of data governance: Data sharing change, public interest, fairness, and inant model. The label comes from seminated in different platforms, currently adopted by only a niche pools (DSPs), Data cooperatives but at the same time data subjects’ self-determination. the broader principle of technolog- and control their sharing with third of users and often fail to scale be- (DCs), Public data trusts (PDTs) and For the most part, these models ical sovereignty, which concerns foster economic parties (Lehtiniemi, 2017). These yond pilots (Ilves and Osimo, 2019). Personal data sovereignty (PDS). could be found in niche initiatives or subjects, public administrations, or growth through an services, which appeared in early Furthermore, as business entities, The models are abstract concep- pilot projects, and there is still lim- governments regaining control of ecosystem of new 2000s, have been strengthened by they may have interested in how tualisations (Kvist, 2006) that do ited research concerning the value technology, digital content and in- Art. 20 of the GDPR (data portabil- to ‘nudge’ users and a few personal not necessarily represent discrete they generate and their sustaina- commercial services frastructures – thus reducing the in- ity). They are expected to remove data spaces might gain more power implementations of data govern- bility over time (although interest in fluence of IT commercial enterprises supporting them. obstacles for individuals wanting to in the market (Lehtiniemi, 2017). An- ance. Nonetheless, they provide a these models is signficantly increas- and of foreign States in which these exchange their data for research or other limit is that citizens have lim- foundation for discussion on al- ingly and many studies will prob- companies reside (Villani, 2018). ” other purposes, acting as trusted in- ited awareness about platforms’ use ternative approaches or “desirable ably be published in the upcoming termediaries and improving citizens’ of personal data for profit and the futures” for accessing and sharing months and years) (Borkin, 2019). At This model promotes a different and model pursues two goals: it increas- ability to make choices about their need for alternative models of value data in the age of datafication (Jas- the current time, the value produc- fairer data economy, echoing critical es individuals’ self-determination, data (Delacroix and Lawrence, 2019). production, and the majority would anoff, 2015). All models highlight a tion and redistribution in the four accounts of the dominant model of granting more opportunities to ac- not be capable, nor have the time concern for redressing the struc- models could be assessed more at surveillance capitalism (Lehtiniemi, cess, share and use personal data, PDS has been especially encour- to, take advantage of the opportu- tural power imbalances between the level of the imaginary, than from 2017). Data subjects are envisioned and engendering a more balanced aged within the context of MyData, nities offered by these intermediary corporate big data platforms and evaluations of tangible outcomes. In as key stakeholders together with relationship between users and dig- an international movement and a services (e.g. Andrejevic, 2014). En- other actors, such as data subjects, DSPs, data is a “market commodity” digital service providers – which de- ital platforms; and it is expected to community of activists, non-profit visioning citizens as ‘market agents’ public bodies, third parties, civil and economic value is redistribut- liver the means for subjects to con- foster a socially beneficial usage of organisations, think-tanks as well (Lehtiniemi and Haapoja, 2020) free society and researchers. There are ed horizontally among data holders trol, use and share their data – and data through the development of as commercial actors, start-ups and to choose from an ecosystem of per- nonetheless substantial differences who join the partnership. PDSs put re-users with whom data subjects new data-driven services centred on SMEs. PDSs are expected to produce sonal data spaces might not fully ad- regarding which stakeholders exert forward important innovations for decide to share their data (Ilves user needs (Ilves and Osimo, 2019; value in the form of data subjects’ dress the asymmetries of power of influence over data, and what value data subjects’ exerting digital rights, and Osimo, 2019). This governance Lehtiniemi, 2017). self-determination, knowledge, and the current data landscape. is pursued through data use. but do not question the datafication 52 Data governance models emerging from the practices of social actors Data governance models emerging from the practices of social actors 53

DCs“ allow The“ more data subjects powerful data to collect and analyzing different data sources to 2019), and the resulting dominant subjects are in a aggregate their data inform policy-making and address model of data governance stirred by data governance model, societal challenges. If in DCs a co- big data platforms, it is advisable to for the public interest, the greater accountabili- operative has to be trusted, in PDTs look at the inventive data practices while PDTs act on is a public body. Yet, in the latter, it of civic society and public bodies as ty is required to the data behalf of citizens, might also be that a trusted external it is from these actors that we have holders, which in turn independent organisation acts as a found more interest in the redistri- aggregating and limits risks and data intermediary between citizens bution of value generated through analyzing different data and a public body; this demonstrates data. data misuses. sources to inform poli- how the abstract models can easily cy-making and overlap in practice. PDTs represent An important dimension to discuss ” a form of public-driven governance is the extent to which these models ance, such as dataveillance, function address societal that could significantly redistribute democratize data governance. To an- creep, technocratic governance, etc, challenges. the value of data and increase fair- swer it we turn our attention to three (Kitchin, 2014). The more powerful ness, but requires the support of a models that involve data subjects. In data subjects are in a data govern- ” new legal framework mandating ac- all cases, data subjects can choose ance model, the greater accounta- and commodification (Van Dijck et cess to data for public interest. Sim- a trusted intermediary for their data, bility is required to the data holders, al., 2018) mechanisms of big data ilarly, DCs are a fairer alternative of being it a commercial service from which in turn limits risks and data platforms. They are oriented to- surveillance capitalism, but struggle an ecosystem of personal data spac- misuses. At the opposite end, DSPs wards the creation of value for the to find financial sustainability and to es (PDS), a cooperative that allow to are only accessible to data holders individual (self-determination) and reach a critical mass of users. There- keep democratic control over data and/or those in a position to pay for new commercial actors (data servic- fore we did not find a single model and share responsibilities (DCs), or data. How does that model guar- es), with public interest as a by-prod- to be “recommended” or “promoted” a public body that is entrusted by antee that needs and interests of uct of these. The remaining models for a fairer data landscape. Instead, citizens to use (their) data ethically data subjects (citizens at large and expressively pursue the public in- a combination of all these models and for the public interest (PDTs). In- marginalized groups) are accounted terest: DCs allow data subjects to should be envisioned for a “desira- volving subjects in the governance for? To address this, and for good collect and aggregate their data for ble future” (Jasanoff, 2015). In par- of data is a key strategy to address, data governance, it may be advisa- the public interest, while PDTs act on ticular, to oppose the privatization and avoid, many of the possible neg- ble to combine DSPs with the others behalf of citizens, aggregating and of internet governance (DeNardis, ative consequences of data govern- models that offer more guarantees, 54 Data governance models emerging from the practices of social actors Data governance models emerging from the practices of social actors 55

at least in principle, in terms of ac- lic sector in a top-down manner to References In: Cardullo, P., Di Feliciantonio, C. countability. earn trust and foster the public in- Abraham, R., Schneider, J. and and Kitchin, R. (Eds.) The Right to terest, as in PDTs. Vom Brocke, J., ‘Data governance: the Smart City, Emerald Publishing The findings of this study highlight A conceptual framework, structured Limited, Bingley, 2019, pp. 85–98. that the same “buzzwords” can be A final consideration concerns the in- review, and research agenda’, In- Collinge, A., A ‘new deal’ for city associated to different rationali- tertwined relationship between the ternational Journal of Information data? 2018. Available at: https:// ties of data governance, since the data practices we have examined Management, Vol. 49, 2019, pp. data.london.gov.uk/ blog/a-new- notion of data intermediaries and and the regulatory frameworks in 424–438. deal-for-city-data/ (accessed 18 data trusts is included somehow in which they exist. These data gov- Andrejevic, M., ‘The big data June 2020). all models. This underlines how im- ernance models can only develop divide’, International Journal of Delacroix, S., and Lawrence, N. portant it is to think critically about further if they are sustained by ap- Communication, Vol. 8, 2014, pp. D., ‘Bottom-up data trusts: Disturb- data infrastructures as socio-tech- propriate legal frameworks, such as 1673–1689. ing the ‘one size fits all’ approach nical products, moving beyond mere the GDPR for personal data or a new Bass, T., Sutherland, E. and to data governance’, International instrumental and technical aspects. legal act to mandate access to com- Symons, T., Reclaiming the Smart Data Privacy Law, Vol. 9, No 4, 2019, Data trusts might be powerful mercial data of public interest. With City. Personal data, trust and the pp. 236–252. means to reduce the power unbal- the recent developments in data pol- new commons, NESTA, London, UK.. DeNardis, L., ‘The social media ances of the current data economy icy (EC, 2020), the European Com- 2018. challenge to internet governance’, if adopted within DCs, while they mission is strengthening its role as Borkin, S., Platform co-operatives. In: Graham M. and Dutton. W. H. may foster very different aims in transnational regulator of technol- Solving the capital conundrum, NES- (Eds.) Society and the Internet: How DSPs. Indeed, in the first case these ogy with repercussions on a global TA and Co-operatives UK, London, Networks of Information and Com- would be “bottom-up data trusts” scale. In doing so, it will be crucial to UK, 2019. munication are Changing Our Lives. that act in behalf of citizens’ inter- engage with the wider set of stake- Carballa Smichowski, B., ‘Alter- Oxford University Press, Oxford, ests and preferences (Delacroix and holders identified in our research native data governance models: 2019, pp. 389–402. Lawrence, 2019), while the latter including local administrations and Moving beyond One-Size-Fits-All European Commission. Communi- would be repeatable frameworks of many actors from civil society who solutions’, Intereconomics. Review of cation from the Commission to the terms and mechanism to facilitate have an important role in shaping European Economic Policy, Vol. 54, European Parliament, the Council, the sharing of data (Hall and Pesenti, the emerging forms of data govern- No 4, 2019, pp. 222–227. the European Economic and Social 2017). Conversely, data trusts could ance that address the asymmetries Cardullo, P., ‘Smart commons or a Committee and the Committee also be a service offered by the pub- of the current data landscape. “Smart Approach” to the commons?’ of the Regions. Shaping Europe’s 56 Data governance models emerging from the practices of social actors Data governance models emerging from the practices of social actors 57

Digital Future. COM(2020) 67 Final, Daly A, Devitt K. and Mann M. (Eds.) Kooiman, J. Governing as Govern- trusts, collaboratives and coops can economy. Report for the Rosa Lux- to contextual integrity’, The Infor- Publications Office of the European Good Data. Institute of Network ance. Sage, London, 2003. help govern data for the maximum emburg Foundation, New York, 2016. mation Society, Vol. 35, No. 1, 2019, Union, Luxembourg, 2020. Cultures, Amsterdam, 2019, pp. Lehtiniemi, T., ‘Personal data spac- public benefit. Nesta, London, 2019. Shkabatur, J., ‘The global commons pp. 36–51. GovLab. ‘Data Stewards: Data 202–213. es: An intervention in surveillance Available at: https://www.nesta.org. of data’, Stanford Technology Law leadership to address 21st centu- Ilves, L. K., and Osimo, D., A capitalism?’, Surveillance & Society, uk/blog/new-ecosystemtrust/ (ac- Review, Vol. 22, 2019, pp. 1–46. ry challenges’. In: Medium. 2018. roadmap for a fair data economy. Vol. 15, No 5, 2017, pp. 626–639. cessed 18 June 2020). Van Asselt, M. and Renn, O., Available at: https://medium.com/@ Policy Brief, Sitra and the Lisbon Lehtiniemi, T. and Haapoja, J., Open Data Institute Data trusts: ‘Risk governance’, Journal of Risk TheGovLab/data-stewards-da- Council. Helsinki, 2019. ‘Data agency at stake: MyData Lessons from three pilots. Report, Research, Vol. 14, No. 4, 2011, pp. ta-leadership-to-address-21st-cen- Jasanoff, S., ‘Future imperfect: activism and alternative frames Open Data Institute, London. 2019. 431–449. tury-challenges-a5e555cd4235 Science, technology, and the imag- of equal participation’, New Media Available at: https://theodi.org/ Wolf, K. D., ‘Contextualizing norma- (accessed 18 June 2020). inations of modernity’. In: Jasanoff & Society, Vol. 22, No. 1, 2020, pp. article/odi-data-trusts-report/ (active standards for legitimate govern- Hall,W., and Pesenti, J., Growing S. and Kim S. H. (Eds.) Dreamscapes 87–104. cessed 20 June 2020). ance beyond the state’, In Grote, J. the artificial intelligence industry of Modernity: Sociotechnical Imagi- Mattioli, M. ‘The data-pooling Pazaitis, A., Kostakis, V. and R. and Gbikpi, B. (Eds.) Participatory in the UK. Independent report, UK. naries and the Fabrication of Power. problem’, Berkeley Technology Law Bauwens, M., ‘Digital economy Governance: Political and Societal 2017. University of Chicago Press, Chicago, Journal, Vol. 32, 2017, pp. 179–236. and the rise of open cooperativism: Implications. Leske und Budrich, Hardingens, J. and Wells, P., De- IL, 2015, pp. 1–33. Micheli, M., Blakemore, M., Ponti, The case of the enspiral network’, Opladen, 2002, pp. 35–50. fining a ‘data trust’. In: Open Data Kitchin, R., The Data Revolution: M., et al. The governance of data Transfer: European Review of La- Van Dijck, J., Poell, T. and De Institute Blog. 2018. Available at: Big Data, Open Data, Data Infra- in a digitally transformed European bour and Research, Vol. 23, No 2, Waal, M., The Platform Society: https://theodi.org/article/defin- structures and Their Consequences. Society. Report of the second Work- 2017, pp. 177–192. Public Values in a Connective World. ing-a-data-trust/ (accessed 18 June Sage, London, 2014. shop of the Digitranscope Project, Ruppert, E., Isin, E. and Bigo, D., Oxford University Press, Oxford, 2020). Kitchin, R. and Laurialt, T. ‘To- JRC114711, Publications Office of ‘Data politics’, Big Data & Society, 2018. Heeks, R., and Shekhar, S., ‘Data- wards critical data studies: Charting the European Union, Luxembourg, Vol. 4, No. 2, 2017, pp. 1–7.. Villani, C., For a meaningful arti- fication, development and marginal- and unpacking data assemblages 2018. Sandoval, M., ‘Entrepreneurial ficial intelligence: Towards a French ised urban communities: An applied and their work’. In: Thatcher, J., Eck- Morozov, E., and Bria, F., Rethink- activism? Platform cooperativism and European strategy, Mission data justice framework. Information’, ert, J. and Shears, A. (Eds.) Thinking ing the Smart City: Democratizing between subversion and co-opta- assigned by the Prime Minister Communication & Society, Vol. 22, Big Data in Geography. New Re- Urban Technology. Rosa Luxemburg tion’, Critical Sociology, Vol. 46, No. Edouard Philippe. 2018. No 7, 2019, pp. 992–1011. gimes, New Research. University Stiftung, New York, 2018. 6, 2020, pp. 801–817. Winter, J. S., and Davidson, E., Ho, C. and Chuangt, T. ‘Govern- of Nebraska Press, Lincoln/London, Mulgan, G. and Straub, V. The Scholz, T. Platform cooperativism. ‘Big data governance of personal ance of communal data sharing’. In: 2014, pp. 83–94.. new ecosystem of trust: How data Challenging the corporate sharing health information and challenges 58 59

Igor Calzada Seeing Platform and Data Co-operatives Through European Pan- demic Citizenship1

Introduction: European Pandemic Citizenship at Stake In Europe, many citizens will likely be unemployed during and probably as a result of the COVID-19 crisis (McKinsey, 2020; Parker, 2020). The coronavirus does not discriminate, yet it has unevenly distributed economic and social impacts across and within state borders by producing a new pandemic citizenship regime that exposes health, socio-economic, cognitive, and even digital vulnerabilities (Calzada, 2020a). By contrast, the COVID-19 pandemic has also shown that the digital platform economy can offer opportunities to continue working and earning even during times of crisis. But how can job quality be ensured for all platform workers while also creating further democratic socio-economic platformised alternatives to revert the algorithmic and data-opolitic (data oligopolies; Stucke, 2018) extractivist business-as-usual hegemonic paradigm (Barns, 2020; Belloc, 2019; De Marco et al., 2019; Dig- ital Future Society, 2019; Fairwork Foundation, 2020; Helberger et al., 2018; Kilhoffer et al., 2019; Lane, 2020; Riso, 2020; Taylor, 2020)?

Nominally, over the last decades, globalisation has led to a new class of global citizenship for workers (Calzada, 2020b). While the access to this global citizenship remains uneven, many have enjoyed unlimited freedom to move, work, and travel. However, COVID-19 has drastically slowed down this global citizenship regime and introduced a new level of ubiquitous vulnerabilities in global affairs by inciting a new pandemic citizenship regime in which citizens—regardless of their locations—share fear, uncertainty, and risks (Taylor, 2020). Furthermore, COVID-19 is deeply and pervasively relat-

1 This chapter draws on the following article: Calzada, I. ‘Platform and Data Co-operatives Amidst European Pandemic Citizenship’, Sustainability, Vol. 12, No 20, 2020, pp. 8309. doi: https://doi.org/10.3390/su12208309. 60 Seeing Platform and Data Co-operatives Through European Pandemic Citizenship Seeing Platform and Data Co-operatives Through European Pandemic Citizenship 61

Under these extreme circumstances,“ the European pandemic citizenship thus could be described as follows: the post-COVID-19 era, on the one hand, has Actually, the responses to this pan- also known as platform co-opera- this chapter refers to it as fostering dramatically slowed down several mundane routines demic emergency have varied ex- tives (Scholz, 2016; Schneider, 2018; social innovation among stakehold- tremely from location to location, Scholz and Schneider, 2017; 2015) ers in civil society (Moulaert and for citizens such as mobility patterns while, on the even within the same state. It is true and data co-operatives (Blassimme MacCallum, 2019). The COVID-19 other hand, it has exponentially increased demanding that the pandemic caused in many et al., 2018; Hardjono and Pentland, crisis has clearly shown that citi- new professional pressures, emotional fears, life countries a lockdown, which then 2019a; Hafen, 2019; Pentland et al., zens are highly dependent on data boosted online work and online de- 2019). Nonetheless, this is not the and the economic value it creates. uncertainties, algorithmic exposure, data privacy livery of goods via platforms, put- only resilient strategy adopted with- The COVID-19 crisis has thus led to concerns, health-related direct risks, and socio- ting further pressure on platform in the literature of data governanc an explicit, necessary revaluation in economic vulnerabilities depending eminently on the workers. But it also allowed many models. society of the roles of both govern- material and living conditions shared by a wide range communities and particularly civ- ment and citizens through extending ic groups and activists to respond There is a growing consensus in Eu- economic and socially innovative of citizens regardless of their specific resiliently, pushing ahead co-oper- rope that it is urgent for governments alternatives to digitization and da- geolocalisation in Europe. atives and reinforcing social capi- to start filling the same role in the tafication (Moulaert and MacCallum, tal. Among the resilience strategies information society that they have 2019). We are referring to co-opera- ” adopted by European governments, traditionally taken in the post-indus- tives (Beckett, 2019). ed to data and artificial intelligence patterns while, on the other hand, collective intelligence stemming trial society: not only fixing market (AI) governance issues, which expose it has exponentially increased de- from a proactive citizenship re- failure but also regulating the digi- citizens’ vulnerabilities in a potential manding new professional pressures, sponse has been highly considered tal power relations and supervising surveillance state and market (Hintz, emotional fears, life uncertainties, to further avoid dystopian meas- actual economic interplay among Dencik and Wahl-Jorgensen, 2017; algorithmic exposure, data privacy ures that could exacerbate existing stakeholders (Calzada, 2020c). This Kitchin, 2020; Morozov, 2020; Zuboff, concerns, health-related direct risks, social inequalities and techno-polit- means not just demanding fair tax 2019). Under these extreme circum- and socio-economic vulnerabilities ical vulnerabilities among European payments by the big tech companies stances, the European pandemic depending eminently on the mate- pandemic citizens (Bigo, Isin and and imposing fines when they violate citizenship thus could be described rial and living conditions shared by Ruppert, 2019). A particular collec- the GDPR or when they abuse their as follows: the post-COVID-19 era, a wide range of citizens regardless tive intelligence response emerging market power. There is much more— on the one hand, has dramatically of their specific geolocalisation in in Europe is the creation of digital and more fundamental issues—at slowed down several mundane rou- Europe. co-operatives (Borkin, 2019; Cher- stake that calls for government at- tines for citizens such as mobility ry, 2016; McCann and Yazici, 2018), tention beyond public intervention: 62 Seeing Platform and Data Co-operatives Through European Pandemic Citizenship Seeing Platform and Data Co-operatives Through European Pandemic Citizenship 63

New forms“ of co-operatives using digital technologies Rationale: At present, against the fragile back- (Calzada and Almirall, 2020)? How Hence, this chapter aims to shed can provide a framework to rethink, renew, Co-operatives as a Collective drop of the pandemic, European cit- does European citizenship (reacting light on how new forms of co-oper- and offer alternatives for how policies on digital Resilient Response to the izens working in tourism, the arts, and therefore self-organising) chal- atives using digital technologies can transformations and AI can help enhance pandemic Pandemic Crisis retail, and education and all infor- lenge data extractivism (Morozov, provide a framework to rethink, re- Historically, co-operatives have mal workers are the hardest hit (Kil- 2019) and surveillance capitalism? new, and offer alternatives for how citizens’ well-being and thus improve the been created when people work to- hoffer et al., 2019; Gramano, 2019). Is there any alternative response policies on digital transformations post-COVID-19 working conditions. gether—now with the help of tech- Further, marginalized low-income to big tech AI-driven data-opolies? and AI can help enhance pandemic nology—to respond with collective working-class citizens and immi- What will be next? citizens’ well-being and thus im- resilience to complex crises and grants are more adversely affected prove the post-COVID-19 working ” to mobilise a wider range of infor- than the average of the standard New possibilities for how Europe conditions of vulnerable and/or al- novative power-building strategies da, 2019). Thus, this chapter will mation, ideas, labour, and insights population (Eubanks, 2017). Income could advance towards and thus ready empowered citizens. This ob- rooted in cooperative ownership of focus on a citizen-driven resilient to address social structural trans- inequality is growing, confidence reinforce its democratic values—be- jective will be addressed through a digital platforms and data storage. post-COVID-19 response through formations through disruptive eco- in governments is eroding, and yond considering the citizen a sim- brief presentation of a taxonomy for Could we imagine Uber owned by ‘platform co-operatives’ and ‘data nomic innovations (Calzada, 2013). increasingly more people are em- ple resource—have already been platform and data co-operatives, as the drivers or Twitter being owned co-operatives’. Both ‘platform co-op- The co-operative movement began bracing populism (Dyer-Witherford, claimed in the widely spread mani- evidenced by 155 ongoing cases. by its users? Ultimately, advocates eratives’—which address alterna- in the UK and France in the 19th 2020). Workers may lose power and festo ‘#DemocratizingWork: Democ- of platform and data co-operatives tive communitarian business/social century. Remarkably, though, sev- a sense of agency over their lives ratize, Decommodify, Remediate’ As the concentration of big tech suggest that a shift from a sharing models—and ‘data co-operatives’— eral unique regionally rooted expe- and consequently their own data (#DemocratizingWork, 2020) signed companies is accelerating, plat- economy to a genuinely participa- which aim to customize protected riences with strong communitarian because the free market has been by relevant academics worldwide. form and data co-operatives are tory, democratically owned economy data stores for their members while identity have flourished in Europe allowed to develop into a data-opo- This manifesto is entirely aligned still challenging surveillance capi- might be possible (Scholz, 2016). generating a non-profit social val- since then, such as the Mondragon ly without regulatory frameworks with the direction of this chapter in- talism; they might equip citizens to ue from doing so—share the same case in the Basque Country (Spain) or rules (Delacroix and Lawrence, sofar as it considers the importance succeed and build an alternative as The ongoing post-General Data Pro- underlying principles: citizen-centric in the 1950s (Bengu, 2018; Clamp 2019). To this end, how can working of empowering citizens in their en- cooperative platform entrepreneurs tection Regulation (GDPR)/COVID-19 collective ownership, decentralised and Alhamis, 2010; Ellerman, 1984; citizens organize, regain control of vironments by owning data, which or activists in the fast-growing gig Europe may require reshuffling the self-governance, transparency, and Gupta 2014; Heales et al., 2017) their data, and participate in build- potentially fosters their co-oper- economy (Alosi, 2016; Hayes, 2019; way data is affecting active citizens’ offering an alternative to platform and the Emilia Romagna case (Italy) ing socio-economic alternatives to ativisation of their work through Lutz, 2019). They might allow mem- work and life and the overall way capitalism (Bastani, 2019; Beard- in the 1970s (Battilani and Zamag- alter the existing data governance more democratic and network-driv- bers of the co-operative to analyse the digital economy might function man, 2012; Beckett, 2019; Blok et ni, 2012; Borzaga and Galera, 2012; extractivism to protect pandem- en forms of organisation (Edenfield, and get involved with a generation more democratically and more local- al., 2017; Como et al., 2016; Srnicek, Gonzales, 2010). ic European citizens’ digital rights 2019). of citizens experimenting with in- ly rooted in cities and regions (Calza- 2017). 64 Seeing Platform and Data Co-operatives Through European Pandemic Citizenship Seeing Platform and Data Co-operatives Through European Pandemic Citizenship 65

