Smarter Intervention in Complex Systems

Manifesto for SMARTER INTERVENTION IN COMPLEX SYSTEMS

Mark Fell, Managing Director, Carré & Strauss with an Introduction by Hanne Melin, Policy Strategy Counsel, eBay Inc.

INTERVENTION PRINCIPLE

STOP?

INTERVENTION OB MECHANISM T S C E A R V

INTERVENTION

E MINDSET

I E

T D

‘SMARTER INTERVENTION’ About the Author

Mark Fell is the Managing Director of Carré & Strauss, a firm that specialises in decision support systems for professional strategists. During the course of a career in strategy and governance he has advised a wide range of organisations, including Arup, BSkyB, Chiquita, Diageo, eBay, Estée Lauder, Europe’s blank recording-media industry, the European Cancer Organisation, the European Cooperative Movement, Hydro, Knauf Insulation, LVMH Group, Mastercard, Nutricia, PayPal, Richemont Group, Shecco, UPS and Vodafone. Mark holds a Masters Degree in Euro- pean Politics and Administration from the College of Europe, Bruges, a First Class Honours Degree in Politics from Edinburgh University and has studied at the Institut d’Etudes Politiques, Strasbourg. In his free time he enjoys testing himself through endurance running, having com- pleted races at the North Pole, in Antarctica, the Sahara, Amazon, Himalaya and Alps, as well as an Ironman Triathlon.

Acknowledgements

This manifesto draws extensively on the work of David Cliff, Mary Cummings, Len Fisher, Tim Harford, Jaron Lanier, Donella Meadows, Melanie Mitchell, Jason Pontin, John Sterman, Steven Strogatz and Duncan Watts, as well as the labour of countless others too numerous to mention here. It is also the fruit of my interaction over the years with clients and colleagues from a wide array of sectors and geographies. My frequent collaborations with Hanne Melin of eBay are of particular note in this regard. As ever, any errors and omissions are solely of my own making.

EXECUTIVE SUMMARY ...... 6

INTRODUCTION BY HANNE MELIN ...... 8

LIGHTS - ILLUMINATING THE CHALLENGE ...... 10

A manifesto for change ...... 10

CAMERA - BRINGING THE CHALLENGE INTO FOCUS WITH “SYSTEMS DYNAMICS” ...... 12

The global system defined ...... 12

‘Systems Dynamics’ introduced ...... 14

The example of product awareness as a system ...... 18

Leverage points in systems identified - ‘Twelve Heuristics’ ...... 20

ACTION - SETTING OUT A METHODOLOGY TO MEET THE CHALLENGE ...... 24

A methodology for ‘Smarter Intervention in Complex Systems’:

- The ‘Mindset’ ...... 26

- The ‘Mechanism’ ...... 34

- The ‘Principle’ ...... 46

CONCLUSION ...... 50

‘Fast Forward to the Future’ ...... 50

ENDNOTES ...... 56 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS

Executive Summary

far more iterative ’trial-and- to be made for why a given ‘Smarter Intervention in Complex Systems’ error’ approach. One in which intervention agent should be means pursuing a new intervention the decisions that we take are involved in the OODA loop - i.e. mindset, mechanism and principle. survivable. the best agent to solve this type of problem. The aim is to get all (2) MECHANISM: The ‘OODA Loop’ of these different agents cycling (‘Observe-Orient-Decide-Act’) is around the loop as a team, at the a practical method for actioning right tempo and to be able to This manifesto is a call for better decision this trial-and-error approach. hold them to account. making in our social, technological, environ- We can cycle this loop by using mental, economic and political systems. experts or non-experts. They may be individuals or groups It marshals a series of disparate concepts of individuals - i.e. ‘crowds’. into a framework. This framework aims to Increasingly the loop is being facilitate ‘Smarter Intervention in Complex cycled by software algorithms. Systems’. It is a methodology. Each of these ‘Intervention Agents’ has its strengths In doing so the manifesto draws extensively and weaknesses. These are on the ideas, words and images of many examined at length in this professionals from a wide array of disci- Manifesto. Agents can emanate plines. ‘Systems Dynamics’ and ‘Complex- from either the public or the ity Theory’ helps to guide us through this private sector. The key message process. is that none of these agents is the panacea to cycling the ‘Smarter Intervention’ is based on three loop. In each specific case we tenets: need to identify the appropriate combination of these agents. (1) MINDSET: The systems we We require an ‘Intervention Mix’. are seeking to shape are often (3) PRINCIPLE: To achieve this mix ‘complex’. They are ‘nonlinear’. A a new ‘Intervention Principle’ is does not always lead to B to C to proposed. To the extent that D to E. Instead they can be full a system can successfully self- of surprises. We cannot ‘predict- organise, it should be left to and-control’ their behaviour. do so. Where intervention is Instead we need to pursue a necessary, then the case has