Discussion: Co-operatives and reasons, clearly underestimating the with the pandemic citizenship re- have long fretted over information the profit to members. In the glob- mind: keep the micro-targeting go- Pandemic Citizenship economic dimension of a sustaina- gime emerging across state borders security, devising passwords to ac- al market today, while big data and ing and the micro-payments flowing. Overall, co-operatives could be con- ble business model in a global com- in Europe (Calzada, 2020a). Further, cess passwords, fearful they might AI do not naturally favour non-mar- As a result, little thought has gone sidered a citizen movement which petitive context. This approach leads this chapter argues that at pres- be hacked or exposed. People are in- ket activities, they do make it easier into building digital technologies accounts for 130,000 enterprises to a high risk of failure beyond the ent, such a citizenship regime might secure in their jobs, homes, and rela- to imagine a post-neoliberal world that produce macro-level anony- in Europe in basically all economic purely altruistic, volunteer, civilian, actually be the seed for creating tionships and on social media. They where production is automated mous insights about the collective sectors, with 127 million members, and grassroots-driven initiatives far post COVID-19 co-operative forms are also insecure about themselves. and technology underpins universal behaviour of non-consumers. Dig- more than 4 million employees, and removed from formal professional in the digital economy and society Co-operatives might thus contribute healthcare and education for all in ital platforms, as they are known nearly €990 billion annual turno- entrepreneurial institutions. The sec- that aim to protect citizens’ digital to a more secure economy and so- the post-COVID-19 era—a world hegemonically today, are the sites ver (Cooperatives Europe, 2020). ond regime appears in the Eastern rights, such as platform and data ciety for everyone. It is a challenge where abundance is shared by peers, of individualised consumption, not However, the understanding, prac- European co-operative tradition. In co-operatives. Obviously, co-opera- Europe cannot afford to ignore (Tay- not appropriated (Bastani, 2019; Dy- of mutual assistance and solidarity tices, and perceptions of co-opera- essence, the communist legacy left a tives existed before COVID-19, but lor, 2020). er-Witheford, 2019; Riso 2020). But (Sandoval, 2019; Siapera and Papa- tives vary significantly from state generalised distrust of the co-opera- the pandemic has accelerated the in less idealistic terms, it could be ar- dopoulou, 2016). Thus, could digital to state—particularly now in the tive concept that is still linked to the willingness of citizens to learn more However, the far-reaching aspira- gued that big data and AI could lead platform innovation in Europe be led post-COVID-19 era—in showing a memory of past communist collec- about this particular form of organ- tions of co-operatives are based on to the entirely opposite result, may- by an asymmetric network of co-op- substantial path dependency on tives and has been clearly replaced ising the digital economy. the idea that digital revolutionaries be even more likely than the libertar- erative SMEs (De Marco et al., 2019; contextual factors derived from the by a general individualistic prefer- should reshape everything but the ian one. Today’s debate on the right Helberger, Pierson and Poell, 2018)? history and from current digital and ence for private ownership of assets Hence, at this stage, any approach central institution of modern life: the technological response to COVID-19 This emerging European pandemic socio-political transformations. As over sharing or direct exchange with to citizenship in Europe needs to be market. The early digital revolution- regarding contact tracing apps and citizenship regime is currently shap- a generalisation of historical trends, other peer citizens. Not surprisingly, analysed through the lenses of the aries of the 1960s in Silicon Valley the economic crisis revolves around ing the potential for the formation two distinct citizenship regimes though, one could argue that this aftermath of COVID-19. Citizenship argued for a nirvana Internet free the trade-offs between privacy and of platform and data co-operatives have framed and shaped perception preference applies to Western Eu- encompasses not only identification of government intervention where public health (Kitchin, 2020) and the formation (Calzada, 2020a). of co-operatives. First, in the West- rope, particularly since the 1980s. and belonging but also power, con- everyone would be equally happy: need to promote innovation by start- ern European co-operative tradition, trol, and techno-politics. Long be- however, they opened the door to ups, respectively. Why are there no Since March 2020, coronavirus has there is sometimes a concern that The notion of co-operatives (Cas- fore COVID-19 swept the globe, inse- the data-opolies of today. Co-op- other options? It is because we have mocked immigration controls, biom- many small and local platform and tilla-Polo and Sánchez-Hernández, curity and social vulnerabilities were eratives are working in the market. let digital platforms and telecom etrics, digital surveillance, and every data co-operatives seem to be ap- 2020) in the digital era—eminently a already ubiquitous. Countless people If they were not, they would not operators treat our entire digital uni- kind of data analytics, and struck proaching the co-operative model transnational and resilient phenom- have faced housing, health, and food survive: they make money, but the verse as their fiefdom (Khan, 2017)? hardest—thus far—in the richest, mainly for ideological and value enon—currently can be associated insecurity. Meanwhile, online, people difference is that they redistribute They run it with just one goal in most powerful states of in the world. 66 Seeing Platform and Data Co-operatives Through European Pandemic Citizenship Seeing Platform and Data Co-operatives Through European Pandemic Citizenship 67

Consequently, the significance of Eu- ing times for European citizenship ★ Produser-led Platforms: Users ropean citizenship might be rapidly be seen as an opportunity to foster and producers own the platform, shifting through a sort of pandemic digital co-operatives in Europe in through which producers can sell citizenship adjustment, with conse- pursuit of a Tech New Deal to allow their work. quences for citizens depending on citizens and communities to own ★ Multistakeholder/Community the state they call home and their and govern their own data and plat- Platforms: living conditions. What might be forms (Bauwens and Pazaitis, 2018; ☆ City-owned platforms: This model called a shared pandemic citizen- Calzada, 2013; Hardjono and Pent- could involve collaboration be- ship—citizens in Europe sharing ex- land, 2019a, 2019b; Pentland et tween many cities, which would Table 4.1. Definitions: Platform Co-operatives and Data Co-operatives actly the same fears—seems to be al., 2019; Schneider, 2020; Scholtz, pool their resources to create a here to stay. This trend has different 2016; Scholz and Schneider, 2015, software platform for any kind levels of techno-political implica- 2017)? of service: short-term rentals, tions as it intersects with another There is a diverse set of taxonomies Finally, platform co-operatives con- utilities, and so on (e.g., Cities Co- global trend—how algorithms are Taxonomy: (Scholz, 2016): sist of four typologies: alition for Digital Rights – CCDR, increasingly shaping everyday life. Shedding Light on Platform ★ Generally speaking, platform co- ★ Consortia Worker Platforms: 2020). and Data Co-operatives operatives focus essentially on ☆ Co-operatively owned online ☆ Co-operatives from within: In this Arguably, the current pandemic crisis According to Bauwens (Bauwens and business models and the social labour brokerages and market model, workers/citizens from a and democracy are deeply related to Vasilis, 2014; Bauwens, Kostakis and impact of their activity, while places: In this most common sharing economy platform like data governance issues, exposing Pazaitis, 2019; Bauwens and Pazai- data co-operatives mutualise co-operative platform, workers/ Uber use the technical infrastruc- citizens’ vulnerability in a potential tis, 2018) and Scholz (2016), data and store data without directly citizens own the company, re- ture of the company to run their surveillance state (Calzada, 2020b; co-operatives can be seen as a sub- focusing on the economic inter- ceive dividends and have a voice own enterprise. Worker co-opera- Lucas, 2020). Should European gov- category of platform co-operatives. play of data. in running the company. tives form inside the belly of the ernments protect citizens from being But generally speaking, data co-op- ★ Regarding the flow, platform ☆ Union-backed labour platforms: sharing economy (Mensakas). infected even if doing so might mean eratives arguably focus merely on co-operatives manage labour Unionised workers/citizens can ★ Data co-operatives can be con- establishing a new digital non-pri- data stores, while platform co-op- exchange and distribute content create their own companies as sidered a sub-typology of plat- vacy norm? Will this pandemic crisis eratives revolve around the whole while aggregating the data of a a result of the collaboration form co-operatives—also known become an algorithmic crisis, with se- business model of workers, services, group of members/citizens. between unions and workers. as a data consortia platform. rious side-effects for governments in and products, which also includes Europe? Could these rapidly chang- data. 68 Seeing Platform and Data Co-operatives Through European Pandemic Citizenship Seeing Platform and Data Co-operatives Through European Pandemic Citizenship 69

To further simplify analysis, the four distribution and revolves around co-operativising community ser- Table 4.3 illustrates a classification typologies of platform co-opera- co-operativising the outcome, re- vices (stemming from healthcare, of each platform co-operative ac- tives are defined as follows: sulting in an exchange between delivery riders, media, rental, cording to its typology (https://ioo. ★ Worker: This typology refers to users and producers (stemming housing, land, etc.). coop/directory). Several cases could the flow of labour exchange and from culture, agriculture, food, ★ Data (this fits into data co-op- be included in multiple typologies, revolves around co-operativising software, websites, hosting, eratives): This typology refers to but the identification process aimed work (stemming from mobility start-up support, videoconfer- the flow of data aggregation and to include each case in only one ty- services). encing, etc.). revolves around co-operativising pology. ★ Produser (as a merge of users ★ Multistakeholder: This typolo- and mutualising data (particular- and producers): This typology gy refers to the flow of content ly data related to finance, health, refers to the flow of content distribution and revolves around security, etc.).

Table 4.3. Case Identification by Typology Table 4.2. Taxonomy for Platform Co-operatives and Data Co-operatives Source: elaborated from https://ioo.coop/directory 70 Seeing Platform and Data Co-operatives Through European Pandemic Citizenship Seeing Platform and Data Co-operatives Through European Pandemic Citizenship 71

Final Remarks First, expert analyses, case identi- iments aligned with data donation References Bauwens, M., and Pazaitis, A., This ongoing exploratory research fications, and preliminary fieldwork and altruism (European Commission, Alosi, A., ‘Commoditized Workers: P2P Accounting for Planetary Surviv- intended to decipher the rationale action research have demonstrat- 2020). Case Study Research on Labor Law al, P2P Foundation, Guerrilla Foun- behind platform and data co-oped that the post-COVID-19 world Issues Arising From a Set of on-De- dation and Schoepflin Foundation, eratives amid a new citizenship re- is reigniting the need to reactivate Third, ultimately, initiatives around mand/Gig Economy Platforms’, Com- 2018. gime presented by this chapter as European civil societies by further platform and data co-operatives parative Labor Law & Policy Journal, Beardman, S., Altruism and the Ex- European pandemic citizenship. Fur- experimentation with digital so- need to find their own strategic Vol. 37, No 3, 2016, pp. 653–690. perimental Data on Helping Behavior, thermore, this ongoing exploratory cio-economic innovations, such as pathway amidst the digital and so- Barns, S., Platform Urbanism: Ethical Theory and Moral Practice, research has provided the point of platform and data co-operatives— cial economy policy agenda of the Negotiating Platform Ecosystems Vol. 15, No 4, 2012, 547–561. departure for leading us to new in- but marginally and at a small scale. European Commission. The ongoing in Connected Cities, Springer Singa- Beckett, A., The New Left Eco- sights on platform and data co-op- institutional arrangements around pore, Singapore, 2020. nomics: How a Network of Thinkers eratives’ advancements in the near Second, consequently, procurement H2020-Smart Cities and Commu- Bastani, A., Fully Automated Lux- Is Transforming Capitalism. 2019. future. This chapter concludes with and public incentives are required to nities have already gathered 17 ury Communism, Verso, London, Available at: https://www.theguardi- three aspects that frame potential push ahead, enhance, and reinforce lighthouse projects (encompassing 2019. an.com/news/2019/jun/25/the-new- future research and a policy agenda platform and data co-operatives 46 lighthouses and 71 fellow cities), Battilani, P., and Zamagni, V., left-economics-how-a-network-of- for platform and data co-operatives. beyond extremely marginal exper- and the still-experimental Digital ‘The Managerial Transformation of thinkers-is-transforming-capitalism Innovation Hubs entirely follow the Italian Co-Operative Enterprises (accessed 1 Dec 2020). new European Data Strategy. In all 1946–2010’, Business History, Vol. Belloc, F., Why Isn’t Uber Work- likelihood, these initiatives can firm- 54, No 6, 2012, pp. 964–985. er-Managed? A Model of Digital Plat- ly aid and foster city-regional connec- Bauwens, M., and Vasilis K., form Cooperatives. CESifo Working tions between social entrepreneurs ‘From the Communism of Capital to Papers, 7708. 2019. and urban activists seeking to scale Capital for the Commons: Towards Bengu, R., What Can South Africa up platform and data co-operatives an Open Cooperativism’, Triple C: Learn from Mondragon? The Answer to better articulate the uneven Euro- Communication, Capitalism and Cri- Lies in Community Ownership. Avail- pean pandemic citizenship. tique, Vol. 12, No 1, 2014, 356–361. able at: https://www.dailymaverick. Bauwens, M., Kostakis, V., and co.za/article/2018-10-26-what-can- Pazaitis, A., Peer to Peer: The Com- south-africa-learn-from-mondra- mons Manifesto, University of West- gon-the-answer-lies-in-community- minster Press, London, 2019. ownership/ (accessed 1 Dec 2020). 72 Seeing Platform and Data Co-operatives Through European Pandemic Citizenship Seeing Platform and Data Co-operatives Through European Pandemic Citizenship 73

Bigo, D., Isin, E., & Ruppert, E., Handbook on Social Innovation: Geopolitics. Edward Elgar, Chel- Cities Coalition for Digital Delacroix, S., & Lawrence, N.D., No 3, 1984, pp. 272–294. DOI: Data Politics, Routledge, London, Collective Action, Social Learning tenhman, UK and Northampton, Rights – CCDR, 2020. Avail- ‘Bottom-up Data Trusts: Disturbing 10.1080/00346768400000025. 2019. and Transdisciplinary Research. MA, USA. 2020a. DOI: 10.13140/ able at: https://citiesfordigital- the ‘One Size Fits All’ Approach to Eubanks, V., Automating Inequali- Blasimme, A., Vayena, E., and Edward Elgar, Cheltenhman, UK RG.2.2.17301.6832/1. rights.org/city-examples-digi- Data Governance, International Data ty: How High-Tech Tools Profile, Po- Hafen, E., ‘Democratizing Health and Northampton, MA, USA, 2013, Calzada, I., Will Covid-19 Be the tal-rights-times-covid-19 Privacy Law, 2019. DOI: 10.1093/ lice, and Punish the Poor. St. Martin’s Research through Data Coopera- pp. 219–229. DOI: 10.13140/ End of the Global Citizen? Apolitical. Clamp, C. A., and Innocen- idpl/ipz014. Press, NYC, 2017. tives’, Philosophy & Technology, Vol. RG.2.2.30748.69766. 2013. Available at: https://apolitical.co/ tus A., ‘Social Entrepreneurship DemocratizingWork: Democratize, European Commission, Data Gov- 31, 2018, pp. 473–479. Calzada, I., ‘Benchmarking Future en/solution_article/will-covid-19- in the Mondragon Co-Operative Decommodify, Remediate. Available ernance Act, Available at: https:// Blok, A., Courmont, A., Hoyng, R., City-Regions beyond Nation-States’, be-the-end-of-the-global-citizen. Corporation and the Challenges at: www.democratizingwork.org ec.europa.eu/digital-single-market/ Marquet, C., Minor, K., Nold, C., Regional Studies, Regional Science, 2020b of Successful Replication’, The (accessed 1 December 2020). en/news/proposal-regulation-euro- and Young, M. ‘Data Platforms and Vol. 2, No 1, 2015, pp 351–362. DOI: Calzada, I. ‘Replicating Smart Cit- Journal of Entrepreneurship, Vol. Digital Future Society, The Future pean-data-governance-data-gov- Cities’. Tecnoscienza: Italian Journal 10.1080/21681376.2015.1046908, ies: The City-to-City Learning Pro- 19, No 2, 2010, pp. 149–177. DOI: of Work in the Digital Era: The Rise ernance-act (accessed 1 December of Science & Technology Studies, 2015. gramme in the Replicate EC-H2020- 10.1177/097135571001900204. of Labour Platforms. DFS, Barcelona, 2020). Vol. 8, No 2, 2017, pp. 175–219. Calzada, I., ‘Technological Sover- SCC Project’, Smart Cities, Vol. 3, Como, E., Mathis, A., Tognetti, M., 2019. Fairwork Foundation. How Is Borkin, S., Platform Co-Operatives eignty: Protecting Citizens’ Digital No 3, 2020c, pp. 978–1003. DOI: and Rapisardi, A. Cooperative Dyer-Witheford, N., ‘Left Populism the Platform Economy Responding - Solving the Capital Conumdrum. Rights in the AI-Driven and Post-GD- 10.3390/smartcities3030049. Platforms in a European Landscape: and Platform Capitalism’, Triple C: to COVID-19? Open Democracy, NESTA, London, 2019. PR Algorithmic and City-Regional Calzada, I., and Almirall, E., An Exploratory Study, ISIRC Confer- Journal for a Global Sustainable Available at: https://www.open- Borzaga, C., and Galera, G., ‘The European Realm’, Regions eZine, Vol. ‘Data Ecosystems for Protecting ence. 2016. Information Society, Vol. 18, No 1, democracy.net/en/oureconomy/ Concept and Practice of Social En- 4, 2019, DOI:10.1080/13673882.20 European Citizens’ Digital Rights’, Cooperatives Europe Available at: 2020, pp. 116–131, 2020. how-platform-economy-respond- treprise. Lesson from the Italian 18.00001038. Transforming Government: People, https://coopseurope.coop/resources/ Edenfield, A.C., ‘Research in ing-covid-19/ (accessed 1 December Experience’, International Review of Calzada, I., ‘Emerging Citizenship Process and Policy, Vol. 14, No 2, news/covid-19-coop-response (ac- Cooperatives: Developing a Po- 2020). Social Research, Vol. 2, No 2, 2012, Regimes and Rescaling (European) 2020, pp. 133–147. DOI: 10.1108/ cessed 1 December 2020). litically Conscious Research Gonzales, V. ‘Italian Social Coop- pp. 85–102. Nation-States: Algorithmic, Liquid, TG-03-2020-0047. 2020 De Marco, C. E., Di Minin, A., Methodology’, Technical Com- eratives and the Development of Calzada, I. ‘Knowledge Building Metropolitan and Stateless Citi- Cherry, M., Legal and Governance Marullo, C., and Nepelski, D., Dig- munication Quarterly, Vol. 28, Civic Capacity: a Case of Cooperative and Organizational Behavior: The zenship Ideal Types’. In: Moisio, S., Structures Built to Share: The Rise ital Platform Innovation in European No 4, 2019, pp. 376–390. DOI: Renewal?,’ Affinities: A Journal of Mondragon Case from a Social In- Koch, N., Jonas, AEG, Lizotte, C., of Platform Co-Operatives. Legat SMEs: An Analysis of Sme Instrument 10.1080/10572252.2019.1621388. Radical Tehory, Culture, and Action, novation Perspective’. In: Moulaert, and Luukkonen, J., (Eds.) Hand- Studies Research Paper Series, Saint Business Proposals and Case Studies, Ellerman, D.P., ‘Entrepreneurship Vol. 4, No 1, 2010, pp. 225–251. F., MacCallum, D., Mehmood, A. and book on the Changing Geogra- Louis University School of Law. Publications Office of the European in the Mondragon Cooperatives’, Gupta, C. ‘The Co-Operative Model Hamdouch, A. (Eds.) International phies of the State: New Spaces of 2016. Union, Luxembourg, 2019. Review of Social Economy, Vol. 42, as a ‘Living Experiment in Democra- 74 Seeing Platform and Data Co-operatives Through European Pandemic Citizenship Seeing Platform and Data Co-operatives Through European Pandemic Citizenship 75

cy’, Journal of Co-operative Organ- tem Case Study, The Young Founda- Toolkit Policy Note, No. 1. Availa- ian.com/commentisfree/2020/ Sandoval, M., ‘Entrepreneurial gal Studies Research Paper No. 349. ization and Management, Vol 2, No tion, London, 2017. ble at: https://goingdigital.oecd. apr/15/tech-coronavirus-sur- Activism? Platform Cooperativism 2018. Available at: https://papers. 2, 2014, pp. 98–107. DOI: 10.1016/j. Helberger, N., Pierson, J., and org/toolkitnotes/regulating-plat- veilance-state-digital-disrupt (ac- between Subversion and Co-Opta- ssrn.com/sol3/papers.cfm?abstract_ jcom.2014.09.002. Poell, T., ‘Governing Online Plat- form-work-in-the-digital-age.pdf cessed 1 December 2020). tion’, Critical Sociology. 2019. DOI: id=3144045 (accessed 1 December Hafen, E. ‘Personal Data Cooper- forms: From Contested to Coopera- (accessed 1 December 2020). Morozov, E., ‘Digital Socialism?’ 10.1177/0896920519870577. 2020). atives – a New Data Governance tive Responsibility’, The Information Lucas, E., Pandemic Democracy, New Left Review, 116/117, March & Schneider, N., ‘Tech New Deal: Poli- Taylor, A. The People’s Platform: Framework for Data Donations and Society, Vol. 34, No 1, 2018, pp. 1–14. Available at: https://www.cepa.org/ June, 2019. cies for Community-Owned’. 2020. Taking Back Power and Culture in Precision Health’. In: Krutzinna, J., Hintz, A., Dencik, L., and pandemic-democracy (accessed 1 Mos, E., ‘Digital platforms and the Schneider, N., ‘An Internet of the Digital Age, Metropolitan Books, and Floridi, L. (Eds.) The Ethics of Wahl-Jorgensen, K., ‘Digital cit- December 2020). (re)organization of solidarity’. Avail- Ownership: Democratic Design NYC, 2020. Medical Data Donation. Springer In- izenship and surveillance society’, Lutz, C., ‘Digital inequalities in the able at: https://platformlabor.net/ for the Online Economy.’ The Zuboff, S. The Age of Surveillance ternational Publishing, Cham, 2019, International journal of Communica- age of artificial intelligence and blog/digital-platforms-and-the-re- Sociological Review Vol. 66, No Capitalism: The Fight for a Human pp. 141–149. tions, Vol. 11, 2017, pp. 731–739. big data’, Human Behavior and organization-of-solidarity (accessed 2, 2018, pp. 320–340. DOI: Future at the New Frontier of Power, Hardjono, T., and Pentland, A., Kilhoffer, Z., Pieter de Groen, W., Emerging Technologies, Vol. 1, No. 2, 1 December 2020). 10.1177/0038026118758533. Profile, London, 2019. ‘Data Cooperatives: Towards a Foun- Lenaerts, K., Smits, I., Hauben, H., 2019, pp. 141–148. DOI: 10.1002/ Moulaert, F., and MacCallum, D. Scholz, T., Platform Cooperativism: dation for Decentralized Personal Waeyaert, W., Giacumacatos, E., hbe2.140. Advanced introduction to social in- Challenging the Corporate Sharing Data Management’. arXiv e-prints. Lhernould, J-P., and Robin-Oliv- McCann, D., and Yazici, E. Dis- novation, Edward Elgar, Cheltenham, Economy, Rosa Luxemburg Stiftung, 2019a ier, S. Study to Gather Evidence on rupting Together: The Challenges 2019. New York, 2016. Hardjono, T., and Pentland, A., the Working Conditions of Platform (and Opportunities) for Platform Parker, M., Life after COVID-19: The Scholz, T. and Schneider, N., Ours ‘Empowering Artists, Songwriters Workers, European Commission, Co-Operatives, NEF, London, 2018. Other Side of Crisis, Bristol Universi- to Hack and to Own, OR Books, NY, & Musicians in a Data Cooperative Brussels, 2019. McKinsey, The future of work. ty Press, Bristol, 2020. 2017. through Blockchains and Smart Con- Kitchin, R. ‘Civil Liberties or Availablet at: https://www.mckinsey. Pentland, A., Hardjono, T., Siapera, E., & Papadopoulou, L., tracts’, arXiv:1911.10433v1. 2019b. Public Health, or Civil Liberties com/featured-insights/future-of- Penn, J., Colclough, C., ‘Entrepreneurialism or Coopera- Hayes, R. ‘Worker-Owned Apps and Public Health? Using Sur- work/the-future-of-work-in-eu- Ducharmee, B., and Mandel, L., tivism?’, Journalism Practice, Vol. Are Trying to Fix the Gig Economy’s veillance Technologies to Tackle rope?cid=soc-web# (accessed 1 Data Cooperatives: Digital Empow- 10, No 2, 2016, 178–195. DOI: Exploitation’. Motherboard, Tech by the Spread of Covid-19’, Space December 2020). erment of Citizens and Workers, MIT 10.1080/17512786.2015.1125760. Vice. 2019. and Polity, 2020, pp. 1–20. DOI: Morozov, E., The Tech ‘Solutions’ Connnection Science, Boston, 2019. Stucke, M.E. ‘Should We Be Con- Heales, C., Hodgson, M., and 10.1080/13562576.2020.1770587. for Coronavirus Take the Surveil- Riso, S., Digital Age: Mapping the cerned about Data-opolies?’ 2 Rich, H., Humanity at Work: Mon- Lane, M. Regulating platform work lance State to the Next Level’. Countours of the Platform Economy, Georgetown Law Technology Review, dragon, a Social Innovation Ecosys- in the digital age, Going Digital Available at: https://www.theguard- Eurofound, Dublin, 2020. Vol. 275, University of Tenesse Le- 76 77

Marisa Ponti and Massimo Craglia Citizen-generated data for policy: A review of EU projects

Introduction The rapid embedding of digital technologies into everyday practices and spaces has made it possible for citizens to generate and share data about local community-relevant problems, ranging from air quality measurement to reporting street problems. The growth of data generated by citizens can give the public sector new opportunities for addressing critical social and econom- This chapter“ ic issues and inform policies. For example, mobile-equipped citizens can com- provides an overview plement digital sensors for real-time reporting and situational awareness, providing public authorities with opportunities for data-driven decision mak- of European projects ing, improved performance management, and heightened accountability (Lin- involving citizen- ders, 2012). Thus, digital technologies constitute the socio-technical means generated data citizens can use to participate on issues that affect their lives, for example by producing data and statistics and creating a new political subjectivity (Rup- (CGD). pert, 2018). This chapter provides an overview of European projects involving citizen-generated data (CGD). We aim to understand how CGD makes it possi- ” ble to experiment with new forms of public participation, rethink relationships between citizens and local governments and explore new emerging roles for citizens and local governments. 78 Citizen-generated data for policy: A review of EU projects Citizen-generated data for policy: A review of EU projects 79

CGD projects“ Data“ often result from is a social Citizen-Generated Data Projects and depend on to complement data granularity to Governance of Data and Data for Digital technologies generate oppor- and political practice CGD has been defined as “data that partnerships between institutional sources. By providing a Governance tunities for producing, managing and engaging participants people or their organisations pro- means for citizens who want to make Involving citizens in producing data using data that citizens may want to citizens and several who are not only duce to directly monitor, demand or their voices heard, CGD projects can can help to experiment with new take, resulting in different forms of drive change on issues that affect organisations, make valuable contributions to un- forms of participation in data pro- data governance. Technologies are “objects of data”, them. It is actively given by citizens, including civil society derstanding and addressing social duction and its governance that, in inextricable components of data about which data is providing direct representations of and economic problems. As report- turn, may lead to new types of rela- governance, as also explicitly indi- organisations, produced, but also their perspectives, and an alterna- ed by Lämmerhirt, Jameson and tionships between citizens and pub- cated in the definition of this concept tive to datasets collected by govern- community-based Prasetyo (2016), CGD projects often lic institutions (Ruppert, Isin & Bigo, provided by the British Academy and “subjects of data”, ments or international institutions” organisations, result from and depend on partner- 2017). This involvement may lay Royal Society (2017). In their report, as they drive the (DataShift, 2015, p. 1). This defini- public sector, and ships between citizens and several the ground for expanding the public they refer to data governance as the how and why data tion denotes two main characteris- organisations, including civil socie- sphere and may create an opening processes of governing data man- businesses. is produced. tics. One is the voluntary participa- ty organisations, community-based for thinking about what Castells agement, data use, and the technol- tion of the public in collecting data organisations, public sector, and (2008) called the “Network State” ogies involved in these processes “to on community-relevant problems. ” businesses. These partners play a that is “characterised by…greater di- inform the extent of confidence” in ” CGD can be considered a form of surfacing and responding to issues decisive role to provide resources, versity in the relationship between these processes (p. 1). of data, involved in the governance user-generated data collected ex- that affect them. support, and knowledge to citizens. governments and citizens” (p. 88). of data, are deployed taking into plicitly for tackling those problems, In return, they can tap into the data This argument resonates with the Dealing with data is not just about account specific objectives, needs such as improving local infrastruc- CGD efforts are typically organised generated by citizens. Thus, citizens more recently argued-for need for producing data. Much discourse on and capacities. Therefore, data “con- tures, tracking environmental issues, as projects. Motivations for setting and their partners can gain mutual governance systems involving mul- data governance focuses on the struct” the world following different or collecting spatial data. The other up a CGD project can be very dif- benefits from the application of CGD ti-stakeholder collaboration (British capacity of data for knowing and visions and interests (Grey, Gerlitz, one is the creation of alternative da- ferent. For example, citizens can be approaches. Academy and Royal Society, 2017, p. representing the world. While data & Bounegru, 2018). Second, data tasets that can complement official stirred by a lack of accurate data 55). This collaboration implies that certainly is a representational re- is a social and political practice data, offering the opportunity for from the public sector or a lack of not only the public sector, but also source, it can also shape the way we engaging participants who are not citizens to make their voices heard trust between public sector author- businesses, academia, and citizens see and think about the world (Gray only “objects of data”, about which within democratic processes at the ities and citizens. In other cases, can provide data publicly (Meijer & & Marres, 2018). Two assumptions data is produced, but also “subjects local level of government (DataShift, citizens collect data to raise aware- Potjer, 2018) to address common underpin this performative view of of data”, as they drive the how and 2015). CGD can help gain new per- ness of a topic that does not receive concerns and meet public good. data. Fist, data is not neutral. The why data is produced (Ruppert, Isin spectives, involving communities in enough attention from institutions or creation, extraction, and analysis & Bigo, 2017). CGD can expand what 80 Citizen-generated data for policy: A review of EU projects Citizen-generated data for policy: A review of EU projects 81

gets measured, how, and for what In the following sections, we first The key-features of purpose. Data generation can cre- present a brief overview of the European CGD projects ate opportunities for citizens to play main features of 18 European pro- Purpose of data collection a more active role in examining a jects CGD in five areas: environment, The sampled projects are distribut- limited scale in time and place, al- situation and taking action, such as, public health, energy, transport, and ed across countries as follows: Ger- though several of them had the po- for example, in local development infrastructure. Then, we summarise many (two cases), Netherlands (four tential for replication and expansion. and collaborative strategies for findings from five interviews related cases), Belgium (one case), Swit- A slight majority of the 18 sampled monitoring, auditing, planning and to data governance, setup and de- zerland (one case), Italy (one case), projects aimed at collecting data decision-making (Lämmerhirt, Gray, velopment of projects, project im- Spain (four cases), and UK (five cas- for environmental monitoring and Venturini, & Meunier, 2019). Läm- pact, project sustainability, and CGD es). Half of the projects were ongo- environmental decision making, as merhirt and colleagues (2019) not- use by the public sector. ing, while eight where a pilot and shown in Table 5.1. ed that CGD projects at a local level one completed. They tended to be of could help facilitate engagement of citizens. For example, concerning Examine the effects of climate change To select relevant cases of CGD projects, we used the following inclusion the achievement of the Sustainable Identify where building new homes criteria (Ponti and Craglia, 2020): Development Goals, localisation has The project had to be about data actively generated by citizens around Improve local infrastructure been acknowledged to connect the issues concerning them. Improve transportation broad global dimensions of SDGs The project had to involve citizens generating data in partnerships with Make a neighbourhood a healthier place with their contextual relevance. This the public sector and community-based organisations at the local level. Map accessibility of public places for disabled connection would help engage resi- The collection of data had to serve the public good primarily (e.g., collect data Map cycling conditions dents because it could provide rele- on air and water quality), and inform policy and create public services. Map favourite local places to influence development plans vance to the intended actions, add The project had to be at a local scale (neighbourhood, municipality, and Map quiet areas value, and create local ownership. city-scale) in the European Member States. Measure air quality The technologies used by citizens to collect data had to be digital devices Measure noise pollution (e.g., cellular telephones, calculators, sensors). Measure odour pollution Besides collecting information about such projects through desk research, Report street problems we also conducted interviews with five people responsible for the devel- Share info about policy proposals opment of five of the above projects. Box 1. Selection of Citizen-Generated Table 5.1. Purpose of data collection Use renewable energies Data projects in a nutshell in the selected projects 82 Citizen-generated data for policy: A review of EU projects Citizen-generated data for policy: A review of EU projects 83

Categories of projects and types egorised as ‘crowdsourcing’ because apps, and locating street problems Use of CGD by the public sector service planning and facilities im- of data we stressed their reliance on a large on maps, among the others. In re- Regarding the use of CGD by the provement, or reporting and plan- Figure 5.1 shows the relative fre- number of contributors. Citizens lation to forms of engagement, cit- public sector, we could find this in- ning environmental actions. Table quency of the primary categories of collected different types of data, as izens were enrolled as sensors, mon- formation for ten of the sampled 5.2 shows how CGD in the sampled the selected projects. Projects aimed shown in Figure 1, using a variety itors, reporters, observers, platform projects. Public sector organisations projects was used by public sector at measuring air quality or noise pol- of methods and devices, including users and co-creators of sensors. Table 5.2. Reported uses of CGD by public organisations were interested in CGD for public organisations. lution were categorised as ‘passive sensors, online platforms, mobile Reporting information sometimes sensing’ because they depend on phones, and maps. Data generation occurred in a very structured way to Project Use of CGD Public Organisation participants using a resource that involved automatic sensing, plot- ensure quality and consistency. Hush City Partial update of the Berlin Plan of Berlin City Council - Berlin Senate Department for they are provided with, or they own ting urban places using mobile apps, Quiet Areas 2018-2023 for the Ber- the Environment, Transport and Climate Protection lin Action Plan of Noise Reduction (e.g., a home sensor), for automatic taking pictures with mobile phones, Curious Noses Reporting, planning and model Flemish Environmental Agency sensing. Thirteen projects were cat- writing observations using mobile Figure 5.1. Types of collected data improvement Samen Meten Official air quality monitoring Dutch Institute for Public Health and the Environ- ment (RIVM) Decoro Urbano Planning maintenance and repair Italian local councils of street problems and resource allocation FixMyStreet Report management UK local councils Catch! Identify existing problems and Coventry City Council, Ipswich Borough Council, develop solutions Oxfordshire County Council, Leeds City Council, Newcastle City Council D-Noses Potential integration in official noise Saõ João da Madeira Municipality and the Mu- pollution measurements nicipality of Sofia, and the Intermunicipal Waste Management of Greater Porto Botellon no me deja Implementation of solutions co-de- Barcelona City Council 7 8 4 1 4 1 dormir signed with residents Cycle Hackney Prioritise investments into cycling Hackney Council geolocated sensor textual audio- photos stories from or designing road infrastructures data data descriptions recordings residents Southwark New Homes Prioritise investments into housing Southwark Council 84 Citizen-generated data for policy: A review of EU projects Citizen-generated data for policy: A review of EU projects 85

CGD was“ collected The projects“ were Financial“ to address public reported to have support mechanisms Findings from interviews pollution, or urban infrastructural issues and effects on citizens, update their plan on quiet areas. In included micro-grants Data governance: problems - and expected to be used expected to be collaborating another case, city councils without or tax exemptions CGD was accessible and findable in for a public purpose, such as raising financial resources to hire dedicat- all the five projects. Data was open awareness, helping local authorities used for a public stakeholders, and ed staff that can collect and process to pay for server and reusable in three out of the five make better decisions, or influence purpose, such as political agendas. citizen reports can use valuable and maintenance and projects. Different reasons for not local governments to respond to raising awareness, accurate reports generated by cit- software development, opening the data included the need community-relevant problems. The izens and made available through helping local ” and European to allow citizens to govern their data, choice of a participatory approach example, they felt empowered from an external platform. Councils can or the need to restrict the use of also differed, ranging from lever- authorities make being able to participate and gener- also use the same platform to ex- contributions for the data to avoid purposes not connect- aging a vast community of citizens better decisions, ate data that can be used directly by plore detailed statistics, for example, creation of data ed to the public interest. In all the using sensors, to using a qualitative or influence local a city council. In another case, the to see how many reports they have platforms to store, projects, CGD was not interoperable, approach involving citizens’ per- careful reports submitted by citizens received and how many they have governments to manage and make although it could be but not without ceptions of urban problems, to the on public street issues suggested processed. These figures can also overcoming some challenges. The need for human “noses” to capture a respond to communi- that they believed in the usefulness be used to support political agendas available collected lack of definition of data standards specific phenomenon like odours. To ty-relevant of contributing their data to their because city councils can show how data even after the and metadata in Citizen Science was involve stakeholders, including the representatives in city councils to many reported problems have been problems. end of projects. considered a problem. Problematic public sector and community-based improve local problems. Regarding fixed over time. was also the tension between the organisations, different modalities political agendas, CGD can have an interest of the data platform used in were used. These modalities are re- ” impact as it brings a critical problem Project sustainability: ” a project to keep the data and make flected in the roles played by public the industry, scientific community – e.g., air quality – to the attention Informants described both financial and make available collected data it compatible with specific stand- organisations. These organisations and the policy influencers) at local, of politicians more strongly than and non-financial support mecha- even after the end of projects. Oth- ards, and the citizens’ right to keep had different roles, ranging from national and global levels. official reports produced by institu- nisms to ensure the project’s viabili- er financial mechanisms used in the the data. being clients paying a fee for the tions. Thus, CGD can point politicians ty, continuity, and scalability. Finan- projects included the creation of a service, to co-creating and imple- Project impact: to matters that their constituencies cial support mechanisms included start-up to replicate project results Setup and development of projects: menting solutions to reduce noise The projects were reported to have are interested in. Benefits for the micro-grants or tax exemptions to and use the same methodology to Projects were set up for different pollution, to providing institutional effects on citizens, collaborating collaborating stakeholders were pay for server maintenance and tackle environmental challenges, reasons but shared two aspects: support and offering mentorship to stakeholders, and political agendas. also reported. For example, the city software development, and Euro- and the charge of a service fee to CGD was collected to address public the project. One project also applied Citizens benefited from participating council of a big city partially adopted pean contributions for the creation be collected from city councils which issues – for example, air and noise a quadruple helix model (the public, in the projects in different ways. For the framework used in a project to of data platforms to store, manage signed a contract with the data plat- 86 Citizen-generated data for policy: A review of EU projects Citizen-generated data for policy: A review of EU projects 87

However,“ this new agency of citizens form owners. The fee was neces- of data. Quality of data for scientific Discussion ogies seem to act as “focal devices” cannot just be between local authorities and cit- sary to cover the increasing costs purposes means that data collected New forms of public participation and (Haklay, 2015). This means they hold assumed because izens where citizens can also con- of software upgrading and reporting are analysed, elaborated, and vali- emerging roles for citizens and local the potential to change the way tribute to driving “the how and why of the use of moderation service. Non-financial dated according to rigorous scientific governments citizens look at their living environ- data is produced” (Ruppert, Isin & support mechanisms included ways standards set for the given purpose Digital technologies, such as inter- ments and facilitate data creation technologies. Bigo, 2017). However, this closer col- in which the EU could play a role. of a study. Data quality for non-sci- connected sensors and mobile ap- as a focal practice, a purposeful and Opportunities must laboration should not be expected or For example, an informant wished entific purposes needs to refer to plications, constitute an important meaningful social activity (Haklay, mandatory. It should not be intend- be actively created the EU would help CGD projects gain trustworthiness in terms of sources driver of participation, by allowing 2015). Most of the sampled projects ed as an attempt to downplay the more credibility and fairly acknowl- and used devices, as in the examples citizens to generate data on commu- suggest the existence of focal prac- through closer role of government agencies, but edge and reward citizen contribution presented here. Solutions to provide nity-relevant problems. Thus, CGD tices, as they engage citizens with collaborations as a way to redesign governance to to data generation. data of sufficient quality included projects become techniques allow- their local environments in various between local integrate citizen’s efforts in broader validation mechanisms along with ing citizens to exert their rights by ways, from recording sounds and institutional settings (Lam, 1996), authorities and CGD use by the public sectorPublic the development of a new method- producing data to evidence local odours to taking photos of street with citizens bringing something sector organisations use CGD for ology for data collection, strict data problems and generate better living problems and plotting places where citizens. essential to the table that would be solving local problems, or for re- collection procedures to ensure the environments (Gabrys, 2019). By col- new houses should be built. This lacking otherwise. porting and planning environmental consistency of data, and sensor cali- lecting data that local governments process of recording and mapping ” actions. Building trust and credibility bration methods. can use to derive relevant insights can become a focal practice (Haklay, Rethinking relationships between among public authorities regarding and informing action, citizens can be 2015). citizens and local governments the accuracy of CGD remains a pri- more actively involved in improving Promoting the agency of citizens mary challenge. Public sector organ- and maintaining the quality of their CGD projects as focal practices hold pression of personal beliefs in the through data collection can open up isations may not trust how citizens living environment. Generating data the potential to bring back agency way that the world should evolve the possibility to “achieve” citizen- generate data – for example, how can offer the opportunity for citizens and control to citizens who become and operate” (p. 4). However, this ship, rather than receiving it (Hintz, they use sensors to measure air to raise and amplify their voices “subjects of data” and not only “ob- new agency of citizens cannot just Dencik & Wahl-Jorgensen, 2019), quality – and question the quality of within democratic processes at the jects of data”, about which data be assumed because of the use of through a reconfiguration of the re- the collected data. Quality remains a local level of government (DataShift, is produced (Ruppert, Isin, & Bigo, technologies, as grassroots CGD lationships between citizen and local critical aspect of CGD. An important 2015). 2017). Following Haklay (2015), we projects can just be populated with authorities. For example, citizen in- distinction needs to be made be- see the act of mapping and record- citizens primarily contributing data. volvement in reporting street prob- tween high-quality data for scientific Environmental sensors and infor- ing in itself as “an act of asserting Opportunities must be actively cre- lems makes a difference for local purposes and the fit-for-purpose use mation and communication technol- presence, rights to be heard or ex- ated through closer collaborations city councils in terms of cost, staff 88 Citizen-generated data for policy: A review of EU projects Citizen-generated data for policy: A review of EU projects 89

time savings, and efficient report- tion practices within participatory the most potent actions from public Generating richer and trusted data to policy context where alignment ers may not (Volten et al., 2018). In ing. Citizen reports help initiate in- approaches hold the potential to authorities to boost the integration address current challenges with monitoring requirements and this respect, according to some key terventions on street problems and open up more systematic interac- of CGD with official datasets would Most of the sampled projects involve regulatory standards is paramount informants, the EU could have a role hold local governments accountable tions between local authorities and simply be to open CGD data (and in a citizens using low-cost sensors and (Brenton, von Gavel, Vogel & Lecoq, to play by enforcing that technolo- for their promises. While data and citizens, as well as between local second step, to do that using shared accessible digital technologies to 2018). Some policymakers have gies used by citizens to collect data technologies in the sampled projects authorities, citizens and business- standards or APIs) so that third-par- conduct indicative monitoring and advocated the notion of fitness for are assessed by certified bodies and place responsibilities on bureaucra- es. In this “two-way traffic” model ty applications can be created. Hav- generate data over a wider spa- purpose approach – at times used deemed to meet EU specifications. cies and politicians who have a role of governance (Kooiman, 2003), the ing applications based on a mix be- tial area or more extended periods. interchangeably with fitness for use Last, CGD projects would benefit to play in fixing urban problems, boundaries between local govern- tween CGD and authoritative data However, this data may not be at – where key aspects like data quality, from adhering to open practices they also displace responsibilities ments and citizens become more (instead of only CGD) would increase the same level of precision or accu- scale, cost, interoperability and data and fully documenting such practic- onto citizens – in particular, to take porous. Local governments become their reliability and impact. racy as data produced for regulato- format must be taken into account es (e.g., objectives, data collection more responsibility for their urban more open to gather data and infor- ry compliance (Gabrys, Pritchard & when evaluating the value of CGD for protocol and analysis techniques), environment, rather than merely mation external to their organisa- Barratt, 2016). The methods used a particular policy question (Holdren, to ensure the trust of participating “blaming others.” In these projects, tions from multiple actors, including by citizens are different from those 2015). While the analysis of the in- citizens and other stakeholders. This collecting data becomes a way of citizens. No single actor, either pub- used by official organisations, and terview data show that projects de- aspect also brings to another criti- taking up responsibility as individu- lic or private, has all the information citizens usually have no legal man- veloped and applied validation pro- cal point, which is the openness of al citizens and can go a long way in and knowledge necessary to address dates to standardise reporting with- cesses to ensure data reliability and software (and hardware) used to solving urban problems and achieve and solve complex problems in a in or across countries (Lämmerhirt trustworthiness, it also indicates generate data. Using open-source active citizenship. fast-changing and diverse society. et al., 2019). CGD may not comply that others aspects such as interop- software in CGD would allow full The interaction between local gov- “with established conventions to ob- erability and data format are not yet control (e.g., reproducibility) of the The act of mapping and recording ernments and other stakeholders, tain the quality, interoperability and fully implemented. Two measures to procedures and workflows, which is becomes both an act of asserting including citizens, are often based on verifiability of data and sometimes help address this problem could be crucial to ensure the reliability of re- citizens’ presence or their rights to the recognition of interdependencies abide by ‘good enough’ standards the development of citizen science sults/measures. be heard, and an act of sharing re- (Kooiman, 2003). for operational use, different from data and metadata standards and sponsibility for the governance of those of established official profes- the use of technologies aligned with the public good – term used here CGD projects may enrich already sional statistics” (Lämmerhirt et al., regulatory requirements (Bonn et al., as shorthand signal for the shared produced data, published by gov- 2019, p. 8). 2018). For example, while some en- benefit at a societal level (Morrell, ernments as open data or otherwise vironmental monitoring sensors may 2009). In this respect, data collec- (Lämmerhirt et al., 2019). One of Data quality is a critical issue in a align with regulatory standards, oth- 90 Citizen-generated data for policy: A review of EU projects Citizen-generated data for policy: A review of EU projects 91

CGD has great“ potential to be used as a resource for the public good. Conclusion nologies can change the way hold local governments account- However, this will happen if the public sector, The examined projects have been citizens look at their living enable for their promises. Collecting citizens, and other stakeholders will work driven by community-relevant prob- vironments and facilitate data data becomes a way of taking together to match policy needs, data sources, lems that affect citizens’ quality of creation as a focal practice, a up responsibility as individual life, and the need to provide evidence purposeful and meaningful social citizens and can go a long way technological solutions, and standards. for local authorities to take action. activity. In turn, CGD projects as in solving urban problems and These projects indicate that CGD can focal practices hold the potential achieve active citizenship. How- ” have an impact by enabling different to bring back agency and control ever, “achieving” citizenship is The sample size of 18 projects ex- be useful to guide the public sector relationships with the public sector. to citizens, moving them closer challenging because it requires amined for this chapter is only a por- to use unofficial data together with They can provide the opportunity to to the role of agents of change a “culture shift” such that citizens tion of currently active CGD projects, official data and private repositories. find novel ways of interaction and in the places where they live. and communities become active and new projects surface regularly. This could lead to a digital ecosys- open up channels of communication This potential does not develop participants. Our goal was selecting represent- tem which can provide evidence that between policymakers and citizens. automatically, though: opportu- ★ Third, the issue of data quality in ative entities rather than achieving can be understood and used by de- This chapter shows that types of nities must be actively created policy contexts. Most CGD is col- exhaustiveness or statistical analy- cision-makers, businesses, and citi- CGD data, types of uses of data and for forms of closer collaboration lected using low-cost sensors and sis while making a smaller sample zens alike (cf. United Nations for the types of technologies used in data between local authorities and cit- accessible digital technologies. size appropriate for this review. The Environment, 2018). collection are interrelated. The inter- izens where citizens can also con- This data may not be at the same number of CGD projects in Europe weave of data, data collection prac- tribute to driving “the how and level of precision or accuracy as is growing. This trend is expected tices and forms of use indicate how why data is produced.” This pro- the data produced for regulato- to continue, along with more influ- citizens can generate data in various cess can be challenging because ry compliance. However, it could ence on decision-making at the local social contexts reflecting their lived it implies ways of shifting agency, raise different concerns and pos- government level. CGD has great po- experiences. accountability and responsibility sibilities useful to describe “data tential to be used as a resource for towards citizens. stories” together with citizens the public good. However, this will The chapter points to three main ★ Second, the potential of CGD to and integrate the representation happen if the public sector, citizens, aspects of CGD reconfiguring the re- enable citizens to “achieve” cit- of reality provided by official data. and other stakeholders will work to- lationship between citizens and the izenship, rather than receiving gether to match policy needs, data public sector. it.Citizen reports help initiate in- sources, technological solutions, ★ First, the potential of CGD to act terventions on problems affecting and standards. National/regional as a focal practice. Digital tech- the places where citizens live and legal and policy frameworks could 92 Citizen-generated data for policy: A review of EU projects Citizen-generated data for policy: A review of EU projects 93