6 7 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS

Introduction by Hanne Melin

policymaking in order to not only step into The traditional answers to these two ques- In parallel, the European Commission has “While all other sciences have advanced, the 21st Century but keep moving in synch tions have been co-regulation and consulta- reorganized one of its Directorates-General that of government is at a standstill - little with it. tions with the public. But are these the best to better “face the challenges of the next ten better understood, little better practiced answers when today’s world requires us to years in a digital world”5 by inter alia setting now than three or four thousand years Take robotic technologies, expected by think less in ‘form’ and more about ‘func- up a Complex Systems unit. ago.” many to revolutionise our lives even more tion’? profoundly than the mobile phone. How I see these actions as products of an evolu- John Adams should the law think about robots, asks a re- Last summer, I was listening to Spotify tion in policy thinking on how to “grapple Second President of the United States cent essay from Washington University. The founder Daniel Ek on the radio. Why think with far more complexity than before” – as authors warn that “the law almost always about the music album as something fixed Chan Lai Fung, then Chairman of Singa- considers new technology as merely a new to 12 tracks, he asked. When artists can sim- pore’s Smart Regulation Committee, defined form of something else”,1 which sounds a lot ply release songs once finalised, function the challenge already in 2006.6 like what happened in Sweden in 1981. rather than form matters. In the early 1980s, commercial attempts to So where do we go from here? European develop mobile phone services in Sweden Indeed, two interrelated questions – both at We could approach policymaking with a Commissioner Neelie Kroes is right when were resisted. The reason being the fright- the heart of good lawmaking in changing similar mentality. If we freed ourselves of she says that we need “new legal frameworks ening legal uncertainty when a new tech- times - have been discussed for years. the constraints of traditional form, how and a new way of thinking”.7 We need the nology was not covered by existing rules. would we go about releasing the outcome framework that pulls together the evolution Experience has surely taught us to better The first is finding the right tools for prob- we’d like to see? in thinking and importantly lets also poli- cope with change but have we actually lem-solving. ”Regulation is one of the many cymaking detach from form, rigid division found the policy framework for doing so ways of implementing public policy but it is The Hans-Bredow-Institute study pointed between regulators and the public, and a with confidence? not necessarily the best way of solving a given towards the need for a system governance preference for control. problem nor is it the only way”, stressed the approach to “work out better ways to achieve The European Commission launched in 2001 Mandelkern Group report that formed the policy goals under changing conditions” To my knowledge, no one has yet proposed 2005 the Better Regulation agenda aimed the basis for the EU Better Regulation in view of it being impossible to control that framework. Now, with this Manifesto at improving lawmaking in a fast moving agenda.2 social systems. someone finally has! Europe. In 2010, the Commission evolved this agenda into a Smart Regulation strate- The second is involving the appropriate ac- Indeed, the European Parliament embarked gy to more effectively embed the objectives tors in decision-making. A 2006 study com- in 2011 on an exercise to identify changes of Better Regulation into the whole policy missioned by the European Commission needed to prepare for the complex environ- cycle. and carried out by the Hans-Bredow-Institut ment resulting from a “multi-polar world stressed that in increasingly complex and where governance is more and more a multi- These are the right steps but I am convinced rapidly changing societies knowledge is level one, involving multiple actors in decision Hanne is Policy Strategy Counsel at eBay and that more evolution – in fact transforma- held by different actors.3 making and implementation”4 and where formerly associate in the EU competition law tion – now needs to happen in the area of technologies are accelerating change. practice of Sidley Austin LLP based in Brussels

8 9 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Illuminating the Challenge