References sphere: Global civil society, communi- Gray, J., Gerlitz, C. and Boune- archives.gov/sites/default/files/ Together%20CGD%20Report_0.pdf ment: Discussion paper. Bringing to- Bonn, A., Hecker, S., Bowser, A., cation networks, and global govern- gru, L., ‘Data infrastructure microsites/ostp/holdren_citizen_sci- Linders, D., ‘From e-government to gether data, algorithms and insights Makuch, Z., Vogel, J. and Haklay, ance’, The ANNALS of the American literacy’, Big Data & Society, Ju- ence _memo_092915_0.pdf. we-government: Defining a typol- for sustainable development, 2018, M., Citizen science to foster innova- Academy of Political and Social ly-December 2018, pp. 1-13, doi: Kooiman J., Societal governance, ogy for citizen coproduction in the https://un-spbf.org/ tion in open science, society and pol- Science, Vol. 616, No. 78, 2008, pp. 10.1177/2053951718786316. edited in I. Katenhusen and W. age of social media’, Government Volten, H., Devilee, J., Apituley, icy, edited by S. Hecker, M. Haklay, 78-92. Habermas, J., Between facts and Lamping, Demokratien in Europa, Information Quarterly, Vol. 29, No. A., Carton, L., Grothe, M., Keller, A. Bowser, Z. Makuch, J. Vogel and DataShift, What is citizen-gener- norms: Contributions to a discourse VS Verlag für Sozialwissenschaften, 4, pp. 446–454, 2012. https://www. C., Kresin, F.… and Wesseling, J., A. Bonn, Citizen science: Innovation ated data and what is the DataShift theory of law and democracy. MIT Wiesbaden, Germany, pp. 229–250, sciencedirect.com/science/article/pii/ Enhancing national environmental in open science, society and policy, doing to promote it?, 2015, http:// Press, Cambridge, MA, 1996. 2003. S0740624X12000883 monitoring through local citizen sci- UCL Press, London, 2018, pp. 1-23, civicus.org/images/ER%20cgd_brief. Haklay, M., Beyond quantifica- Lam, W., ‘Institutional design of Meijer, A. and Potjer, S., ‘Cit- ence, edited in S. Hecker, M. Haklay, doi: 10.14324/111.9781787352339 pdf (accessed 14 February 2019.). tion: a role for citizen science and public agencies and coproduction: izen-generated open data: An A. Bowser, Z. Makuch, J. Vogel and Brenton, P., von Gavel, S., Vogel, Gabrys, J., Data citizens: how to community science in a smart city. A study of irrigation associations in explorative analysis of 25 cases’, A. Bonn, Citizen science: Innovation E. and Lecoq, M-E., Technology reinvent rights, edited by D. Bigo, Prepared for the Data and the City Taiwan’, World Development, Vol. 24, Government Information Quarterly, in open science, society and policy, infrastructure for citizen science, E. Isin and E. Ruppert, Data politics: workshop, 31st August and 1st Sept No. 6, pp. 1039–1054, 1996. Vol. 35, pp. 613-621, 2018. UCL Press, London, pp. 337-352, edited by S. Hecker, M. Haklay, worlds, subjects, rights, Routledge 2015, Maynooth University, Ireland, Lämmerhirt, D. Jameson, S. and Ruppert, E., Socio-technical imag- 2018. A. Bowser, Z. Makuch, J. Vogel and Studies in International Political So- 2015, https://discovery.ucl.ac.uk/id/ Prasetyo, E., Making citizen-gener- inaries of different data futures. An A. Bonn, Citizen science: Innovation ciology, London, 2019, pp. 248-266. eprint/1470344/ ated data work, 2016. http://civicus. experiment in citizen data. 3e Van in open science, society and policy, Gabrys, J., Pritchard, H. and Bar- Hintz, A., Dencik, L. and org/thedatashift/wp-content/up- Doornlezing. Erasmus University UCL Press, London, 2018, pp. 63-80, ratt, B., ‘Just good enough data: Wahl-Jorgensen, K., Digital citi- loads/2017/03/Making-Citizen-Gen- Rotterdam, 2018. https://www. doi: 10.14324/111.9781787352339 figuring data citizenships through air zenship in a datafied society. Polity erated-Data-Work_short-report_. eur.nl/sites/corporate/files/2018- British Academy and Royal So- pollution sensing and data stories’, Press, Cambridge, UK, 2019. pdf 06/3e%20van%20doornlezing%20 ciety, Data management and use: Big Data & Society, Vol. 3, No. 2, Holdren, J.P., Addressing societal Lämmerhirt, D., Gray, J., Ven- evelyn%20ruppert.pdf Governance in the 21st century: A 2016, pp. 1-14. and scientific challenges through turini, T. and Meunier, A., Ad- Ruppert, E., Isin, E. and Bigo, joint report by the British Academy Gray, J. and Marres, N. The data citizen science and crowdsourcing. vancing sustainability together? D., ‘Data politics’, Big Data and the Royal Society, 2017, https:// as public experiment, London Ideas, Memorandum to the Heads of Ex- Citizen-generated data and the & Society, pp. 1-7, 2017, doi: royalsociety.org/-/media/policy/pro- Centre for London, 2018. http:// ecutive Departments and Agencies. sustainable development goals, 10.1177/2053951717717749 jects/data-governance/data-man- jonathangray.org/wp-content/up- White House Office of Science and 2019, http://www.data4sdgs.org/ UN Environment Science Policy agement-governance.pdf loads/2018/07/GrayMarres_DataCi- Technology Policy, Washington DC, sites/default/files/services_files/ Business Forum, The case for a Castells, M., ‘The new public tyPublicExperiment.pdf 2015, https:// obamawhitehouse. Advancing%20Sustainability%20 digital ecosystem for the environ- 94 95

Marina Micheli Commercial sector data for the public interest? A qualitative research on data sharing practices in EU cities

Introduction Private sector data collected by commercial entities (such as mobile phone Public bodies,“ especially operators, social media platforms, transport services, accommodation web- including local govern- sites, energy providers, and so on) might offer insights and valuable opportu- ments, might have a key nities to public authorities supporting their efforts to address both short-term and long-term societal challanges. This chapter discusses the findings of a role in the current qualitative research that examined in particular how municipalities in Europe European data landscape, are gaining access to commercial sector as a means to further pursue their as they could help public interest mission. In particular, the study consisted in semi-structured promoting a more interviews with city’s managers and project leaders (e.g. chief data officers) that are working in the field of data, technology and urban innovation. Overall, balanced data economy the study aimed at investigating public bodies role’ in contemporary forms of based in which the value data governance. Public bodies, especially including local governments, might produced with personal have a key role in the current European data landscape, as they could help promoting a more balanced data economy based in which the value produced privately held data is with personal privately held data is redistributed across society (Couldry & redistributed across Powell, 2014; Morozov & Bria, 2018; Bass et al., 2018; Adalovelace Institute, society. 2019).

” How local administrations access to private sector data Private sector data is often described as part of the “urban data landscape” (Kitchin, 2018). Yet, notwithstanding the expectations, the practice of data sharing between businesses and governments is currently sporadic and lack sustainability (HLEG, 2020; Martens & Dutch-Brown, 2020). The European Commission’s High Level Expert Group on Business-to-Government (B2G) 96 Commercial sector data for the public interest? Commercial sector data for the public interest? 97

Data “sharing is not examined as a data sharing - established to pro- competition, market regulation). Fi- for the public interest. This mode for private sector consists in the intro- a contractual agreement, remote ac- technical issue, neither vide recommendations on how to en- nally, public bodies lack a “culture” accessing private data is referred to duction of data-sharing obligations cess by a trusted intermediary, etc. as an economic activity, hance the sharing of privately held on data sharing (e.g. how to create as data donorship (Huyer & Cecconi, as part of subcontracted services (HLEG, 2020). Drawing from a set of but as a socio-technical data for the common good - identi- value with it), have limited resources 2019; HLEG, 2020). Otherwise, pub- (HLEG, 2020): cities might include interviews, in this chapter we con- fied several challenges. The lack of and deficiency of skills and also lim- lic administrations might purchase data sharing clauses in their tender textualize the various operational practice. governance frameworks means that ited trust from both the private sec- data through public procurement contracts “specifying that a service modes for accessing private sector private companies have to face vari- tor and citizens on public bodies ac- (HLEG, 2020): triggered by specif- provider must make any data that data described in this section in the ” ous uncertainties when sharing their countable use of data (HLEG, 2020). ic needs, public bodies request to may be of public value available to experiences and opinions of Euro- Methodology data - in relations to liability regimes The regulations currently available acquire a specified set of data, or the city council in machine-readable pean city’s managers, directors and This study examines how the prac- (who is responsible if inaccurate or for privately held data sharing vary data-driven insights, from a data format” (Bass et al., 2018: 28). All project leaders. tice of accessing privately held data biased data is shared that leads to by EU country and are sector specif- supplier through a contract (Huy- modes could lead to access different is “constructed” throughout relation- discrimination), intellectual prop- ic, while at the EU level key regula- er & Cecconi, 2019: 16). A different types of data: raw, pre-processed ships between actors. Data sharing erty, and competition law. Further- tions are currently being discussed approach is that of data sharing (e.g. cleaned, re-sampled, normal- is not examined as a technical is- more, companies face operational (EC, 2020). pools (Shkabatur, 2019; Micheli et ised), processed (aggregated and sue, neither as an economic activi- and technical challenges for the al, 2020) in which a public author- combined) or insights derived from ty, but as a socio-technical practice. preservation of sensitive commer- Notwithstanding the difficulties, ity establishes a partnership with the data (HLEG, 2020). Private com- The methodological approach is in- cial information and the protection some cities are developing their own other actors to pursue mutual in- panies might be more willing to sell formed by the tradition of research of customers’ personal information strategies to access private sector terests, and commercial companies, (or donate) insights deriving from in media domestication and the (HLEG, 2020). In a study on the eco- data collected by commercial enti- government entities, data plat- internal data analysis (“intelligence social shaping of technology (Sil- nomics of B2G data sharing, Martens ties. The operational models adopt- forms, and/or research institutions sharing”, such as dashboards, apps, verstone & Haddon, 1996; Lievrouw, and Dutch-Brown (2020) identified ed by local administrations to access exchange data in a collaborative reports), instead of actual datasets, 2006). This method is adopted to un- the following economic barriers: privately held data are diverse and way. A related mode is that of data as a way to keep control of infor- derstand how public actors envision monopolistic data markets (compa- denote very different relations be- research partnerships, when public mation and reduce risks (Shkabatur, their ‘power to set the terms’ on how nies can charge high prices for data), tween these actors. Private compa- bodies collaborate with research/ 2019; HLEG, 2020; Micheli et al., privately held data is shared and high transaction costs and perceived nies, for instance, might share data scientific institutions for a project of 2020). Furthermore, data could be what strategies they put forward to ex-post risks for data providers, and with public bodies at no cost on a mutual interest to analyze private- shared in various technical solutions, facilitate access. The study focuses lack of incentives for private firms voluntary basis as corporate social ly held data that the latter have at such as Application Programming in particular on the perspectives of to contribute to the public good if responsibility, such as during an disposal (HLEG, 2020). A different Interfaces (APIs), limited release of specific actors from the public sec- it might affect them negatively (e.g. emergency or to support initiatives relevant mode for accessing data of data under conditions stipulated in tor: cities’ Chief Technology Officers, 98 Commercial sector data for the public interest? Commercial sector data for the public interest? 99

Chief Data Officers, or project leaders working on a city’s innovation/smart The interviews, which lasted 70 min- context. During the interviews it was Results tioned in the interviews are often city agenda. Twelve semi-structured interviews have been conducted with utes on average, have been conduct- taken into account that digital inno- Local administrations’ access to pilot projects, activities at the “ear- representative of as many cities during the course of 2019. A combination ed remotely using voice and video vation is a highly marketed issue for privately held data is a sporadic ac- ly stages”, if not still in preparation of purposive and snowball sampling procedures has been adopted for the across the Internet via a synchro- cities, for instance by explicitly ask- tivity. Data companies often have (“figuring out”). In a few cities the selection of cities to be included. The participants were chosen in a way to nous connection. Conducted by the ing about obstacles and unrealized no interest in selling data, and nei- topic was rather novel, as access have a diversified group by city size, area in Europe, and tradition of innova- author, the interviews investigated projects, and being self-reflexive ther in sharing it with a municipality to private data was not part of cur- tion (Table 6.1). the “concrete realities” of working in during the conversations. (cfr. Section 2). The practices men- rent/planned activities. Companies this area, digging into the actual ex- with mobility data were cited more City Country Size EU zone periences of these professionals, and often, both as potential, past or Amsterdam NL Large North West simultaneously analyzing discourses actual data providers, highlighting Barcelona ES Large South and imaginaries, investigating their how access to privately held data Ghent BE Small West opinions on the topic. The transcrip- might be promising especially for Ljubljana SI Small Central East tions did undergo a qualitative the- this sector. London UK Large North West matic analysis through manual cod- Milan IT Large South ing; the documents have been coded The types of data providers most Rennes FR Small West following the main themes of the interesting according to those who Rjeka CR Small East interviews. The findings discussed in participated in the study are: (1) Tallin EE Mid-size North East this contribution focus in particular utilities companies, and (2) tele- The Hague NL Mid-size North West on the analysis of the codes that re- com operators and online platforms. fer to access to private data, which Utility companies are depicted as Vilnius LT Mid-size North East were labeled: “Operational modes the ideal candidate for access to Zaragoza ES Mid-size South of access”, “Discourses and per- privately held data, but also as a Large: > 1 million inhabitants; Mid-size: Between 300.000 and 1 million; Small: < 300.000. Table 6.1 – List of cities involved in the study1 spectives”, “Relationships between difficult one to deal with, due to lack actors”, “Power to set the terms”, of human resources and interest “Strategies to negotiate power”. The in data sharing. Online platforms analysis aims to delineate common and telecom operators feature less trends in the experiences and dis- prominently. According to a partic- courses, as well as key differences ipant, big platforms are difficult to

1 To protect participants’ identities, the number in the quotes does not correspond to the order of cities in this table. and how these relate to the specific reach because they do not have 100 Commercial sector data for the public interest? Commercial sector data for the public interest? 101

representatives working at the lo- Data donorship es (...) In this case they were gain- question, do we want to have a free Public procurement of data is expensive, scaling up is expensive, cal level. Furthermore, they seem Some respondents recognized the ing some free promotion from these lunch if others are paying for it?” While almost all interviewees dis- the knowhow is a long process (…) I to have less to gain from engaging possibility to access privately held experimental samples (...) And we, (city02) cussed the possibility to acquire data never heard that a product is becom- with a municipality. Several other data or information at no cost, as yes, we didn’t pay them any money.” directly through public procurement, ing cheaper over the years.” (city04) actors, beyond data holder com- companies/data providers occasion- (city09) [During the meeting of a national most were against this solution and panies, are involved for enabling ally make it available for free on a group of cities] “He said, ‘Okay, for the remainders experienced it with “One of the most important things in public bodies to access (and use) voluntary basis. This data donorship Companies use the reputation of us, in cityX, the conditions under great circumspect. This operational the equation is that we are not put- privately held data. Start-ups, pub- mode, however, was often associat- (smart) cities as promotional ma- which we deal with the great com- access mode is contrasted with ideo- ting money in it, so if we bought the lic universities, research institutions ed to a specific discourse. Instead of terial. Thus, being a well-known panies is that we deal for nothing. logical arguments: (1) data produced data that would be easy (…) some- and civic organizations also have an being described as a philanthrop- city seems to be a key enabler for They come and they develop some in public spaces should be accessed times it is their business model, so important role in this context, help- ic move, it was acknowledged as a such form of access to private solutions, and we work together in by public bodies and not be treated they don’t want to give the data for ing with data stewardship and ana- marketing strategy used by private data. This creates a double source partnerships, and it’s free for us’. as a commodity; (2) local administra- free, they want to have money. And lytics. Collaborations with research companies, which favored already of disadvantage for smaller cities And the other one in the room, they tions have to serve the public interest well, we don’t have that kind of mon- institutions or PhD students, is privileged “smart cities”. Companies because they lag behind and are said, ‘Okay, it’s free for you, but it’s and should not invest economically is ey and it’s also some kind of a princi- mentioned as an enabler for work- that freely share data with cities – proposed the same service to a not fair. You have money, more than acquiring data; and (3) it is important ple discussion that the data has been ing in this field, especially by cities and collaborate with them to devel- price. This phenomenon emerged as we have’. And when they get to us, that cities keep sovereignty over data, collected public space. Data collect- with less economic resources. op products or services valuable for an ‘ethical dilemma’ in a couple of they say, ‘Okay, we developed a solu- becoming a buyer to a private plat- ed in public space is from everyone, the municipality without asking any- interviews in which managers question with cityX.’” (city06) form (especially if a big corporation) it’s not just from the company who In the remaining sections we illus- thing in return - do so because this tioned their city’s position (its privi- might undermine their autonomy. happens to put a sensor” (city05) trate the most common operational allows them to market new products lege or lack of thereof) in relation to This operational mode for access modes adopted within these cities and services to other cities in the that of others: tend to be associated to smaller or “I’m very reluctant to pay for data (…) Those engaging in public procure- for accessing commercial sector future. one-time-only projects; one inter- first of all; we need to keep a cer- ment of data also questioned its data and the perspectives of the “They (companies, ndr.) can say to viewer described that as an “inciden- tain amount of independence from effectiveness. Companies, in fact, study’s participants (See Table 2 “A company will approach us and say other cities, “hey cityX did this use tal partnership” to stress the volatile third parties when it comes to infor- often sell data packaged with limit- for a summary). Hopefully these “hey, we’ve got this…”, but it will be on case, our data is very valuable, so nature of the initiative. mation on your city. Because data is ed options for personalization, they findings could shed light on the hin- a pilot form only, because they want our product is also more valuable, not neutral and if we become very send finished products (such as drances, as well as the promising to say that they work with the, you so you can pay more”. This is for us dependent on a tracker, we know dashboards or PDFs) that curtail the avenues, for public bodies’ access to know, the Mayor of cityX, in order to a way to work with these compa- there’s not a lot of competition on possibility of intervention on data. privately held data. market their products in other plac- nies. But again, there is the ethical the market, because the technology Furthermore, they are not transpar- 102 Commercial sector data for the public interest? Commercial sector data for the public interest? 103

ent regarding data quality and rep- you agree to their terms, you can’t Data partnerships ductive relations with private com- resentativeness (“we’ve also found get your hands on the data. So, A different attitude emerged when panies and to address societal chal- companies over-promise”). we’ve had quite a few discussions respondents claimed that they es- lenges more effectively. This relation that have gone round and round and tablished (or wish to establish) data is described as a form of “co-crea- “In the best scenario, we receive data round, maybe for a year, maybe for a partnerships with private compa- tion” in opposition to “buying data” in a PDF, but not in an XML format, year and a half.” (city09). nies. In this operational mode local and being “just a client” of data hold- so it’s very difficult for us to process administrations identify shared in- er companies. A key enabler seems the data in our systems and well, ba- Some respondents suggested that terests with the private companies to be to work with people already in sically it’s not of any use at all if we cities should do collective bargain- holding data, seeking a win-win col- ones network with whom a personal receive a PDF.” (city10). ing to deal with the issues of the laboration. Both parties are involved relationship has already been estab- long negotiations and the high pric- in the project and in the analysis, at lished. Respondents who obtained data es enforced by private companies. time also sharing the objective for through public procurements stress This strategy, which consists in cities which the data is used. Local ad- “We have a history with the people. the experimental setup of such ac- creating a coalition and relating with ministrations eventually give admin- I mean the people working in com- tivities, described as “evaluation companies altogether, would allow istrative data in exchange, creating panyX, I know her for, I’ve knowing phases” to assess the quality of the municipalities to strike better deals. a data sharing pool with the private her for now five years maybe. Had data and the opportunities it affords. company (Shkabatur, 2019; Micheli discussion on different topics, and This mode for access is criticized “We think that cities have a real role et al., 2020), or are simply a partner now I know where she wants to go. mostly because it does not allow in basically collective bargaining in the development of a product (e.g. She knows where we want to go. We municipalities to be involved in de- on this and telling companies what a public service). know where we could go together- it fining the information needed and they’ll pay for it, rather than the oth- is easier. With companyY, it is the the kind of analysis performed, leav- er way round.” (city08) “We try to talk to them, like, ‘what same. We are working with them on ing great decision-making power in are your incentives and how can we data since 2010.” the hands of the company. “With a collective, in a sort of a col- help you?’ We didn’t go to them and lective effort with other stakehold- say, okay, we need your data. We say Another important enabler is the so- “The company has set the rules and ers, to share also the cost (…) we ‘how can we work together?’” (city02) cietal relevance of the projects on we’re not at a stage yet where we’ve could also imagine that the data, which these collaborations are based. set the rules in any of the projects which is for sale, we can go and buy According to the respondents, this According to some respondents, the I’ve been involved in. (...) But until it under a collective.” (city06) approach allows establishing pro- new generations of developers are 104 Commercial sector data for the public interest? Commercial sector data for the public interest? 105

interested in working on socially rel- they have, and they see what we do something we buy, belongs to the Operational mode Discourses Public/private relations Strategies to evant issues. Therefore, establishing with it, how we enhance it, which city of X.” (city02) support access personal relationships with them (for makes their product better. So, it’s Data donorship “Free lunch” The promotional city Reputation instance during hackatons) pave the that iteration.” (city02) “We are thinking about something “Incidental partnership” way for future collaborations. Occa- more systematic, like how to intro- Public procurement “Reluctant to pay for data The city as a client Negotiations sionally respondents highlighted that Data-sharing clauses duce data questions in our contracts, of data to keep independence” Evaluation phases a partnership originated from a com- Another way to access private data in our agreements on different pub- “Data as a product” Collective bargaining mon goal (between the municipality consists in putting data sharing lic policies. It’s quite a different per- and a private company) to impede clauses within tender contracts with spective. It’s not how to reach new Data partnerships “Win-win collaborations” The city as a business Data in exchange the dominance of big tech corpo- suppliers so that data collected as partners on data, but how to intro- “Societal relevance” partner Internal know-how rations in a certain sector, such as by-product has to be accessible to duce data with our historical part- Personal relations Google Maps as mobility app. Never- the city council. Only a minority of ners.” (city06). Societal aims theless, private companies join such respondents already adopted claus- Hackatons data partnerships to develop a busi- es in their tender, and a few others Respondents explanations for adopt- Data-sharing clauses “Data as a public good” The sovereign city A standard legal framework ness model or a commercial product were considering it. This operation- ing data sharing obligations are sim- “Responsible use of data Collective of cities to offer to other cities/clients. There- al mode allows accessing data of a ilar to the motives for not relaying for the public interest” fore this operational mode of access city’s suppliers and, theoretically, of on public procurement. They claim “Independence from com- might also lead to inequalities be- any company that has a contractu- that data collected as a by-product panies” tween cities (such as for data donor- alised relationship with the munici- for delivering public services should Table 6.2. Summary of the operational modes for accessing private data contextualized in the discourses and experiences of twelve local administrations. ship), since municipalities with ad- pality (e.g. public transport, waste be available to public bodies, this vanced knowledge and expertise are collection, etc.). will allow data sovereignty and di- business is to provide a service that access consists in working collec- in which we can come up with the more likely to find companies willing rect the digital transformation at the you are contracted for, so the data tively with other cities and jointly juridical text where we can use that to collaborate with them. The more “We have done that [paid for data] in service of public interest. you are collecting within the service define the same contractual frame- to make a contract with these busi- “experienced” a city is, the more it a couple of situations where it was needs to be available for everyone to work to be used with private com- nesses upfront. So, there is no dis- has to offer to private companies in not specified well in the tender (…) “Thinking about contract services, provide, or for the city to provide, or panies: cussion about a data, but it would be terms of data and support. What we try to do now is prevent tenders, saying that you are pro- to improve the service.” (city12) every city in the country is using the that by making our contracts better viding services as you were the city “we are working together with the same contract, so it’s no use to go “The collaboration so far is more that and have a warrant in our contract council, so it’s not your business to A strategy put forward by a few re- association of cities and we want shopping to another city because it’s it’s a win-win, that they give us what that says all the data being used in collect data about the city. Okay, your spondents to enhance such mode of to come up with a model contract very similar.”(city10) 106 Commercial sector data for the public interest? Commercial sector data for the public interest? 107

Bigger“ and smart cities might Discussion Another underlying issue, emerg- businesses for tenders or partner- have more chances to Huyer, E., Cecconi, G., and Cap- tions Office of the European Union, This short chapter summarizes the ing from the discourses of partici- ships. These tactics could help lev- access commercial gemini Invent, Analytical Report Luxembourg, 2020. findings from a qualitative study with pants, is that of data sovereignty: elling the playing field, lessening the 12. Business-to- Government Data Micheli, M., Ponti, M., Craglia, M., sector data. innovation/data managers of twelve respondents are wary in buying data inequalities described above, and in- Sharing. Report, European Data and Suman A.B., ‘Emerging European cities. Access to commer- through public procurement, both creasing cities’ strength in demand- Portal, 2019. models of data governance in cial sector data has been examined because that would place them in ing access to privately held data ” HLEG, Towards a European strategy the age of datafication’, Big Data as a socio-technical practice that is a dependent position (economical- with a public interest. on business-to-government data & Society, 2020, pp. 1-15, DOI: still “in-the-making”. The chapter de- ly) to commercial companies, and References sharing for the public interest. Final 10.1177/2053951720948087 scribes the four operational modes because there is a lack of transpar- The study provided qualitative in- Adalovelace Institute, Rethinking report prepared by the High-Level Morozov, E., and Bria, F., Rethink- for accessing private data most fre- ency regarding privately held data sights in the experiences of cities’ in- data. Changing the data governance Expert Group on Business-to-Gov- ing the Smart City: Democratizing quently mentioned by the cities in- quality and limited possibilities to novation and data managers in rela- ecosystem. Report, Adalovelace ernment Data Sharing, Publications Urban Technology. Rosa Luxemburg novation/data managers who took control how it is shared. To preserve tion to access to commercial sector Institute, 2019. Office of the European Union, Lux- Stiftung, New York, 2018. part in the study. From the results we control, respondents imagine to en- data as a way to reflect on the role Bass, T., Sutherland, E. and embourg, 2020. Shkabatur, J., ‘The global com- learned that bigger and smart cities gage in collective bargaining with of public bodies in the current Euro- Symons, T., Reclaiming the Smart Kitchin, R. ‘Data-driven urbanism’, mons of data’, Stanford Technology might have more chances to access companies as a means to strike bet- pean data ecosystem. Yet, this chap- City. Personal data, trust and the In Kitchin, R., and Lauriault, T. P., Law Review, Vol. 22, 2019, pp. 1–46. commercial sector data. Their rep- ter deals when acquiring data. The ter only provide some hints about new commons, Report, NESTA, Lon- Data and the city. Routledge, New Silverstone, R. and Haddon, L. utation, their professional network, access modes that allow to keep the issue of ‘getting access’. Further don, 2018. York, 2018, pp. . 44-56. ‘Design the Domestication of their resources and expertise, put control of data, according to the re- avenues of research in this area in- Couldry, N., and Powell, A., ‘Big Lievrouw, L. A., ‘New Media Information and Communication them in a favorable position in order spondents, are data partnerships or clude: an analysis of how cities are Data from the bottom up’, Big Data Design and Development: Diffusion Technologies: Technical Change and to be contacted by private companies data-sharing obligations in tender using or planning to use commercial & Society, Vol. 1, No. 2, 2014, pp. 1-5. of Innovations v Social Shaping of Everyday Life’, In R. Mansell and R. for data donorship or to be welcomed contracts with suppliers. sector data; a systematic review of European Commission (EC), Com- Technology’, In L. A. Lievrouw & S. Silverstone (Eds.), Communication as partners for data sharing pools. the resources available to local ad- munication from the Commission Livingstone (Eds.), Handbook of New by Design: The Politics of Informa- Private companies, then, use such Overall, the strategies described ministrations and how they relate to to the European Parliament, the Media, Sage, Thousand Oaks, 2006, tion and Communication Tech- “use cases” to market their services by the respondents to facilitate ac- access and use of private data; how Council, the European Economic and pp. 246– 265. nologies, Oxford University Press, and products to other cities. Further cess to commercial sector data are citizens, and public trust, are taken Social Committee and the Com- Martens, B., and Duch-Brown, N., Oxford, 1996, pp. 44 -74. research could see to what extent a collective efforts in which cities join into account when gaining access to mittee of the Regions “a European The Economics of Business-to-Gov- divide between cities regarding their forces for the cause: from collective privately held personal data. Strategy for Data” COM(2020) 66 ernment Data Sharing. JRC Tech- chances to access data and use ex- bargaining, to develop a common Final. Publications Office of the Eu- nical Report, JRC Digital Economy ists and with what implications. contractual framework to use with ropean Union, Luxembourg, 2020. Working Paper 2020-04, Publica- 108 109

PART B: on the Governance with Digital Data 110 111

Jiri Hradec, Margherita Di Leo, Massimo Craglia Probabilistic Synthetic Population modelling for EU policy support

Introduction This stream of activity of the project started from the recognition that there In Digitranscope“ is a widening gap between the accuracy and timeliness of the data available we started therefore to the big commercial platforms to personalise services to their customers to consider how we and influence their behaviours via recommendation algorithms, and the slow could use some of the pace of official statistics released at area-level aggregates and often years out of date. Governments of course do collect and have at their disposal in- techniques of the dividual-level data about their citizens and residents but most of the times commercial sector of are not allowed to use it and link it to other individual-level data because consumer profiling and as a society we value our privacy and confidentiality (from government). For this reason, government policy is normally based on the ecological fallacy of targeting and apply them assuming that people living in the same arbitrary spatial units at which sta- to publicly available tistical data is released all share the same characteristics. In Digitranscope administrative data we started therefore to consider how we could use some of the techniques of to develop more the commercial sector of consumer profiling and targeting and apply them to publicly available administrative data to develop more “personalised” policies “personalised” policies targeted to those who need it most. The key idea to develop these personal- targeted to those who ised profiles without having to deal with real personal data was to create a need it most. probabilistic synthetic population from disaggregated official statistics.

Synthetic populations in the form of disaggregated individual data represent the ” main input entities to several multi-agent models and micro models, used in many different contexts in policy design and evaluation, from economic simula- tions (labour market policies, tax benefit, poverty-reduction policies, multi-country microsimulations, etc.) to agricultural and environmental policies, education, health, demographic and social well-being, just to mention a few. In policy decision-making, the levels of complexity that come into play include the heterogeneity of the population and its often under represented different subgroups; 112 Probabilistic Synthetic Population modelling for EU policy support Probabilistic Synthetic Population modelling for EU policy support 113

the behavioural response of individu- Synthetic populations are a powerful Any individual is characterised by a a very limited data sample that rep- missing records (individuals) using als; and difficulty of assessing ex-ante tool because they can be extremely set of variables, or features: I = . Some are independent, ulation of disaggregated data avail- ing sure that the statistical figures subgroups, over time and space. To try vacy of citizens, but still reflect the like age and gender. Others are cor- able at individual level. These data of the population are reflected in the and disentangle such a level of com- complexity of the structure of the related, e.g. education, family compo- are also known as PUMS (Public Use modelled population, within some plexity, it is often necessary to recur to population, and the characteristics sition, place where they live etc. The Microdata Sample), or simply micro- level of accuracy. It is normally nec- models (Aaberge et al. 2014). of the individuals that influence their system is characterised by a state data. This is an anonymized subset essary to operate a selection of the behavioural response. A synthetic that can be updated with ancillary of census data that Statistical Of- features that are important to the Review of Methods population is designed to reflect the data and constraints deriving from fices put at disposals of research- problem under investigation, ne- Micro models based on synthetic heterogeneity of the real population, additional data sources. ers, after making sure of removing glecting or relaxing the constraints population can simulate the effects including minorities and under-rep- every location detail and blurring on the remaining attributes (Nama- of proposed policy implementation resented individuals that would not Several methods have been pro- other information that may allow to zi-Rad et al. 2014). on subgroups, as well as estimating be characterized considering just the posed for generating synthetic pop- reverse engineer the identity of the program costs and caseload (Na- general statistics of the population. ulations (Farooq et al. 2013; Ye et al. individuals. The most popular SR methods are It- tional research Council, 1991). The 2009; Ironmonger et al. 2000; Joro- erative Proportional Fitting (IPF) and spatial dimension is important to In Digitranscope we aimed at re- sz, 2013; Antoni et al. 2017; Arentze Synthetic Reconstruction is based Iterative Proportional Updating (IPU) consider whenever the model is spa- constructing a synthetic population et al. 2007; Lenormand and Def- on microdata and cross-classifi- techniques. Among Combinatorial tially targeted. Datasets with a spa- that would serve as a baseline to be fuant, 2012; Gargiulo et al. 2010; cation tables, also released by the Optimization methods, Hill Climb- tial component are usually available used as an input to different kinds Delhoum et al. 2020; Thiriot and Statistical Offices, that present the ing (HC) is one of the more frequent, at census area (or coarser level of of models. This baseline should be Sevenet 2020; Sajjad et al. 2016; statistical figures of the population proposing a random solution and detail). The input data to the models flexible enough to be successive- Stevens et al. 2015; Namazi-Rad et at various levels of details of 1 – 3 iteratively tries to improve it maxi- may be rich in contextual data relat- ly enriched and updated whenever al. 2014). Most methods tackle the attributes, like for example unem- mizing an objective function, meas- ed to persons or households and lack- more data becomes available. The generation of the synthetic popula- ployment by level of education by uring the performance at each loop. ing on the spatial information, or the population has to carry all possible tion as a fitting problem. Two main gender. From such tables it is possi- Hilltop is reached when the prede- sample size may be too limited to be information until a model is chosen families of techniques are Synthetic ble to derive conditional probability fined errors are lower than a certain representative at a fine spatial scale, and the relevant features can be se- Reconstruction (SR) and Combina- for a certain co-occurrence of attrib- predefined threshold. Each method or vice-versa, data at fine-grain spa- lected accordingly. The baseline is torial Optimization (CO) (Farooq et utes (“marginals”). Starting from the has its strengths and weaknesses, tial resolution may have gaps in con- characterised by all features present al. 2013; Namazi-Rad et al. 2014). microdata (also called “seed”), both and for a full review see Hradec et textual data (Aaberge et al., 2014). in the source. Both families of methods start from SR and CO methods reconstructs the al. 2021. 114 Probabilistic Synthetic Population modelling for EU policy support Probabilistic Synthetic Population modelling for EU policy support 115

Applications Having prepared the base to target of the first wave was reached and et al. (2020). Here we discuss briefly In Digitranscope we used the IPF/IPU techniques to generate the synthetic these policies, we planned a survey of passed, there was a need to identify step 4 which involved the creation of population of Amsterdam, assigning it by household to properties, and iden- people’ attitudes to the energy tran- which economic sectors to open first a synthetic population for the whole tifying vulnerable groups such as elderly people living alone or single parents sition in collaboration with Housing to allow the restart of the economy of France, as data from the French with small children that should be the priority target for policies supporting Europe, a federation of social hous- whilst reducing risk of second waves Statistical Institute (INSEE) was access to healthcare and education (see Fig 7.1), or the energy transition (in- ing organisations in Europe, to iden- of infection. found to be the most readily availa- sulation, sustainable energy sources). tify the policy options more appro- ble for this task. It was not necessary priate to trigger the support towards To answer this question, several steps to use any of the methods reviewed this transition. This survey has been were needed: in Section 2 to create this synthetic delayed due to the Covid pandemic 1 Create a model of the likely num- population as the INSEE makes avail- and we will therefore complete this ber of daily contacts of each per- able anonymised detailed files that usecase in 2021 after the end of the son based on both economic and make it possible to carry out explor- Digitranscope project. social activities, atory analyses of data, to model be- 2 For the economic activities, iden- haviours, or simply to tabulate on a Another application we developed tify the relative number of daily particular subpopulation defined ac- was in the context of the JRC Coro- contacts of each worker by eco- cording to certain criteria: belonging na Virus Task Force supporting the nomic sector, taking also into ac- to a geographical area and / or sta- European Commission and the EU count for each sector the potential tistical unit presenting certain char- member states to assess the relative for telework and the proportion of acteristics. risks of reopening different economic workers commuting daily by public sectors after the lock-down period in transport in “normal” circumstanc- The key datasets used in this usecase Spring 2020. During the initial lock- es. were: down in the Spring of 2020, which 3 Assess the socio-economic impact ★ INDCVI (Individus localisés au can- was more or less stringent in differ- of risk (by gender and income) ton-ou-ville): Table containing the ent EU countries, only essential eser- 4 Assess the spatial distribution characteristics of the individuals, vices were kept running at all times of risk, based on socio-economic such as age, sex, level of educa- (e.g. utilities, food production and characteristics of different regions, tion, household composition, etc. distribution, pharmaceuticals, es- and commuting patterns. Each record contains attributes Note: Properties in green are well serviced while those in yellow and red and not. Source: JRC Figure 7.1: Access to education and health for single-parent families and elderly people sential infrastructures). As the peak Steps 1-3 are described in Craglia of an individual aggregated by 116 Probabilistic Synthetic Population modelling for EU policy support Probabilistic Synthetic Population modelling for EU policy support 117

means of a weight (IPONDI) that to an individual described accord- working in France are taken into gives a measure of the frequency ing to the characteristics of his account. with which every record (profile of trips to go to work (home-work the individual) is found in the pop- trips), his main socio-demographic Additional datasets included the map ulation in a certain area. Records characteristics, as well as those of of the census tracks used by the also contain the 5-digit code of the household to which he belongs. INSEE, and data from the cadastre the census area (IRIS code). Small ★ MOBSCO (Mobilités scolaires): about properties cross linked with census areas, where the privacy is Table containing the information geographic data form the French at stake, are grouped into larger about the mobility related to edu- Geographic Institute (IGN) and Open- areas and the attributes are given cation. Each record in the file corre- StreetMap to create as detailed a at a coarser resolution. sponds to an individual described map with the distribution of dwell- ★ LOGEMT (Logement); Every re- according to the characteristics of ings by type. cord in the table corresponds to his trips to go to attend an edu- an ordinary dwelling described cation institute (home-study trips), Data about the location of educa- according to its location, its char- his main socio-demographic char- tional establishments was extracted acteristics (category, type of con- acteristics, as well as those of the from the Ministry of Education1 while struction, comfort, surface area, household to which he belongs. the location of economic activities number of rooms, etc.), and the ★ MOBZELT (Fichier Activité profes- was obtained by cross referencing socio-demographic characteristics sionnelle des individus (localisa- the data from MOBZELT which covers of the household residing there. tion à la zone d’emploi du lieu de 64 different economic activities with Household information is provided travail). Each record in the table the buildings for the OpenStreetMap only when the accommodation is corresponds to an individual lo- database. The linkages between the

occupied as the main residence. cated at the workplace described datasets are shown in Figure 7.2. Fig 7.2 Conceptual model of the synthetic population for France..Source: JRC Information are aggregated by according to the characteristics the weight. of their trips to work (home-work 1 https://data.education.gouv.fr/explore/dataset/ By linking the datasets above it was zation is known as the Variable Size This problem can be tackled in differ- fr-en-adresse-et-geolocalisation-etablissements- ★ MOBPRO (Mobilités profession- trips) as well as their main so- premier-et-second-degre/table/?disjunctive.nature_ possible in the first instance to cre- Multiple Knapsack Problem. Some ent ways, no solution is perfect but uai&disjunctive.nature_uai_libe&disjunctive.code_ nelles): Table containing the infor- cio-demographic characteristics. departement&disjunctive.code_region&disjunctive. ate families and households and authors (Thiriot, S., and Sevenet, M., there is always a trade-off between code_academie&disjunctive.secteur_prive_code_ mation about professional mobility. All active individuals with a job, type_contrat&disjunctive.secteur_prive_libelle_type_ then to attribute them to individual 2020) propose a probabilistic ap- precision and computational intensi- contrat&disjunctive.code_ministere&disjunctive. Each record in the file corresponds aged 15 or over, registered and libelle_ministere buildings. This combinatorial optimi- proach to pair households to housing. ty. Aiming at a better precision is only 118 Probabilistic Synthetic Population modelling for EU policy support Probabilistic Synthetic Population modelling for EU policy support 119

We“ were The“ clear able to model the advantages of using synthetic population of those of France (because of lack of Conclusions synthetic data is that 63 million people, in data in other countries) it was pos- In this chapter we have introduced we are able to retain sible to arrive at a cumulative esti- the concept of probabilistic synthet- 35 million households mation of the relative risk or reopen- ic population and shown applications the full richness of allocated in 10 million ing the economy by sector, and the at both urban and national/Europe- the geographic and houses in France social and geographical distribution an level. The clear advantages of socio-economic/demo- of potential impacts. This helped to using synthetic data is that we are including their travel graphic distributions at informed the European Commission able to retain the full richness of to work and study and the EU member states about the geographic and socio-econom- the individual level behaviour, and the possible policy options. Ultimately, ic/demographic distributions at the without encroaching proportion of people us- the choice of what to open, where individual level without encroaching into personal data and how is political as it needs to into personal data on the one hand ing public transport on the one hand balance the health vs. the econom- nor loosing detail through the tra- by economic sector. ic and social risks, but this example ditional approaches of aggregation. nor loosing detail shows the opportunities offered by Pseudo-anonymization (i.e. by mask- Fig 7.3: Influx-outflux of French commuters 2016 through the traditional Source: JRC Note: Blue = inflow, Red = outflow the application of AI methods on ing or obfuscating) does not protect ” approaches of available official data to support pol- against de-anonymization as ad- icy. vances in machine learning lead to aggregation. possible when the input data add up larger housing surfaces, which is ob- iour, and the proportion of people ever smarter reidentification attacks. useful information. Sometimes least viously not always the case2. using public transport by economic ” computationally intensive solutions sector. The figure 7.3 below shows We have recreated the synthetic graph properties help design com- offer reasonable results as well. In Notwithstanding these limitations, the model of the commuting pat- population as a directed network munities of common features (e.g. our case, having any additional at- we were able to model the synthetic terns of 26 million French in 2016 graph where synthetic individuals same city in simplest example) and tribute to houses, e.g. year when population of 63 million people, in with blue/red scale showing com- belong to synthetic families and meta-population connects these built, would make people positioning 35 million households allocated in muting balance (blue = positive in- households and live in real houses communities. Other sectors and pol- much more precise. Another source 10 million houses in France including flux and red = outflux). with real workplaces, schools and icies may benefit accordingly. We are of uncertainty is that, in the absence their travel to work and study behav- shopping and leisure places. Such a now working with the Central Bureau of better information, we assumed Assuming the same commuting pat- population is fit for epidemiological of Statistics of the Netherlands to 2 For a full discussion of the method and limitations that larger families would inhabit see Hradec et al. (2021) terns by public transport by sector as studies, where either attributes or replicate and validate this approach 120 Probabilistic Synthetic Population modelling for EU policy support Probabilistic Synthetic Population modelling for EU policy support 121

This framework“ would allow to design “personalised” policies that are target to those who need support most and could be loca- be consistently used to better under- legal umbrella. The existing legal tion-specific. In this way, also the feedback on policy stand the dynamics of an infectious framework settled into a certain lo- disease outbreak, socioeconomic cal optimum. We started discovering outcomes could be localised and personalised and impacts and policy response. This where our simulated population has allow a much richer understanding of what worked framework would allow to design needs while the legal analysis shows well and what can be improved, opening the door “personalised” policies that are tar- the boundary conditions to be able get to those who need support most to satisfy these needs. Ultimately, in the future for policies-that-learn by design. and could be location-specific. In this we may be able to completely rede- way, also the feedback on policy out- sign the policy cycle so that it is built ” comes could be localised and per- bottom-up with people really at the sonalised and allow a much richer centre of government intervention. to synthetic population modelling sition in social housing. A further understanding of what worked well in the Netherlands. We plan then to possibility is to enrich the synthetic and what can be improved, open- work with other statistical agencies population with the highly detailed ing the door in the future for poli- and then EUROSTAT to bring this ap- and comprehensive behavioural cies-that-learn by design. proach EU-wide and have a robust models developed by the large web population base for comparative platforms for marketing purposes, In another line of work, presented in modelling and policy design. if they were made accessible, or to the next chapter, we have used ar- encourage the public to “adopt” their tificial intelligence tools to identify One of the key future opportunities synthetic personal digital twin and commonalities and patterns in the is then to link the synthetic popula- endow it with their behavioural re- European legal and technical doc- tion with behaviour as inferred from sponses to stimuli or problems, for uments, helped bridge the domain either EU-wide data collections like example in a gaming environment. jargons and started extracting and the Eurobarometer3 series or nation- The possibilities are endless, but the verifying facts from texts. This tool al/local survey, like the one we are outcome could be a very rich and has helped us to move towards anal- currently doing with Housing Europe detailed framework for agent-based ysis of legislation on all levels of the on the attitude to the energy tran- modelling, where unified population European administrations, from the used in different domains will lead to local and regional bylaws to nation- 3 https://www.europarl.europa.eu/at-your-service/en/ be-heard/eurobarometer comparability of the results. It can al legal frameworks to the European 122 Probabilistic Synthetic Population modelling for EU policy support Probabilistic Synthetic Population modelling for EU policy support 123

References Activity-Based Demand Modeling global projections of household National Research Council. Im- ancillary data. PloS one, 10(2), 2015, Aaberge, R., Bach, S., Bargain, O., for a Future Urban District. Sus- numbers and size distributions proving Information for Social Policy e0107042. URL: https://journals. et al., A. Handbook of Microsimula- tainability, 12(14), 2020, 5821. for LINK countries and regions. In Decisions--The Uses of Microsimula- plos.org/plosone/article?id=10.1371/ tion Modelling. Emerald publishing, URL: https://www.mdpi.com/2071- Project LINK meeting, Oslo, Norway tion Modeling: Volume I, Review and journal.pone.0107042 2014. 1050/12/14/5821 2000., pp. 3-6. URL: https://pdfs. Recommendations (Vol. 1). National Tanton, R. Spatial microsimulation: Antoni, J. P., Vuidel, G., & Klein, O. Farooq, B., Bierlaire, M., Hurtu- semanticscholar.org/832a/8931ae- Academies Press, 1991. developments and potential future Generating a located synthetic pop- bia, R., & Flötteröd, G.). Simu- 225cf0148e564efaf8d5f29dc- Sajjad, M., Singh, K., Paik, E., & directions. International Journal of ulation of individuals, households, lation based population synthesis. 67d7a.pdf Ahn, C. W. A data-driven approach Microsimulation, 11(1), 2018, 143- and dwellings. Luxembourg Institute Transportation Research Part B: Jorosz, B. Estimating House- for agent-based modeling: simulat- 161. URL: https://www.microsimula- of Socio-Economic Research (LISER) Methodological, 58, 2013, 243- holds by Household Size Using the ing the dynamics of family formation.org/IJM/V11_1/IJM_11_1_4.pdf Working Paper, 2017 URL: https:// 263. URL: https://econpapers.repec. Poisson Distribution. In Population tion. Journal of Artificial Societies Thiriot, S., & Sevenet, M. Pairing papers.ssrn.com/sol3/papers.cf- org/article/eeetransb/v_3a58_3ay_ Association of America. 2013, URL: and Social Simulation, 19(1), 2016, for Generation of Synthetic Pop- m?abstract_id=2972615 3a2013_3ai_3ac_3ap_3a243-263. https://paa2013.princeton.edu/pa- 9. URL: http://jasss.soc.surrey. ulations: the Direct Probabilistic Arentze, T., Timmermans, H., htm pers/130204 ac.uk/19/1/9.html Pairing method. arXiv preprint arX- & Hofman, F. Creating synthetic Gargiulo, F., Ternes, S., Huet, S., Lenormand, M., & Deffuant, G. Soille, P., Burger, A., De Marchi, D., iv:2002.03853, 2010. URL: https:// household populations: problems & Deffuant, G. An iterative ap- Generating a synthetic population of Kempeneers, P., Rodriguez, D., arxiv.org/pdf/2002.03853.pdf and approach. Transportation Re- proach for generating statistically individuals in households: Sam- Syrris, V., & Vasilev, V. A ver- Ye, X., Konduri, K., Pendyala, search Record, 2014(1), 2007, 85- realistic populations of house- ple-free vs sample-based methods. satile data-intensive computing R. M., Sana, B., & Waddell, P. 91. URL: https://journals.sagepub. holds. PloS one, 5(1), 2010, e8828. arXiv preprint arXiv:1208.6403, platform for information retrieval A methodology to match distri- com/doi/10.3141/2014-11 URL: https://journals.plos.org/ 2012. URL: http://jasss.soc.surrey. from big geospatial data. Future butions of both household and Craglia M. (Ed.), de Nigris S., plosone/article?id=10.1371/journal. ac.uk/16/4/12.html Generation Computer Systems, 81, person attributes in the generation Gómez-González E., et al. Artifi- pone.0008828 Namazi-Rad, M. R., Mokhtarian, 2018, 30-40. URL: https://www. of synthetic populations. In 88th cial Intelligence and Digital Trans- Hradec J, Craglia M., & Di Leo M. P., & Perez, P. Generating a dy- sciencedirect.com/science/article/pii/ Annual Meeting of the Transporta- formation: early lessons from the Multipurpose Probabilistic Synthetic namic synthetic population–using an S0167739X1730078X tion Research Board, Washington, COVID-19 crisis. EUR 30306 EN, Population for Policy Applications. age-structured two-sex model for Stevens, F. R., Gaughan, A. E., DC., Jan. 2019. URL: http://citeseerx. Publications Office of the European JRC Technical Report, 2021 (forth- household dynamics., Plos One 9(4), Linard, C., & Tatem, A. J. Disag- ist.psu.edu/viewdoc/download?- Union, Luxembourg, 2020. coming) 2014, e94761. URL: https://journals. gregating census data for pop- doi=10.1.1.537.723&rep=rep1&- Delhoum, Y., Belaroussi, R., Ironmonger, D., Jennings, V., plos.org/plosone/article?id=10.1371/ ulation mapping using random type=pdf Dupin, F., & Zargayouna, M. & Lloyd-Smith, B. Long term journal.pone.0094761 forests with remotely-sensed and 124 125

Jiri Hradec Semantic Text Analysis Tool (SeTA)

Introduction: What is SeTA and why it is important Searching for information is the story of our lives. It is like trying to find your The“ idea house keys on a Monday morning after you’ve spent the weekend with your to develop SeTA, or kids in the mountains. You try desperately to find it in the huge pile of stuff Semantic Text Analysis which, as tired parents, you have yet to clean up after a late Sunday night ar- tool, came from brain- rival. You usually put it next to the front door in its designated box, but where on earth could it be this time? storming in Digitranscope

on how we could use AI Information extraction suffers from exactly the same problem: usually or- and related technologies ganised, sometimes not; high-value facts are usually stored in designated databases, sometimes not. However, ‘usually’ is not enough, especially in sit- to develop new forms uations that require quick action and a comprehensive overview. What we of more responsive need is a powerful assistant that provides us with this overview, and gives us and citizen-centred quick access to the information we need. policies. The idea to develop SeTA, or Semantic Text Analysis tool, came from brain- storming in Digitranscope on how we could use AI and related technologies to ” develop new forms of more responsive and citizen-centred policies. Policies that could “learn” from the feedback provided by the recipients of the policy intervention and that would be dynamic and flexible in achieving the objectives agreed at policy level.