A Manifesto for Change

To do so this Manifesto: “… the challenges we face are real. Finance is in tumult, and while worries about a (1) Begins by framing the big banking crisis may have ebbed, fears of a picture, namely defining the crisis in public finances are running high. global system in which we Even without additional financial reversals, operate and from which our overall economic prospects look bleaker challenges emanate than just a few years ago. The global order also seems more uncertain. Prosperity (2) Proceeds to introduce ‘Systems and power are shifting to new places and Dynamics’ as a lense through peoples. Old political doctrines and divi- which to understand any system, sions no longer seem viable. Technology be it social, technological, and media are changing before our eyes. environmental, economic or So, apparently is the natural environment political itself.” (3) Identifies 12 ‘Leverage Points’ that have surfaced over the 1 Pankaj Ghemawat, World 3.0 decades as experts have studied all kinds of different systems

(4) Sets out a methodology for practically applying this Time and again we are failing to manage systems knowledge, namely the ‘complex systems’ in an increasingly inter- requisite intervention ‘Mindset’, dependent world, be they complex social, ‘Mechanism’ and ‘Principle’ technological, environmental, economic or political systems. This is unsustainable.

This Manifesto therefore makes the case for ‘Smarter Intervention in Complex Systems’.

Not because Smarter Intervention is the panacea for a more sustainable future, but because without it, the chances of success are even slimmer.

10 11 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Bringing the Challenge into Focus with “Systems Dynamics”

The Global System Defined

FIGURE 1 (opposite page) - THE GLOBAL SYSTEM L ECO LOBA SYSTE ‘How Everything Fits Together’ G M Source: Adapted from “The Limits of Growth” by Meadows, Randers and Meadows1 solar energy high- “THE BIG PICTURE” al relations and institutional arrangements. Planetary grade It is here that we can evaluate whether To make the case for ‘Smarter Intervention ‘social value’ is being created or destroyed. Sources energy in Complex Systems’ we first need a concep- tual framework - a lense through which to In turn, our society is nested in a bigger materials bring the challenge into focus. global ecosystem that will or will not ulti- & fossil fuels low- mately sustain it. At this level we need to ‘Our Society’ This begins with a view of how social, concentrate our efforts on conserving and A Subsystem grade technological, environmental, economic nurturing ‘environmental value’. energy and political (‘STEEP’) systems relate to one another. ‘Globalisation’ entails the increasing interaction of all these systems regardless of wastes Planetary The standard way to define the economy is geography. & Sinks heat as a “system of production, distribution and pollution loss consumption of goods and services”. This is ‘Intervention’ means any initiative that seeks where “economic value” is created or de- to influence a system’s behaviour. This may stroyed. take many forms, including government regulation, subsidies and fiscal incentives. It However this Manifesto adopts a more ho- may be a decision made by business or an listic view of the economy, seeing it as being action by civil society. Increasingly it mani- not just about production, distribution and fests itself as algorithmic intervention - think consumption, but rather as “the set of sys- of speed cameras, driverless trains and ‘High Political tems through which a society seeks to satisfy Frequency Trading’. its needs”. This expanded definition there- Systems fore also includes our political and technol- What therefore is this recurrent theme, ogy systems. namely a ‘system’? Economic Technology GLOBALISATION Our society is an expression of how all of Systems Systems INCREASING INTERPLAY OF ALLACROSS THESE BORDERS SYSTEMS these systems interact - the resulting cultur- ‘The set of systems through which society seeks to satisfy its needs’

12 13 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Bringing the Challenge into Focus with “Systems Dynamics”

Systems Dynamics Introduced

WHY SYSTEMS DYNAMICS “The feedback loops can involve risk- management systems, and can be Pioneered at MIT, ‘Systems Dynamics’ is an driven by changes in market volume or approach to understanding the behaviour volatility, by market news, and by delays of complex systems over time. in distributing reference data.”2