These initial ideas (see Craglia, Hradec and Troussard, 2020) resonated with the very practical needs of colleagues in the JRC Competence Centre for Mod- elling who have responsibility for the policy impact assessment of European policies. In their work it is important to understand the relationships between different related policies and identify approaches, models, and data that might help assess the contribution of each policy intervention to the overall 126 Semantic Text Analysis Tool (SeTA) Semantic Text Analysis Tool (SeTA) 127

corpora of public European Com- The technology mission documents (EURLEX, the The English linguist J. R. Firth (1957) EU Bookshop, etc.) since 1953. We postulated that ‘You shall know a have chosen English-language texts word by the company it keeps’. SeTA due to the sheer volume of available is a vector space – imagine a cube material and because the majority – where the position of words and of important documents exist in an phrases determines their mean- English version. The key advantage ings and similarities. Thus, the most of SeTA – which is currently a pro- similar terms to the word auditor totype – is its ability to grasp the are external auditor and internal meaning of terms, and the changes auditor. As we have trained the neu- in those meanings over time. Using ral network using our policy-related this ability, it builds up a comprehen- document corpus, the most similar sive ontology which makes it possi- term to eca is European Court of Au- ble to carry out semantic searches ditors. And while Google gives hat, in more than 500,000 Commission lid or limit as synonyms for cap, it documents. A dedicated Technical obviously means Common Agricul- Report contains a detailed descrip- tural Policy to us. tion of the application, together with a host of examples of its application Our approach goes beyond mere impact measured. Their needs pro- SeTA is now a web application that in real-life policy support scenarios term similarities. A well-known ap- vided an excellent use-case to apply is accessible on the European Com- (Hradec et al. 2019) plication of vector logic trained on AI methods to structure and extract mission’s network to provide support general texts is queen-woman+man, knowledge from the body of legisla- for its staff. It depends on a set of which gives king. Although this does tion of the EU and develop an impor- neural networks1 that have been not work in our vector space as our tant building block of this concept of trained on the practically complete texts are generally gender-neutral, “policies-that-learn”. Water Framework Directive – water 1 Word2vec and FastText word embedding are explained here: https://towardsdatascience.com/ + waste yields Waste Framework word-embedding-with-word2vec-and-fasttext- a209c1d3e12c Directive. And if we know that Wa- 128 Semantic Text Analysis Tool (SeTA) Semantic Text Analysis Tool (SeTA) 129

Figure 8.1 - On-the-fly generated semantic map surrounding the term ‘audits’. Source: JRC Figure 8.2 Evolution of the term ‘auditors’ over decades, reflecting contextual change in policy documents and technical reports. Source: JRC ter Framework Directive is actually a By training networks by half dec- and all terms used within it (includ- directive and we are interested in in- ades to capture the language specif- ing slang2), plus fast access to rele- dicators that are linked in the vector ic to each new Commission, we also vant documents, are prerequisites space, we get water indicators, water extract how the meaning of a term for allowing experts to concentrate quality indicators, but also spatial changes over time in a given policy on generating added value, instead indicators and ecosystem service in- context (see Figure 8.2). dicators. Sounds like science fiction? 2 According to Wikipedia, ‘slang exists because we must come up with ways to define new experiences No: just the application of a mathe- SeTA’s key users are Commission ex- that have surfaced with time and modernity’. The meanings of words develop in different ways, matical principle. We learn by learn- perts working in the field of policy im- depending on whether they are used by politicians, scientists, engineers or policy makers. However, con- ing to ask, and the learning curve pact assessments, where an ability text and frequency remain the same, so SeTA can tell that waste water treatment plant is very similar can be pretty steep (see Figure 8.1). to learn quickly about a new domain to municipal sewage treatment facility. 130 Semantic Text Analysis Tool (SeTA) Semantic Text Analysis Tool (SeTA) 131

A different approach to tifying all phrases (e.g. audit trail) [phrase], Belgium [location], 2007 summarising text and mentions of date, location or [date] and 3.4% [quantity, percent- There are several methods for ex- institution, etc. The neural network age]. However, sentences are usually tracting meaningful information predicted whether the 2000 in the written by real people and, as par- from a large text without having to sentence is a year, a quantity or part aphrased in the very first sentences read everything. For instance, we of a phrase (e.g. Natura 2000). of this article, are often too complex can combine word embedding with and difficult for an algorithm to com- the TextRank algorithm to extract This approach allowed database prehend: ‘The Federal Statistical Of- the most important words or sen- queries such as [eurovoc: health risk fice released figures for GDP growth tences from a document. In theory, AND sentence:PM2.5 AND sentence:- and the general government deficit we can train a deep neural network exposure AND entities:(PERCENT)] in 2005, at 0.9% and 3.3% of GDP, on a large corpus of text/abstracts that yields 18 sentences in 10 doc- respectively’.3 Where an analyst can to obtain a solid model for gener- uments. We can refine the query get slightly confused for a second, ating summaries from a document. results further by adding children, the machine still fails miserably to However, we have taken a different which returns one sentence from the understand how the numbers con-

Figure 8.3 – Features common to impact assessments and audit. Source: JRC approach that is best suited to the final report of the JRC project SIN- nect to the terms … so far. But we needs of policy analysts: claim ex- PHONIE on school indoor pollution. are working on it. of performing mundane tasks. How- One of the most significant side-ef- ments have in common with audit? traction and fact checking. The whole procedure took less than ever, SeTA’s potential user commu- fects of building SeTA was finding In less than two seconds, SeTA finds 30 seconds. Why search for informa- nity is much broader: archivists, pol- that a digital assistant can support similarities, components and vari- Imagine being tasked with checking a tion when you can simply find it? icy analysts, anyone responsible for domain experts by offering compre- ants directly from the vector space, draft impact assessment about popu- checking a large document collection, hensive coverage of their domains. with no need to search for anything lation exposure to PM2.5. The search Such an approach therefore helps or even the leader of a newly estab- Thus, SeTA provides on-the-fly ontol- in the texts. It also satisfies the fre- “PM2.5 exposure population in health us to construct a knowledge graph lished cross-domain team trying to ogy creation, where the expert’s job quently expressed desire for a more risk” yields 220 results on the Pub- as the key to yet another and more help people understand their inter- is to say exactly where to stop. This holistic viewpoint, as comparing lications Office website. About two important advance: automated fact locutors’ slang would benefit from an approach also helps to explain and terms from different domains will weeks’ reading time, you might think? checking. Natural language-pro- instrument that explains the mean- explore the relationships between unearth relationships with which the cessing algorithms can easily parse ing behind Commission policy terms different terms, as in Figure 3, which reader may have been unfamiliar. We have split our EU corpus into the sentence GDP growth in Belgium and their context. simply asks: What do impact assess- roughly 500 million sentences, iden- in 2007 was 3.3% into GDP growth 3 CELEX:32006D0344 132 Semantic Text Analysis Tool (SeTA) Semantic Text Analysis Tool (SeTA) 133

Our ultimate“ goal is to create synthetic Stepping into the future Knowing what claims have previous- and air pollution are the very Conclusions legislation that References One of the main reasons we are ly been made about a topic would much the same thing. In this chapter we have highlighted understands needs Craglia, M, Hradec, J. and Trous- working on knowledge extraction is help avoid contradictions. Ex-post ★ Third and most important aspect some of the key features and ideas sard, X. “The Big Data ad Artifi- and human behaviour automated fact checking. While the analysis will be much simplified if all is how many use cases the SeTA behind the development of SeTa. For cial Intelligence: Opportunities and first sentence in the previous para- facts and claims upon which policy covers. Originally trained to solve more details see Hradec et al, 2019. and calibrate Challenges to Modernise the Policy graph can easily be structured into formation is based are already avail- less than twenty problems, the intervention to Cycle”. In Soucha V. and Sienkewicz a Eurostat database RESTful service able. New problems where informa- new domains will significantly Extending our scope to national, re- (Eds.) Science and Policy Handbook respond to the stated query to obtain information that tion quality yields a totally different improve what all analysts can ex- gional and local levels of policy and Elsevier. 2020. https://www.science- Belgian GDP growth in 2007 was meaning will result from this process. tract from the text. legal documents will bring new in- needs. direct.com/book/9780128225967/ actually 3.4%, the second sentence sight into imbalances, inconsistencies science-for-policy-handbook. requires the attention of an expert. There are several limitations to our SeTA has been trained and continu- and inequalities hiding in gaps intro- ” approach we know of: ously improved in order to support duced in all layers of administration scribed in the previous chapter is Firth, J. “A Synopsis of Linguistic Automating fact extraction and ver- ★ First, it is based on technology analysts exactly where we think it when top down European legislation to create synthetic legislation that Theory 1930-1955,” in Studies in Lin- ification will help us in the very near established in 2015-16, eons ago will help them the most. Finding all transposition process was meeting understands needs and human be- guistic Analysis, Philological Society, future to build an AI assistant that in AI development. Representa- facts, citations and relevant doc- the bottom up national laws and even haviour and calibrate intervention to Oxford. 1957. will offer policy analysts the facts tion learning has evolved beyond uments will reveal fundamental the practical detailed local bylaws. respond to the stated needs. If then they need while they are actually simple word vectors but still has truths. We may not yet have reached augmented with user feedback and Hradec J., N. Ostlaender, C. Mac- writing their sentences. GDP growth not redefined the purpose – un- a situation of on-the-fly data ex- SeTA’s tools, even in their current interaction, it can really open up the millan, S. Acs, G. Listorti, R. To- in Ireland in 2015 was … And, ping, derstanding the word in context traction from published news and state, either as a web application or possibilities of a new contract be- mas, X. Arnes Novau, Semantic the computer automatically inserts so analyst can learn ten times Member State reports, but we are a web service for information sys- tween the citizen and his/her com- Text Analysis Tool: SeTA, EUR 29708 the eye-watering figure of 25.6%. faster. working on it. We may still have a tem integration, provide a major op- munity. In the words of artificial EN, Publications Office of the Eu- ★ Second, it has been trained only low fact-extraction ratio, but 80% portunity for alleviating the burden intelligence community, huge mul- ropean Union, Luxembourg, 2019, However, more importantly, such on the European corpus and only is better than nothing. PDF is a print for the institution’s analysts/gener- ti-domain policy shifts may help go https://publications.jrc.ec.europa.eu/ information can be extracted to cre- on English texts. Multilingual em- format and a lack of information on alists, who can understand the facts beyond local optimum achievable by repository/bitstream/JRC116152/ ate a database of facts and claims. bedding was not available at the text/sentence/paragraph flow means when they see them instead of read- legislation evolution but by finding kjna29708enn_1.pdf Computational models (e.g. for GDP time of the SeTA design, a task that extracting text from PDFs is a ing reams of self-confirming text. global optimum, of fair resilient so- forecasts) can benefit hugely from still ahead of us. Without proper real headache. To compensate, we ciety. We are not there yet but exper- having comprehensive information multilinguality, where Luftver- have also written de-hyphenators, Our ultimate goal along the simu- imenting and getting feedback is the for back-casting to improve models. schmutzung, znečištění ovzduší spellcheckers and phrase formatters. lation of synthetic population de- way in which we can make progress. 134 135

Corentin Kusters and Henk J. Scholten Digitranscope Experiments: Digital Twins and Smart Cities, case studies of Amsterdam and Duisburg

Introduction At the present time, around 80% of the population in Europe and North Amer- ica lives in cities. Asia and Africa are estimated to host 2/3 of their population in urban areas by 2050. Those figures are expected to increase in the next 30 years with a population growth rate reaching 5% a year or more in certain urban agglomerations around the globe. In that perspective, there are growing concerns on how to accommodate such a growing population and maintain a sustainable environment and a great quality of life in urban areas.

In this context, cities’ officials are facing an increasingly complex environment. Preoccupation is now towards the design and operation of sustainable cities, cities that combine environmental, social and economic balance and it can be hard for decision makers to respond to such requirements in an effective way. We must provide them with the necessary tools that will help them apprehend this complex environment and build efficient processes.

Smart City projects are at the crossroad of those imperatives using digital solutions and advanced analytics to tackle the growing concerns mentioned above. It is believed to provide a greater understanding of a city’s dynamic and help decision-makers in their tasks in various domains. Smart city projects leverage the growing pool of ICTs and data to benefit inclusive, sustainable and liveable cities.

In this study, we investigated the potential of smart city projects and digital twins to address issues on mobility, liveability and energy. Two case studies 136 Digitranscope Experiments: Digital Twins and Smart Cities Digitranscope Experiments: Digital Twins and Smart Cities 137

form the base for this research: Amsterdam case study ● the Amsterdam case study Problem Statement ● the Duisburg case study Amsterdam’s population is growing and is estimated to host nearly 100 Both case studies explore key as- 000 additional citizens in the next pects of digital twining and of the 10 years. Beyond its population, the smart city movement at large with city attracted approximately 19 mil- the inclusion of technologies such lion tourists in 2018, roughly 3 mil- as the Internet of Things (IoT), Big lion more compared to 2017. In this Data, Artificial Intelligence (AI) and context, the city is facing challenges Machine Learning. Beyond techno- as to accommodate such a dense logical implementation, we also aim population, especially in terms of to demonstrate the importance of mobility. Indeed, the transportation open standards for the development network is more and more pressured of flexible and interoperable smart resulting in traffic congestion, pub- cities. lic transport delays, safety issues and more. In addition, in 2018, the The experience of these case stud- Dutch government has initiated the ies has provided Digitranscope with National Climate Agreement which practical insights into the complex aims at reducing greenhouse gas Figure 9.1. Intelligent Dashboard Amsterdam: Mobility view. Source: authors relationships between data, technol- emissions in the Netherlands by ogy, social and economic problems 49% by 2030 compared to 1990 Facing those challenges, the Munic- structure investment, a greater un- in urban environments and stake- levels. It also includes an action plan ipality of Amsterdam has chosen derstanding and management of the holders perspectives. It has allowed to facilitate the energy transition as to invest in the implementation of city is believed to a have a positive us to see the opportunities of the much as possible. This policy consid- digital twins and smart city dash- impact on GHG mitigation, mobility digital transformation but also its ers a regional approach where public board. The objective here is to devel- issues and energy consumption. limitations and complemented in- authorities such as the Amsterdam op a platform that convey relevant sights from the activities of the pro- Metropolitan Region have a big role insights for assets and services ject described in the other chapters. to play. management. Indeed, beyond infra- 138 Digitranscope Experiments: Digital Twins and Smart Cities Digitranscope Experiments: Digital Twins and Smart Cities 139

Technological readiness Implementation the objective of actionable insights dict traffic flow, games engines have own. They need strong private part- area with an improved inclusion of For a bit more than a decade, the For this case study, we collaborat- (that we can act upon). been used to develop virtual reality ners with a good expertise on ICT socio-economic, environmental and city of Amsterdam has well inte- ed with the city of Amsterdam and interfaces and efforts have been and IT infrastructure to fill the gaps. digital requirements. It has resulted grated the smart city vision in its the Johan Cruyff ArenA stadium, and That information is then integrated made to integrate point cloud data In the frame of our project, our coop- in the creation of a coherent and in- development agenda. The munici- used the Arena and its surrounding within a digital twin of the city of Am- in the system. eration has been built on the basis of clusive development plan that surely pality has deployed efforts to create as a living laboratory. With hundreds sterdam. This faithful representation a common desire to develop impact- grants the project its “smartness”. an inclusive, communities-centered of thousands of visitors each year, of Amsterdam, in 3D, has been made Finally, this software solution is sup- ful solutions that will serve citizens. Moreover, in an effort to bring digi- city where citizens, learning institu- this entertainment, shopping and possible by the use of open BIM and ported by cloud-based technologies Beyond simple client-provider agree- tal twins to the public, Geodan has tions and private partners can ac- business area perfectly fit the scope GIS standards that aimed at digitalis- that allows us to store a large quan- ments, we jointly designed solutions. launched in close cooperation with tively participate in the city’s digital of the smart city. After consultation ing building and city environment. Be- tity of data and perform analytics This calls for stronger inter-organ- the Vrije Universiteit and the JRC transformation. Indeed, the 2016 with the different actors, stakehold- yond cosmetic, this 3D model is now tasks in an efficient way. We devel- isations ties with a great emphasis “Ecocraft” (see chapter 10): a comput- Europe’s Capital of Innovation has ers and also citizens, three core enhanced with the information about oped GOST, an IoT platform that can on communication and accessibility. er game with an educational angle, created structures such as the Am- themes have been selected: Smart the city and its mechanisms. be deployed with IoT devices, data For example, we worked following an based on the popular Minecraft. The sterdam Smart City that facilitates Mobility, Smart Energy and Sustain- and applications on the Web. GOST Agile methodology which encourag- virtual world in Ecocraft is equal to the transition toward a data-driven able environment. The digital twin is then displayed in a is an open-source, certified, OGC es regular communication over the the digital representation of Amster- city. Those efforts resulted in well- dashboard, presenting the different standard, that allows the storage tasks and requirements and re-en- dam. Such tools allow the youngest ground infrastructures and protocol Data has been collected from di- KPIs and features in a user-friendly and processing of the real-time data forces cooperation between partners. generation to familiarize themselves for data collection and storage. In verse data providers (both public manner. This 3D interface enables and makes them available to parties. with the digital representation the Netherlands we can cite for in- and private) in and around the Arena. an intuitive navigation and helps de- Citizens through an interface and game that stance the Nationale Databank We- Information on traffic flow, parking cision makers in taking effective and Partners In 2017, the municipality of Amster- they know and love. And perhaps gverkeersgegevens (NDW), a con- occupancy, public transports as well immediate actions. In this project, we partnered with the dam invited strategic stakeholders, more importantly, through play they sortium of 19 authorities that work as solar energy production, wind tur- City of Amsterdam and in the Johan including residents and associations can participate in designing their fu- together to collect, store and dis- bines energy production, building en- Note that aside the formal require- Cruyff ArenA for the development of to contribute to the elaboration of ture neighbourhood and community. tribute road traffic data. The techno- ergy labels, air quality and pollutant ments, room has been made for re- our solution. This type of public and the vision and ambition plan for the This tool has been used to digitalise logical readiness of the city of Am- concentration are collected, cleaned search and innovation and the pos- private partnership is set to be the Amstel III/ArenAPoort area. Thanks the entire Netherlands and present- sterdam greatly contributed to the and analysed. Following the require- sibility to develop AI solutions, new basis of any smart city projects. In- to this participative approach, the ed to citizens in events such as the development of smart city projects, ments and based on the datasets KPI’s, improved 3D environment. For deed, public organisations are in the Johan Cruijff ArenA has seized this WeMakeTheCity festival for greater digital twin and dashboards. available, key performance indica- instance, machine learning algo- best position to lead the smart city opportunity to address the need outreach of the work being made in tors (KPIs) have been designed with rithms have been developed to pre- movement but cannot do it on their for a holistic transformation of the our communities. 140 Digitranscope Experiments: Digital Twins and Smart Cities Digitranscope Experiments: Digital Twins and Smart Cities 141

Duisburg case study IoT platform based on RhineCloud, a needed to be addressed. Due to its Problem Statement flexible cloud infrastructure, and has important harbour, mobility is the Duisburg has one of the biggest in- extended its Wifi and soon 5G ac- prime challenge Duisburg is facing. land harbours in Europe and a direct cessibility across the city. Moreover, The number of lorries transiting connection via train and road to Ein- efforts have been deployed for the through the city is important which dhoven and Rotterdam. Furthermore, creation of a digital model of the city can cause strong disturbance in the city is part of the New Silk Road, via the collection of point cloud data traffic. Therefore, the prime aspects an International Freight railway that taken by Lidar technology. to be considered were toward traf- connects China and Europe. The city fic flows, parking’ availability, public being an important hub for freight Beyond infrastructures and the on- transport and EV charging stations. transport in Europe, it is facing going effort to collect more informa- Along with mobility, the municipal- congestion of the ring roads due to tion, Duisburg has made its datasets ity was also interested in investi- heavy traffic in and out the harbour, available through an Open Data gating the environmental impact of especially during peak hours. Moreo- platform, the Smart City Duisburg such traffic. Consequently, some en- ver, Duisburg is concerned by the ef- open Data that aims to ease access vironmental conditions such as the fect that the heavy load of lorry ve- to relevant information, invite cit- temperature, air quality or noise are hicles have on the environment, both izens and organisations to partici- equally part of the project’s speci-

in terms of GHG emission and noise. pate and fuel innovation. fications. Figure 9.2 Smart City Platform Duisburg. Source: Authors

Technological readiness The technological and organisation- Following the requirements, we have and 3D generated GIS models have Innovation has taken a great place of the study and are believed to en- In 2018, the city has initiated a mas- al readiness of Duisburg grounded investigated the different datasets also been used for the design of our in this project. Indeed, cutting edge hance the understanding on mobility terplan for digitalisation with the a solid basis for the development of available that could help us in the digital twin’s 3D environment. machine learning algorithms have and help in making more efficient aim to become a leading smart city smart city projects and digital twining. design of our key performance indi- been developed in order to predict decisions. This is a great example in Germany and Europe. Duisburg cators. Live data on traffic flow, EV In this project, we have taken advan- traffic flow a day-ahead. Addition- of how open data can fuel innova- won the silver medal, in 2019, for Implementation stations availability, parking spot tage of the RhineCloud to host our ally, new KPIs such as the congestion and how we can design relevant “Best digitalization project in cities/ The first step of this case study availability as well as environmental IoT solution. Disperse data are col- tion index (the % of congestion over data analytics from such projects. regions“ with the Smart City con- was to engage conversation with recording on air quality, noise and lected and centralised into this plat- the city road network in km/km) has cept. Following this masterplan, the the different stakeholders in order temperature have been included in form where they can be transformed been implemented. Those addition- Finally, the whole has been inte- city of Duisburg has launched an to prioritise some key aspects that our digital solution. Point cloud data and analysed by our services. al features fit perfectly the scope grated within a dashboard (Figure 142 Digitranscope Experiments: Digital Twins and Smart Cities Digitranscope Experiments: Digital Twins and Smart Cities 143

Without“ the engagement of 9.2) that enables a user-friendly Citizens Findings el and nature of participation from stakeholders, a city experience and an instinctive under- Before engaging in infrastructure The use cases of Amsterdam and people. One must offer the possibil- can never be Smart, standing of the KPIs. The gathering development and project definition, Duisburg have grounded a number ity to all parties to actively partici- no matter how much of information that was so far dis- the first step has been for the city of good practices that we think are pate in its smart city project devel- tributed gives a global view of the and its partners to gather ideas. essential in the good implementa- opment, inviting citizens, business ICT shapes its data … issue of mobility and greatly helps This has been done through a series tion of smart city projects. Hence, owners, research organisations to The starting point of stakeholders in making effective of workshops with citizens, public our findings will come in the form of collaborate. This can be done for in- [Amsterdam Smart City] decisions. For instance, agents can partners, universities and firms. The recommendations: stance via the deployment of urban track congestion in real-time and workshops have resulted in the crea- platforms and/or schemes that will is not the (technical) even predict upcoming congested tion of 271 ideas from which the 18 Beyond technology, smart is about trigger participation and co-creation. solutions, but the road segments. They can therefore most relevant have been selected as people collaboration, deploy efforts into redirecting some core requirements in the project. In The definition of smartness in the Think ubiquity and seek co-creation, and motorists in alternative paths. More, an effort of continuous innovation, city is inevitably linked to one’s un- infrastructures ecosystem with environmental data at hand, an online platform has equally been derstanding of the city needs. On that Data collection, transmission, stor- partnering of they can spot in an effective manner deployed where everyone can send topic, Amsterdam is cited as a good age and processing are the core stakeholders (both temporally and spatially) the new ideas. Those are continuously example to follow: components of the IoT. However, within the city effect of those measures on GHG evaluated to integrate the scope of those parts are rarely handled by a emission and/or noise. In our per- the smart city project. “Without the engagement of stake- single organization. The IoT covers of Amsterdam. spective, a digital twin is not solely holders, a city can never be Smart, a large range of industries and is a visualisation tool but is seen as an Beyond consulting citizens, Duisburg no matter how much ICT shapes its applied to cases of various natures, ” information “bank”. Its smartness is has partnered with the University of data … The starting point of [Am- scales and capabilities. Trying to therefore bound to the information it Duisburg-Essen to create a Smart sterdam Smart City] is not the (tech- merge all those different actors and conveys to the stakeholders. City program to train current and fu- nical) solutions, but the collabora- technologies as one interoperable ture administrative workforce to the tion, co-creation, and partnering of system is certainly the biggest chal- digitalisation of our environment. In- stakeholders within the city of Am- lenge that the industry is currently deed, training workforce on the new sterdam” facing. Indeed, a crucial characteris- digital technologies is essential for tic of the smart city paradigm is its their penetration within public or- Consequently, the “smartness” of a ubiquity, the creation of a single ho- ganisations. project is directly related to the lev- mogenous system. 144 Digitranscope Experiments: Digital Twins and Smart Cities Digitranscope Experiments: Digital Twins and Smart Cities 145

Smart city projects “certainly benefit from standardized technologies as it prevents monopoly To do so, the hardware implemented and unlocks the market capability, leaving the door Innovative Data Analysis from Finally, research and innovation must present capabilities to easily open to smaller players and entrepreneurs. the cloud should not be undermined and op- operate with other systems. This is The requirements and scale of a portunities should be given to de- also important for scalability as it It is an efficient way to break limitations smart city project call for the adoption velopers to innovate upon the data eases the progressive integration of and fuel innovation. of cloud-based solutions. For instance, whether internally or from external new components and services. This the SensorThings API is an open- sources. For instance, a research is equally true for software solu- ” source, certified, OGC standard, that group could be assigned with tasks tions and data. The adoption of open allows the storage and processing of such as developing AI solutions, new standards enables flexibility and the real-time IoT data and makes it KPI’s, improved 3D environment and keeps solutions vendor-independent. available to parties over the web. more. Adopting common standards and protocols at every infrastructures’ Data integration is an important Care for visualization levels that will ensure a seamless step to unify heterogeneous sourc- One should not neglect the value of ecosystem. es into a single, homogeneous sys- dashboard design. It is essential to tem. A successful smart city project present the end-user a user-friendly In addition, consortiums and allianc- should be able to scale up and pro- interface and to invest in front-end es have emerged intending to unify gressively integrate new features development. The dashboard should the IoT industry landscape as well as and data sources. KPIs should be be interactive and straightforward software domains such as GIS, BIM, designed based on the existing IT in- with an easily understandable dis- big data, machine learning etc. frastructure, available documenta- play of KPI’s. Besides, a dashboard tion and gathering all relevant data. can benefit from 3D digital twins of Moreover, smart city projects certain- A strong emphasis is on the notion the city for enhanced user experi- ly benefit from standardized technol- of actionable insights which serves ence. ogies as it prevents monopoly and as a guideline when it comes to de- unlocks the market capability, leav- veloping valuable data analysis. In Be transparent and ensure privacy ing the door open to smaller players such a perspective, data format and Organisations that process data and entrepreneurs. It is an efficient metadata must follow standardized must prove they are accountable. To way to break limitations and fuel in- frameworks to deploy an interopera- do so, full transparency is a key el- novation. ble digital environment. ement. When seeking personal data, 146 Digitranscope Experiments: Digital Twins and Smart Cities Digitranscope Experiments: Digital Twins and Smart Cities 147

It is “central Initiatives“ to favour citizen to engage citizens sovereignty over data, The different parties must jointly in digital life making data openness, transpar- participate in setting clear objectives, should flourish. It is schedules, communication protocols, ency, privacy protection arrangements for data management a crucial element to and to mitigate data and sharing within partners, knowl- build democratic monetization and edge transfer, risk and revenue shar- smart cities. ing and policies compliance. technological lock-in. Involve citizens The practice-oriented” recommenda- Lastly, a great number of profession- tions above provide useful elements ” als in the field, academics as well as of reflection together with the more business leaders have stressed the theory-informed findings identi- importance to integrate citizens fied in Chapters 3 and 4. They add wthin the loop. The city of Amster- to the variety and richness of the consent must be freely given, specif- Finally, when developing a smart city Build strong partnerships dam, for instance, has promoted Digitranscope project and will be ic, informed and unambiguous which project, one must keep in sight the PPPs must follow the smart city vi- inclusive development, inviting citi- discussed further in the concluding means that it is the organization‘s prime reasons for such development sion with long term, coherent strate- zens to have an active role in data chapter of this book. accountability to clearly specify the which are to lessen our environmen- gies and collaboration plans. A first collection, strategies definition and use made of personal data and all tal impact, reach social harmony, step in the creation of such a collab- advisory inputs. Same goes for the parties involved. help decision-makers, and help cit- orative perspective would be simply city of Barcelona that has put citi- izens in living a good life. To do so, to re-considered the status of each zens at the core of the decision-mak- Measures should be taken to pre- it is central to favour citizen sov- party. We must step away from the ing process and by giving citizens serve the privacy of personal data. ereignty over data, data openness, traditional client-supplier schema sovereignty over the data produced. Pseudonymization and encryption, transparency, privacy protection and toward a more “organic” approach Overall, initiatives to engage citizens regularly testing, assessing and to mitigate data monetization and where every party is considered as in digital life making should flourish. evaluating IT systems as well as or- technological lock-in. a partner that pro-actively serves It is a crucial element to build demo- ganisational systems, and following common objectives. cratic smart cities. strict guidelines from the GDPR is a must. 148 149

Jaap Boter, Magdalena Grus, Sanne Hettinga, Henk Scholten Involving children in the renewable energy transition: exploring the potential of the digital transformation

Introduction Digitally“ In the European Union, the built environment is responsible for 40% of the transformed energy consumption (IEA, 2017). To reach internationally agreed climate goals, we not only need energy saving measures for individual buildings, but European society offers also must address the issues at a city or neighbourhood scale. At this level, new opportunities to including all stakeholders in the planning process, creates more support for fuel participatory implementation of renewable energy technologies and minimizes NIMBY effects. There are several strategies to engage these different stakeholders in process. spatial planning, including Geodesign (Steinitz, 2012) and Public Participation GIS (PPGIS) (Brown & Kyttä, 2014; Reed et al., 2018; Sieber, 2006), that allow ” large groups of stakeholders to be involved.

A digitally transformed European society offers new opportunities to fuel many of the steps of a participatory process. Not only are vast amounts of data more readily available, but the digital transformation also offers new possibilities for integrating and representing big data. Rather than a large, two-dimensional map with abstract symbology, now a virtual representation of the real world can be created, based on accurate geospatial data: a digital twin of reality. Herein, one can visualize and model all kinds of processes as well as relations between these different processes, as well as the impact of future scenarios, facilitating multi stakeholder collaboration and citizen participation.

Several authors mention the inclusion of children as one of stakeholders in cases (e.g., Drazkiewicz et al., 2015; Eiter & Lange Vik, 2015). Children are advocated as ‘adults of tomorrow’, who need to learn to take responsibility for 150 Involving children in the renewable energy transition Involving children in the renewable energy transition 151

Recent“ activism on a global scale, their living environment; or even as from children. As a non-negligible such as climate protests the context of adult decision making. one in Warsaw and one in Amster- ‘full citizens of today’, with the right side effect, children’s’ involvement initiated by children like Stakeholder engagement, one of the dam, have been developed by local to be heard (Levine, 2015). With a in participatory planning may indi- primary drivers in the process, for parties utilizing the environment of different perspective on their local rectly also engage others, such as Greta Thunberg, suggest instance, is likely an important dif- the popular computer game Mine- environment, their involvement may parents, family, and the wider public. a great need to build ferentiator; with children requiring craftTM. In Minecraft, users are sup- also lead to improved decision mak- trust from children. (quantitative) information to be pre- plied with a large array of different ing and more innovative solutions. Involving children in a public par- sented in a more accessible and en- types of building blocks, to create Finally, recent activism on a global ticipatory process, however, is not tertaining format to capture engage- houses, gardens, waterworks, and scale, such as climate protests initi- self-evident, as many of the prin- ” ment. Like in regular participatory other planning elements. Convert- ated by children like Greta Thunberg, ciples in the participatory planning planning, representativeness is also ing real-world land use data and 3D suggest a great need to build trust literature have been developed in an issue with children, if not more so; point cloud data, an empty Minecraft with a voluntary approach potential- world can be populated with a rep- ly leading to self-selection. Minors resentation of the real (built) world, can more easily be reached through referred to as ‘Geocraft’ by Scholten certain channels such as schools, et al. (2017). Children can then use but this does require specific consid- this familiar environment to generation of time and group dynamics erate data driven future scenarios (Buchy & Hoverman, 2000) of the by simulation and visualization of context, for example, in the form of future developments and their im- a lesson plan. plications. As the two experiments focused on the energy transition, Given how digital twins resemble the environment was renamed as the idea of contemporary computer ‘Ecocraft’ by the organisers. The two game environments, the question is experiments have somewhat differ- whether a digital twin in computer ent aims, use different strategies in game format can be a particularly execution, and rely on different tool- accessible way for children to par- ing; allowing to explore the effects ticipate. To test this idea, two digital of context, aim and supporting ac- twins of European neighbourhoods, tivities. 152 Involving children in the renewable energy transition Involving children in the renewable energy transition 153

The aim of “the campaign “Ciepło dla Pragi” (“Heat for Praga”) is The Warsaw Experiment to inform and engage all residents of the Warsaw The organization had setup several Teachers, volunteers, and local ex- In Warsaw, smog, as side effect of Praga-Północ district who want to actively work on support activities to achieve this, perts worked with 80 school children household heating, is a pressing issue. such as educational materials, a mi- in total, divided up into 18 teams their neighbourhood. The project wants to stimulate In the old Praga-Północ district of War- ni-audit to be carried out on the own from five elementary schools. There saw, on the eastern side of the Vistu- creative thinking about a “dream district”, with household, workshops, and online were several items of the project in la river, for instance, a relatively high sustainable energy, green areas, and friendly meetings with experts to get help. the local newspaper “Mieszkaniec” number of houses use coal as source as well as other Warsaw newspa- environment. While trying to reach all residents, of heat and especially in the winter it Using the Ecocraft environment, pers; estimated total newspapers has a large impact on air quality in the organisers see children as vital stakeholders. they then designed and implement- circulation is about 40,000. The the city. Awareness about renewable The organizers particularly want to create ed their ideas for a residents-friend- campaign also had a dedicated web- sources of energy as cleaner alterna- a sense of engagement. ly and energy effective area for a site with information about the com- tives is still limited, certainly among chosen part of the district. Addi- petition as well as on energy efficien- the middle-aged group of citizens. tionally, teams had to prepare a cy in general (www.cieplodlapragi. The aim of the campaign “Ciepło dla ” final presentation of the project in pl), which has been visited by 992 Pragi” (“Heat for Praga”) is to inform The Warsaw experiment had a strong on current energy use was not avail- the form of short movie, with the unique users to date. The final event and engage all residents of the War- “campaign” stance. Technical tooling able in the game. authors explaining the proposed saw more than 200 visitors. saw Praga-Północ district who want and infrastructure were kept at a ba- idea and changes in the district. to actively work on their neighbour- sic level, with most of the effort put The experiment was run as an out- A jury reviewed the projects and In line with the ‘campaign’ stance hood and who are not indifferent to into a large competition, outside the of-school competition that ran over chose ten finalists who presented taken in the Warsaw experiment, the issues related to environmental pro- classroom, with prize money avail- an eight-month period, starting in their projects during a “Ciepło dla key insights generated from the ex- tection and energy saving. The project able. Technically, participants had August 2018 and ending mid March Pragi” Open Day (March 9). The win- periment all centre around ‘engage- wants to stimulate creative thinking access to a basic three dimension- 2019. Parents and/or teachers were ning teams received a money prize ment’: about a “dream district”, with sustain- al Minecraft model of the buildings, approached to submit participation to implement some of their propos- ★ Choice of problem area. The or- able energy, green areas, and friendly streets, and green objects in Praga forms for their children or pupils. als. During the Open Day event, also ganizers observe that part of environment. While trying to reach all district. Information was confined to Then, in November, participants were older generations of inhabitants the success depends on the fa- residents, the organisers see children parameters such as height and size prepared for the challenge, focusing were invited to discuss issues con- miliarity and attachment partic- as vital stakeholders. The organizers of the buildings, basis infrastructure on both background knowledge on cerning electric efficiency with the ipants and other stakeholders particularly want to create a sense of and available greenery. Unlike the energy and environmental themes, experts and to get advice on energy already have with the project engagement. Amsterdam experiment, information as well as the Ecocraft environment. efficiency at home. area, whether familiarity with 154 Involving children in the renewable energy transition Involving children in the renewable energy transition 155

its physical appearance, or emo- also helps if children are work- The Amsterdam Experiment In contrast to the campaign stance oped on the foundation of the Ge- tional attachment. Attachment ing on real problems of their Like many large European capitals, of the Warsaw experiment and open ocraft environment of Scholten et to the area helps to increase the district and see the follow-up of Amsterdam has a high diversity in in- enrolment, here a specific group al. (2017) and deals with the energy engagement. It thus may also their work, such as implemen- come, ethnicity, educational level, and of school children was chosen in a situation of a specific (real) location. help to focus on a relatively small tation of the proposed ideas/ quality of buildings. This is particular- class setting; with a strong focus on As there is a one-to-one relation- area, that participants know well. solutions. This requires upfront ly true for the district of Amsterdam education on how energy efficiency ship between this virtual world and ★ Involving other stakeholders early commitment of authorities for South-East; a rather complex district and renewable energy work, and the the real world, clicking on a virtual on. While children are an eventu- change. Here, UNEP/GRID-War- with a relatively large proportion of use of an extended version of Eco- house allows for identification of the al link to reach older generations, saw, together with Veolia as well flats, undergoing structural redesign. craft that allows for calculation of real house and model any choices such as parents and grandpar- as experts from the Praga-Północ The City of Amsterdam and local real metrics for each real building. on the real situation; such as the ents, these other generations district office and the authors of social housing corporations (owning While the Warsaw experiment thus angle of the (real) roof top and the could already be involved in some the winning projects, will choose, about half of the housing in the city) had a limited technical design and subsequent (real) efficiency of a so- way early on in the project, be- finance and implement elements have made serious commitments to was elaborate in the procedure, here lar panel. This was incorporated by fore a final presentation of the of the proposed by children ideas. more energy efficient housing, and the procedure is more straight-for- linking to another spatial data infra- results. In a next project in Poland, One of the ideas presented that to increase clean and renewable ward (part of a lesson plan), with the structure described by Hettinga et al. for instance, the organization is will be implemented is the facade production of energy consumed by technical design and governance as (2018). The modelling capabilities looking for the possibility to cre- of the building growth by plants; households. Any change, particularly more important. Key question here described in that paper are used, as ate family groups (with parents, to be realized on the one of the for high-rise buildings will have to be is whether these children are not well as data on for instance energy grandparents etc), rather than schools of the participants. done as larger project per block. Also, just engaged, but also become a val- labels, solar potential and energy just the children as participants. Dutch legislation requires a minimum uable source for co-decision making, consumption. To make the material If anything, parents’ involvement On a more practical note, the or- of 70% of the tenants approving any thus allowing for participatory deci- more suitable to work with children, from the beginning of the project ganizers observe that using a digital change that increases monthly rent. sion-making and generation of sup- the insulation step was simplified to seems highly beneficial. twin ideally is supported by good, So here, a successful strategy to im- port despite the technical nature. include individual insulation meas- ★ Real problems, real agency. It 24/7 IT support, with dedicated peo- prove housing in energy efficiency ures – floor, roof, wall, and window. ple and control and enough work- and renewable production of energy In Amsterdam, the participants were Additionally, a dashboard keeping shops. Also, while teachers don’t must rely on soliciting grassroots provided with instant quantitative track of the total investment cost, have to be eco-experts, it would be support as well as careful selection feedback on their ideas through a electricity produced, gas consump- good to have very detailed online tu- of building blocks, based on metrics plugin. The plugin, an addition to tion avoided, and CO2 emissions torials available. like costs and gains. the virtual environment, was devel- avoided are displayed, as well as a 156 Involving children in the renewable energy transition Involving children in the renewable energy transition 157

Key question“ here was whether the small dashboard that appears when- gave time in class to work through. children are a valuable Investment Reduction in gas Electricity Reduction in Payback ever a player alters a building, show- They were also available to answer source for co-decision cost consumption production CO2 emission period ing the costs and benefits of that questions and to provide help when (€) (m3) (kWh) (kg CO2-eq) (years) individual change. needed. A total of 31 teams of three making. Thus, allowing Solar panels 6,512,100 - 6,423,219 3,211,609 5.3 children each worked on designing for participatory decision All insulation 81,908,719 8,878,861 - 17,580,145 15.3 The children involved in this project plans in Ecocraft. The children were making and generation All technologies 88,420,819 8,878,861 6,423,219 20,791,754 12.6 were in the second year of the Dutch first given time to come up with a of support, despite the secondary school system and about strategy to make their plan. After- Table10.2: Impact analysis of the overlaid plan of the children 14 years old. The project was embed- wards, they were provided with 4 technical nature. ded in geography classes, with chil- lessons to construct it in the game, and generation of support, despite Table 10.2 shows several indicators that the children have not made ir- dren of a highly diverse educational supervised by the teacher. ” the technical nature. The conclu- of the plan for solar panels, for in- rational or random choices, but that level, as well as from the diverse sion of this project is, therefore, sulation and for a summary of all the payback period of insulation is backgrounds that the district South- Key question here was whether the not an open day, but an analysis of technologies. It shows that the in- significantly higher than the pay- East is known for. The teachers children are a valuable source for the results. Table 10.1 shows the vestment cost of solar panels is 6.5 back period of solar panels. The fact provided a booklet with information co-decision making. Thus, allowing descriptive statistics of the differ- million euros and replaces 3.2 mil- that the children have also selected and assignments to the children and for participatory decision making ent results per technology choice lion kWh of fossil electricity with insulation as a technology despite for the groups. The differentiation electricity from solar panels. The its worse economic performance, presented in this Table implies that locations for solar panels the chil- indicates the added value the chil- Indicator Mean St. Dev. Min. Max. the children had their own individ- dren have selected have an average dren see in this technology, for in- Number of solar panels 1,400 1,570 38 7,612 ual visions and ideas for the area payback period of 5.3 years. The instance the added benefit of comfort Buildings with solar panels 62 63 6 270 and use different approaches. There vestment in insulation is significant- or noise reduction, as was explained Average number of solar panels per building 28 30 2 143 are, for instance, groups that have ly higher, and mainly has a relatively in the lessons the children received Buildings insulated 71 87 0 344 only placed solar panels and have high payback period compared to the before they obtained access to the Buildings addressed 132 135 8 521 not applied any insulation. Further- payback period of solar panels: 15.3 environment. more, there is a group that made years. However, when analysing the changes to a total of 521 buildings payback period of insulation for the % buildings addressed with solar panels 47 27 19 100 within the allotted time, while an- buildings in Amsterdam South-East, % buildings addressed with insulation 53 27 0 81 other group has only selected eight it has indeed an average payback

Table 10.1: Overview of the spread in different indicators between the groups participating in the public participatory planning session. buildings. period of 15 years. This indicates 158 Involving children in the renewable energy transition Involving children in the renewable energy transition 159

The impact“ analysis shows that the children

can make reasonable choices for the area around their school, which they are familiar with.

” Conditions & The impact analysis shows that the also those that never wanted to be a Context Game Digital Twin Requirements children can make reasonable choic- part of it or have no affinity with the es for the area around their school, topic of the energy transition. It was which they are familiar with. The found in this study that the Ecocraft payback periods of the buildings and environment was usable by all chil- technologies the children selected dren included in the pilot. First, the are in line with the average payback lesson plan provided enough knowl- periods of the district. Additionally, edge and skills to teach them about they have chosen buildings specifi- climate change and the renewable Figure 10.1. Context, conditions, game and digital twin as hierarchical set of elements. Source: Authors cally for certain technologies, show- energy transition to participate in ing that this was a well-considered the planning process. Second, the Using gaming in digital twin en how relatively new gaming and ements of context, conditions, game, choice. decision support system in the fa- setting: opportunities and digital twinning are as objects of and the digital twin. While definitions miliar gaming environment with the limitations academic interest, the exploratory of games are plenty and evolving, With the Amsterdam approach a gaming element managed to en- In line with the forward-looking na- nature is well-justified, but then as core elements often include, for in- wide range of children of differ- gage all children to share at least ture of the Digitranscope project, the qualitative research may also bene- stance, tools or tokens, rules and aims, ent background, learning style and some of their ideas and plans. The two experiments were not aimed at fit from post-hoc reflection on how players, or a story. The digital twin is, learning level was reached, enabling differentiation between the plans in empirically proving literature-based this can be understood and what les- like a chess board, only the platform children of all backgrounds and lev- terms of technologies applied, build- hypotheses from well-established sons can be drawn. or board on which the game takes els to participate. Consequently, this ings selected, etc. indicate that they streams of research; but rather at- place, with tokens. The rules and aims, case included children that have designed the plans by their own in- tempting to explore the possibilities As visualised in Figure 10.1, the set- the type of players, and the story – never been included in the public sights. of new technologies and fields, and up of both experiments may be seen the other elements of a game – de- participation planning process, but the new questions they raise. Giv- as an interplay between the four el- termine how the digital twin is used. 160 Involving children in the renewable energy transition Involving children in the renewable energy transition 161

Effective play of the game – certainly Mutually dependent on each other; project may be complementary to Participation level Possible game elements Possible features digital twin with concrete aims like creating en- they influence each other in both the other game elements: Awareness Rewards and punishment Visualisation of the issue gagement or soliciting consultation directions and/or can compensate Engagement Competition Realism/familiarity – is dependent on a range (practical) for each other; as symbolised by the Compared to a traditional public par- Consultation Field trip Interaction conditions and requirements; ranging dashed line. E.g., an advanced digital ticipation (GIS based) setup, here, the Partnership, Power & Control Community Reporting, dashboard from the availability of data to build twin with interactive, quantitative map has been replaced by a high-tech, Manipulation, etc Rules Impression management the local digital twin; the availabili- data may allow for a game design 3D representation with much more ty of computers to access the digi- with point scoring and rewards. But possibilities for input (building blocks) Table 10.3. Possible contribution of game elements vs digital twin at different levels of public participation tal twin as board; the availability of a lack of such features in the digital and (in the case of Amsterdam) direct time in or outside school to play; the twin may be compensated by other feedback. The process, traditional- right or wrong; judges and prizes will the public participation process with There are many smaller and big- capabilities of the umpire or coach game elements, such as rounds with ly a group discussing around a map, fuel participant behaviour; consulta- children, but as stated above, both ger differences in how the two ex- (staff); or the knowledge and skills expert feedback. has been replaced by a game envi- tion may happen as field trips and are interrelated and mutually de- periments have piloted the use of of the players. Finally, the project is ronment, where individuals or teams interaction in real life; and ongoing pendent. It is a combination of dig- Ecocraft in children’s participation dependent on a local context of the Since Arnstein (1969), different lev- mark proposed changes; or even alter partnership may be established in ital twin and game elements that in energy transition. The two key theme, the political and urgency of els of public participation have com- and (re)design the environment. In the community forms, such as meetings, needs to be designed for a particular distinguishing factors, however, are the theme, or whether there is a cul- monly been distinguished, such as process, the digital twin may contrib- newsletters, voting, and town hall situation and level of participation. the underlying difference in aims ture of public participation; all driving ‘awareness’, ‘engagement’, ‘consul- ute in visualising the issue; engage meetings. The two cases are particularly exem- and the balance between game ele- the motivation of players, design of tation’, and ‘partnership (plus pow- more in its realism; allow for inter- plary in how different situations and ments versus digital twin, as visual- the game, and quality of conditions. er, control)’; with the latter implying action like a multiplayer online game; In combination, a digital twin and levels of participation may drive the ised in Table 10.4. Note that these elements are: more participation than the first. On and serve as a presentation platform game elements can greatly support design of the two parts. the negative side, low levels of public of a design or ongoing dashboard in a Hierarchically embedded, such that participation have been labelled as monitoring situation. Participation level Possible game elements Possible features digital twin the digital twin is a part of the game; either ‘manipulation’, ‘decoration’, or Awareness Rewards and punishment Visualisation of the issue the game is dependent on several ‘tokenism’. The intended level of pub- But as the experiments show, a vi- Engagement Competition Realism/familiarity conditions and requirements to be lic participation will drive the design tal part of the process also happens Consultation Field trip Interaction successful; and these all operate in of the project and hence the role of outside the digital twin. Setting re- Partnership, Power & Control Community Reporting, dashboard the context of a particular theme, in the digital twin and use of game ele- wards or punishments (‘rules of the Manipulation, etc Rules Impression management a particular city, in a particular coun- ments surrounding it. The role of the game’) will steer particular partici-

try and culture. digital twin in a public participation pant behaviour and give feedback on Table 10.4. Warsaw (green) vs Amsterdam (blue) 162 Involving children in the renewable energy transition Involving children in the renewable energy transition 163

The Warsaw“ case Here“ strongly focuses on the digital twin achieving engagement, lenging volunteers – outside school was a key factor in the twin was a key factor in the concrete uals and communities; expectations challenging volunteers – hours – to imagine a ‘dream neigh- concrete assignment to assignment to try out various forms on improved decision-making when bourhood’, that is not only ‘green’ in of insulation and solar panel in the more are heard, local knowledge is outside school hours – try out various forms of the use of sustainable energy, but virtual environment with quantita- leveraged, or more innovative solu- to imagine a ‘dream also in its living conditions (green- insulation and solar tive feedback on the quality of the tions are developed; the creation neighbourhood’, that is ery). There is a strong competition panel in the virtual choices; with a report as deliverable of trust in policies and other stake- element, with an extensive judging instead of a video impression. The holders as a result of the coopera- not only ‘green’ in the environment with process and audience vote as well Amsterdam case strongly focused tion, but also as more knowledge use of sustainable as monetary prizes; and a large cel- quantitative feedback on consultation, and hence was also is exchanged in the process (learn- energy, but also in its ebratory, concluding event. Judging on the quality of the evaluated by the organisers in terms ing). While the cases share some of living conditions by the very professional video pres- choices. of the quantitative quality of the the rationale, the Amsterdam case entations of the teams online, the advice (e.g., payback period of pro- leaned more towards ideas of inclu- (greenery). Ecocraft digital twin with its familiar posed solar panel/insulation solu- sive decision-making, whereas the Minecraft look has certainly helped ” tions). The different reports were not Warsaw case seemed more focused ” the children to present their ideas, evaluated against each other with a on normative motivations of activat- Both cases pay ample attention but true participatory consultation ‘best report’ presented to external ing children to make sure they are to preparing the participants in does not seem the ultimate aim. parties, but the focus was on ag- heard. providing information and a prio- gregating the quantitative input for ri knowledge building. There are The Amsterdam case has no compe- a next decision making level. The Note that this does not imply any- brochures, lesson plans and, in the tition element. There is no judging children enjoyed the use of the Mi- thing about the role of public partic- case of Warsaw, home assignments beyond teachers grading assign- necraft environment, though some ipation for either city administration and field trips. Both cases try to ments and there was no prize to be found it somewhat monotonous. in general, but only these two spe- ensure sufficient awareness of the won. The project did not lead up to a cific experiments on the use of Eco- participants. However, in the next public event with presentations and Several reasons have been men- craft in involving children in the en- stages of the experiments, the two school children were required to par- tioned to choose for a broader, more ergy transition. Also, in spite of the diverge. ticipate, as this was part of a regular inclusive decision-making process ‘ladder with rungs’ metaphor often lesson plan. Here, also in absence of (e.g., Gluck et al., 2013; Reed et al., used to describe the different levels The Warsaw case strongly focus- field trips or broader aims of improv- 2018); including normative motiva- of participation, higher is not neces on achieving engagement, chal- ing the neighbourhood, the digital tions on democratic rights of individ- essarily better and, hence, neither 164 Involving children in the renewable energy transition Involving children in the renewable energy transition 165

Should involvement“ in energy transition case, Warsaw nor Amsterdam, is per on the type of object (product) and and participatory plan- References decision support system for local the people of the place se ‘better’. They are different, and situation. Should involvement in ning follow a form of Arnstein, S.R. A ladder of citizen neighbourhood energy planning. involved - youth included. these differences point to the design energy transition and participatory participation. Journal of the Amer- Energy Policy 116 (May), 2018, Quality Innovation Prosper- standard (cognitive) parameters relevant in considering planning follow a form of standard ican Institute of Planners 35 (4), 277-288. ity 21 (1), 2017, 119-150. the use of a digital twin with gaming (cognitive) learning or a form of ex- learning or a form of 1969, 216-224.. IEA, 2017. World Energy Outlook. Sieber, R. Public partici- elements. periential learning? Facts or feelings? experiential learning? Brown, G., Kyttä, M. Key issues Paris: International Energy Agency. pation geographic informa- As always, more research is needed and research priorities for public Levine, P. 2015.The Future of De- tion systems: a literature Facts or feelings? On a more abstract level, the two ex- to start formulating an answer to participation GIS (PPGIS): A synthe- mocracy: Developing the Next Gen- review and framework. periments seem to have a different this question. As always, more sis based on empirical research. Ap- eration of American Citizens (Civil Annals of the Association choice in type of attitude formation research is needed to plied Geography 46, 2014, 122-136. Society: Historical and Contempo- of American Geographers among children. A ‘standard learning start formulating Buchy, M., Hoverman, S. Under- rary Perspectives). University Press 96 (3), 2006, 491-507. hierarchy’, such as in the Amsterdam standing public participation in for- of New England, Lebanon, NH. Steinitz, C. A framework an answer to this case, is based on cognitive informa- est planning: a review. Forest Policy Reed, M.S., Vella, S., Challies, E., for geodesign: changing tion processing. In this learning hier- question. and Economics 1 (1), 2000, 15-25. De Vente, J., Frewer, L., geography by design. ESRI archy, attitude starts with learning Drazkiewicz, A., Challies, E., Hohenwallner-Ries, D., Huber, T., Press, Redlands, CA. 2012 facts and information about ener- ” Newig, J. Public participation Neumann, R.K., Oughton, E.A., United Nations. Paris gy and a resulting effect on what is and local environmental planning: Sidoli del Ceno, J., van Delden, H. Agreement. United Nations, ‘right’, eventually affecting behav- Testing factors influencing decision A theory of participation: what Paris, 2015. iour. One might view the Warsaw quality and implementation in four makes stakeholder and public strategy of involving children as case studies from Germany. Land engagement in environmental man- focusing more on an ‘experiential Use Policy 46, 2015, 211-222. agement work? Restoration Ecology attitude formation’, where attitude Eiter, S., Lange Vik, M.. Public 26 (1), 2018, 7–17. formation starts with an emotional participation in landscape planning: Scholten, H., Fruijtier, S., Dias, E., response (affect rather than cogni- Effective methods for implementing Hettinga, S., Opmeer, M., van tion), and behaviour, later followed the European Landscape Conven- Leeuwen, W. S., Linde, M., Bos, S., by (cognitive) learning about the tion in Norway. Land Use Policy 44, Vaughan, R., van Kaam, H., topic, if and when relevant. Both 2015, 44-53. van Manen, N., Fruijtier, C., learning hierarchies have their place Hettinga, S., Nijkamp, P., Geocraft as a means to support the in consumer behaviour, dependent Scholten, H.J. A multi-stakeholder development of smart cities, getting 166 167

Craglia M., Scholten H., Micheli M., Hradec J., Calzada, I., Luitjens, S., Ponti M., Boter J. Conclusions

Digitranscope set out to explore the challenges and opportunities that the digital transformation is posing to the governance of society. We focused our attention in Part A on the governance of data as a key aspect to understand and shape the governance of society. Data is a key resource in the digital economy, and control over the way it is generated, collected, aggregated, and We“ have value is extracted and distributed in society is crucial. We have explored the explored the increasing awareness about the strategic importance of data and emerging models to distribute the value generated more equitably in society. These increasing awareness findings have contributed to the new policy orientation in Europe on techno- about the strategic logical and data sovereignty and social inclusion. importance of data and The digital transformation, and the rise of artificial intelligence and the Inter- emerging models to net of Things, offer also new opportunities as shown in Part B for new forms distribute the value of policy design, implementation, and assessment providing more persona- generated more lised support to those who need it and being more participative throughout equitably in the policy cycle. The use of digital twins, gaming, simulation, and synthetic data are just at their beginning but promise to change radically the relation- society. ships among all the stakeholders in governance of our society.

” As indicated in the Introduction in Chapter 1, we did not realise at the start of the project is that what we thought was going to be an exploratory project looking 5-10 years ahead would turn instead into one providing already direct input to policy as policy priorities shifted much faster than we anticipated.

On Data Governance The most significant event that occurred during the lifetime of the project was the emergence of Artificial Intelligence (AI) as a key geopolitical battleground, particularly between the US and China. This brought AI also at the forefront of the European political attention and with it came an increasing 168 Conclusions Conclusions 169

recognition of the strategic impor- ropean strategy for data. Its find- bodies in the redistribution of value tance of data as the key asset ne- ings have also been cited to set the generated through data. A lesson cessary to develop AI applications. scene for a stakeholder workshop learned is to keep looking at the re- Technological and data sovereignty held by DG CONNECT on the speci- lationships established between citi- became key objectives of the new ficities of local data ecosystems for zens, civic society organisations, the Commission which took office in No- climate-neutral and smart commu- public sector and/or businesses, for vember 2019. nities, as part of the common Euro- controlling and using data, not only pean Green Deal dataspace. because they are highly informative The new focus on AI and data brought of the (un)balances of the current several new policy initiatives review- Overall, the outcomes of this Part of data landscape, but also because ed in Chapter 1 including a European the project contribute to public and they can be useful to advise on fu- Strategy for AI, a plan of investments scientific thinking concerned with ture policy measures. in AI coordinated with the EU Member fostering democratic forms of data States, a European Strategy for Data, governance in which more actors, On new forms of policy design a Data Governance Act, and in 2021 beyond the usual Big Tech, access, Significant policy shifts have also a risk-based regulatory framework share and use data, especially for emerged in this area during the last for AI. societally beneficial aims. With the few years and become increasingly research activities of Digitranscope, mainstream. Notably, the increasing Digitranscope contributed direct- we encouraged policy makers to con- use of big data analytics to profile ly also to the proposed regulation sider more thoroughly the social and and nudge voters following the ex- on European data governance by political implications of data sharing, ample of the commercial sector re- sharing the research results repor- beyond economic and technical as- cognised not only the power of data ted in Chapters 3-5 with the colle- pects. On the one hand, we explored but also the emotional side of decis- agues in the Commission in charge Data Governance Act. The research how citizens can be (active) subjects ion making. The mantra of evidence- of these data policy initiatives and on data sharing between businesses of data that regain control of their based decision-making that was all thus informing about emerging data and local governments (chapter 6) information, organize, and adopt the rage in the 1990s has come un- governance models including data supports the case for the creation of mechanisms to control and use data. der increasing scrutiny together with altruism and data cooperatives, new common data space for public One the other hand, we highlighted the scientific method when applied which are two key concepts of the administrations as part of the Eu- the role of civic society and public to social and political phenomena. 170 Conclusions Conclusions 171

We have seen therefore a greater te documents into a coherent and sisted Chapter 9 and gain a deeper “Of course, it is best for children to acknowledgement of the multi-face- usable repository of the European understanding through practice and be outside playing in nature. But if ted dimensions of rationality, decisi- Commission’s knowledge. direct interactions with local admi- they do sit behind the computer, this on-making, and post-normal science. nistrators and industry the main fa- game can tap their creativity to de- Communication, participation, and The Probabilistic Synthetic Popula- cets of the so-called smart cities. We sign future living spaces in harmony the use of narratives have gained tion modelling described in Chapter were also able to leverage the digital with the environment. The next ge- currency exploiting also the new op- 8 contributed directly to the work twin of the Netherlands and the Eco- neration of urban planners will see portunities of the digital transition, of the JRC Corona Virus Task Force craft plug-in developed by the Dutch how much our lives depend on this from the booming of citizen-genera- which advises the Commission on EduGIS Foundation to raise the awa- balance,” said UNEP’s Europe Direc- ted content for science and policy to potential policies and strategies to reness of young adults on the trade- tor Jan Dusik at the ‘Liveable Smart the development of digital twins for address the pandemic and its epi- offs needed in the energy transition Cities by Design’ event held in the policy simulation, co-creation, and demiological and socio-economic in two schools in Amsterdam and Dutch capital. communication. Within this chan- effects. A new project has now star- Warsaw as described Chapter 10. ging landscape, Digitranscope has ted with the Dutch Central Bureaus Moreover, Digitranscope was able Environmental education is vital to contributed in three main ways: for Statistics to develop the model to contribute to a big event in the raising awareness on and achieving further, validate it against the sta- stadium of the Ajax football team in the Sustainable Development Goals. SeTa, the semantic text analysis tool tistical data held by the Bureau and Amsterdam where 500 kids used the A target for citizens to participate developed in Digitranscope descri- provide advice to the Dutch govern- digital twin of their city to design a more in sustainable urban settle- bed in Chapter 7 started as a pilot ment on covid19-related policies. new sustainable neighbourhood. In ment planning is included under project but has now been institutio- The ambition is then to extend this that occasion, UN Environment Pro- Goal 11 on ‘sustainable cities and nalised in the JRC by the Unit respon- collaboration between the JRC and gram (UNEP) signed a partnership communities,’ while Goal 7 aims for sible for text and data mining and statistical agencies further, invol- agreement with the Dutch EduGIS ‘affordable and clean energy’. With has been made available throughout ving also EUROSTAT, to create a Eu- Foundation. Under the agreement, these experiments and the follow- the European Commission to sup- rope-wide synthetic population base geospatial data tools will allow the up promoted by UNEP worldwide, port the work of all colleagues requi- for policy simulation and analysis. game to map territories around the Digitranscope was able to show the red to do an impact assessment, ex- globe and simulate environmental value of digital twins and gaming as ante or ex-post of European policies. Digital Twins: the project was able challenges related to achieving the key assets for policy co-creation and Its enormous value is to have turned to experiment with the digital twins Sustainable Development Goals. testing, and for engaging the new hundreds of thousands of separa- of Amsterdam and Duisburg as de- generations of citizens in the decis- 172 Conclusions Conclusions 173

We are“ only A take-away“ at the beginning of message from the ion-making of today that will affect the digital transition of rent digital transformation. On the pean level thorough the work of the Digitranscope journey above all their futures. society, and at the early contrary, we witness increasing po- JRC, at the national level through the is that the governance of larization in society and the political new projects we have started with stages of equipping our digitally-transforming As the Digitranscope project comes discourse, and growing inequality the Dutch geographic council and to its conclusion we are conscious ourselves with the between rich and poor, and among the statistical agency, and locally society is challenging and that there is still much work to do. necessary theoretical different regions, nations, and con- through the network of wonderful complex, full of oppor- We are only at the beginning of the tinents. The covid-19 pandemic has and committed colleagues we have frameworks, regulatory tunities and pitfalls, but digital transition of society, and at illustrated these dangers well with developed during the three years of the early stages of equipping oursel- instruments, and the effects of both the health crisis the project. that ultimately it is up to ves with the necessary theoretical networks of partnerships and the increasing transitions to- all of us to shape it, frameworks, regulatory instruments, and international wards digital platforms and services If there is a take-away message we cannot afford and networks of partnerships and hitting the most vulnerable groups from the Digitranscope journey we alliances necessary to leave it to international alliances necessary (the elderly, children, migrants, mi- have described in this volume is to try and shape our futures effec- to try and shape norities) worst. How we can channel that the governance of our digitally- others. tively. As Steven Luitjens reminded our futures the digital transformation so that it transforming society is challenging us in Chapter 2 governments needs helps reduce inequality and injustice and complex, full of opportunities effectively. ” to step up their actions to help gui- rather than increase them remains and pitfalls, but that ultimately it is de the process, build capacity inside a key challenge. We were only able up to all of us to shape it, we cannot public administrations and society ” to address these issues partially in afford to leave it to others. through education and investments Digitranscope looking for example in research and innovation, develop at emerging models to redistribute greater capacity for foresight stu- more equitably the added value of dies to try and anticipate change data, or ways to engage citizens and and foster a culture of experimenta- children in particular in taking the tion without fear of making mistakes. advantage of digital tools to sha- The necessary conditions for this are pe their future. There is much more however openness, transparency work to do, but we are fortunate that and inclusiveness. These important we will be able to continue the work principles are not a given in the cur- started in Digitranscope at the Euro- 174 175

List of boxes List of tables 1. Selection of Citizen-Generated Data projects 3.1 Analytical dimensions...... 23 in a nutshell...... 44 4.1 Definitions: Platform Co-operatives and Data Co-operatives ...... 33 List of figures 4.2 Taxonomy for Platform Co-operatives and 1.1 Evolutionary landscape in data analytics...... 10 Data Co-operatives ...... 35 3.1 Vignette of the data governance models 4.3 Case Identification by Typology...... 36 examined in the chapter...... 25 5.1 Purpose of data collection in the selected projects. . 44 5.1 Types of collected data...... 45 5.2 Reported uses of CGD by public organisations. . . . 45 7.1 Access to education and health for single-parent 6.1 List of cities involved in the study...... 53 families and elderly people...... 62 6.2 Summary of the operational modes for accessing 7.2 Conceptual model of the synthetic population private data contextualized in the discourses and for France...... 64 experiences of twelve local administrations...... 57 7.3 Influx-outflux of French commuters 2016...... 65 10.1 Overview of the spread in different indicators 8.1 On-the-fly generated semantic map between the groups participating in the public surrounding the term ‘audits’...... 69 participatory planning session...... 84 8.2 Evolution of the term ‘auditors’ over decades, 10.2 Impact analysis of the overlaid plan of the children. . 85 reflecting contextual change in policy documents 10.3 Possible contribution of game elements vs and technical reports...... 70 digital twin at different levels of public participation. 86 8.3 Features common to impact assessments and audit. . 71 10.4 Warsaw (green) vs Amsterdam (blue)...... 87 9.1 Intelligent Dashboard Amsterdam: Mobility view. . . 75 9.2 Smart City Platform Duisburg...... 78 10.1 Context, conditions, game and digital twin as hierarchical set of elements ...... 85 176 177

About the Authors

Massimo Craglia is a Senior Expert at the European Commission Joint Research Information Agency, and between 2000 and 2004 Head of the Informatics De- Centre, Digital Economy Unit, responsible for projects addressing Digital Trans- partment of the Czech Ministry of the Environment. formation and the socio-economic impacts of Artificial Intelligence in different economic sectors, new forms of governance in digitally-transformed societies, Igor Calzada, is Scientific Project Officer at the European Commission’s Joint and the evolution of the space data economy and the geospatial sector. He is Research Centre (JRC), Centre for Advanced Studies. Since 2012 he is a sen- the coordinator of the Digitranscope project. ior researcher at the University of Oxford, Urban Transformations ESRC and Future of Cities Programmes at COMPAS, leading several research projects Henk Scholten is professor in Spatial Informatics at the School of Business and on smart cities and city-regions: (i) EU-H2020-Smart Cities and Communi- Economics of the Vrije Universiteit Amsterdam. He is also founder and CEO of ties-SCC-01-2015-691735-Replicate (2016<), (ii) ESRC-Urban Transformations Geodan Holding, one of the largest European companies specialised in Geo-IT. (2016-2019), and (iii) City-Regions project funded by Ikerbasque (2012-2014) He is the lead scientist of the Digitranscope project at the Centre for Advanced and the RSA (Regional Studies Association) (2014-2015). His main research Studies of the European Commission’s Joint Research Centre. Further, he is pro- interest draws on how digital transformation processes driven by AI disruption gramme director of UNIGIS Amsterdam, distance-learning MSc. in GIS. in the post-GDPR current context are altering techno-political and democratic conditions of data governance for the emergence of new algorithmic citizenship Marina Micheli is Scientific Project Officer at the European Commission’s Joint regimes in European (smart) cities and regions. Research Centre (JRC) Centre for Advanced Studies since 2018. Previously she was a senior researcher and teaching associate at the Institute of Commu- Steven Luitjens is program director within the Dutch government in a variety nication and Media Research of the University of Zurich and a post-doctoral of digital transformation projects. Formerly he was head of the department of fellow at the Department of Sociology and Social Research of University Mi- information society and government within the Ministry of Interior Affairs and lano-Bicocca, where she earned her PhD in 2013. Her research revolves around Kingdom Relations, and of an agency responsible for a range of generic solu- technology innovation, adoption and use examined with a socio-cultural and tions for public service delivery. Within the Digitranscope project, Steven espe- critical perspective. cially helps to bridge the gap between politicians, policy makers and scientists.

Jiri Hradec is Scientific Project Officer at the European Commission’s Joint Re- Marisa Ponti is Assistant Professor at the Department of Applied IT, University of search Centre (JRC) Centre for Advanced Studies exploring personalised policies Gothenburg, Sweden, and was a scientific officer at the JRC in 2018 contributing through data mining, machine learning and econometric modelling. He spe- to the activities of the Digitranscope project with respect to data governance cialises in the use of text mining, machine learning, and deep neural networks and citizen-generated content. Marisa is interested in how citizen-generated in support to policy across a range of projects and departments at the JRC. data can serve the public good by contributing to policy making and the public Between 2005 and 2012 he was Director of CENIA, the Czech Environmental sector. 178 179

This publication is a report by the Joint Research Centre (JRC), the European Commission’s science and GETTING IN TOUCH WITH THE EU knowledge service. The scientific output expressed does not imply a policy position of the European Commission. Neither the European Commission nor any person acting on behalf of the Commission is In person responsible for the use that might be made of this publication. All over the European Union there are hundreds of Europe Direct information centres. You can find the address of the centre nearest you at: https://europa.eu/european-union/contact_en More information on CAS can be found on the EU Science Hub at https://ec.europa.eu/jrc/en/research/ centre-advanced-studies On the phone or by email JRC119959 Europe Direct is a service that answers your questions about the European Union. You can contact this service: European Commission, 2020 ● by freephone: 00 800 6 7 8 9 10 11 (certain operators may charge for these calls), © European Union, 2020 ● at the following standard number: +32 22999696, or ● by electronic mail via: https://europa.eu/european-union/contact_en The reuse policy of the European Commission is implemented by Commission Decision 2011/833/EU of 12 December 2011 on the reuse of Commission documents (OJ L 330, 14.12.2011, p. 39). Reuse is authorised, provided the source of the document is acknowledged and its original meaning or message is not distorted. The European Commission shall not be liable for any consequence stemming from the FINDING INFORMATION ABOUT THE EU reuse. For any use or reproduction of photos or other material that is not owned by the EU, permission must be sought directly from the copyright holders. Online Information about the European Union in all the official languages of the EU is available on the Europa All content © European Union, 2020, except: page 1: coroichi999; page 6: Vjom; page 10, 14, 18, website at: https://europa.eu/european-union/index_en 22, 26, 30: sunward5; page 13: Musicman80; page 21: Drobot Dean; page 25: sdecoret; page 27/28: saicle; page 28/29: kras99, source: stock.adobe.com (unless otherwise specified) EU publications You can download or order free and priced EU publications from EU Bookshop at: https://publications.eu- How to cite this report: M. Craglia, E. Gomez Gutierrez, J. Glovičko, S. Manzan, H. Scholten, L. Barbaglia, ropa.eu/en/publications. Multiple copies of free publications may be obtained by contacting Europe Direct I. Calzada Mugica, V. Charisi, S. Consoli, J. Hradec, M. Micheli, M. Miron, L. Tiozzo Pezzoli, S. Tolan and E. Tosetti, CAS, G. Amran and S. Sutherland editor(s), Publications Office of the European Union, Lux- or your local information centre (see https://europa.eu/european-union/contact_en). embourg, 2020.

The authors would like to thank Stephan Lindner for his work on layout and design of the printed version. KJ-02-20-215-EN-C

180

The European Commission’s science and knowledge service Joint Research Centre

JRC Mission As the science and knowledge service of the European Commission, the Joint Research Centre’s mission is to support EU policies with independent evidence throughout the whole policy cycle.

EU Science Hub ec.europa.eu/jrc

@EU_ScienceHub EU Science Hub- Joint Research Centre EU Science, Research and Innovation EU Science Hub

DOI 10.2760/345877 ISBN 978-92-76-17590-2