Donella Meadows was a leading proponent This is the language of Systems Dynamics - of systems dynamics and a key voice in one of ‘feedback loops’ and ‘delays’. what has become known as the ‘sustainability movement’, the international effort Equipped with the ‘lense’ afforded by this to reverse damaging trends in the social, discipline Meadows foresaw this result al- technological, environmental, economic or ready back in the early 1990s stating that: political systems. “...the great push to reduce information Her work is widely recognized as a formative and money-transfer delays in financial influence on hundreds of other academic markets is just asking for wild gyra- studies, government policy initiatives, and tions.” 3 international agreements. She observed that: THE BASICS This Manifesto will draw extensively on ing’ system that caused this behaviour has Meadow’s work (particularly her book “The industrial society is just beginning recently discovered that: With this said and using Meadows explana- “Thinking in Systems”)1 to set out the fun- to have and use words for systems, be- tion (see her paper “Leverage Points - Places damentals of systems and their leverage cause it is only beginning to pay atten- “... in specific circumstances, a key type to Intervene in a System“)5 let us start with points. tion to and use complexity”.4 of mechanism can lead to significant the basic diagram on the next page: instability in financial markets with What follows in this section at first sight ap- A fact brought home by the September computer based trading (CBT): self- “The ‘state of the system’ is what- pears abstract. However consider the fact 2011 edition of the Harvard Business Review reinforcing feedback loops (the effect of ever standing ‘stock’ is of importance: that on 6 May 2010 ‘The Flash Crash’ wiped whose cover story headlined with, “Embrac- a small change looping back on itself amount of water behind the dam, nearly $1,000bn off the value of US shares ing Complexity - You Can’t Avoid It, But Your and triggering a bigger change, which amount of harvestable wood in a forest, for a period of several minutes. Had the Business Can Profit From It”. again loops back and so on) within well- number of people in the population, losses not been recovered, the Dow would intentioned management and control amount of money in the bank, what- have suffered one of its biggest one-day Think of ‘systemic risk’, a term that policy- processes can amplify internal risks ever. System states are usually physical falls in history. makers are just starting to grapple with as and lead to undesired interactions and stocks, but they could be nonmaterial they seek to tackle the financial crisis. outcomes. ones as well: self-confidence, degree of And that the UK government’s subsequent trust in public officials, perceived safety investigation into the ‘High Frequency Trad- To properly understand complex systems of a neighbourhood. we need to look at them through the Sys-

tems Dynamics lense.

14 15 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Bringing the Challenge into Focus with “Systems Dynamics”

FIGURE 2 - THE IMPACT OF THE 6 MAY 2010 FLASH CRASH ON THE DOW ‘Not understanding complexity and ‘systems dynamics’ can have potentially serious consequences’ In ow Out ow Source: Bloomberg STATE OF THE SYSTEM FIGURE 3 - HARVARD BUSINESS REVIEW - EMBRACING COMPLEXITY ‘We are just starting to wake up and pay attention to complexity and systems dynamics’ Source: Harvard Business Review Perceived State FIGURE 4 - THE FUNDAMENTALS OF A SYSTEM ‘Systems consist of stocks, flows, feedback loops and goals’ Discrepancy Source: Adapted from “Leverage Points - Places to Intervene in a System” by Donella Meadows

Goal

“There are usually ‘inflows’ that increase tap and outflowing drain. If the inflow tween the goal and the perceived state temperature. Suppose the hot inflow is the stock and ‘outflows’ that decrease rate is higher than the outflow rate, the is zero). Then you turn the water off. connected to a boiler way down in the it. Deposits increase the money in the water gradually rises. If the outflow basement, four floors below, so it doesn’t bank; withdrawals decrease it. River rate is higher than the inflow, the water “If you start to get into the bath and respond quickly. And you’re making inflow and rain raise the water behind gradually goes down. The sluggish re- discover that you’ve underestimated faces in the mirror and not paying close the dam; evaporation and discharge sponse of the water level to what could your volume and are about to produce attention to the water level. The system through spillway lower it. Births and be sudden twists in input and output an overflow, you can open the drain for begins to get complex, and realistic, and immigrations increase the population, valves is typical; it takes time for flows of awhile, until the water goes down to the interesting.” deaths and emmigration reduce it. water to accumulate in stocks, just as it desired level. Political corruption decreases trust in take time for water to fill up or drain out There is also a second type of feedback public officials; experience of well func- of the tub. Policy changes take time to “Those are two negative feedback loops, loop, namely a positive or ‘reinforcing’ one. tioning government increases it. accumulate their effects. or correcting [“balancing”] loops, one These are self-enhancing, leading to expo- controlling the inflow, one controlling nential growth or to runaway collapses over “In sofar as this part of the system con- “The rest of diagram shows the informa- the outflow, either or both of which you time - a snowballing avalanch effect. Think sists of physical stocks and flows - and tion that causes the flows to change, can use to bring the water level to your of the escalating noise when a microphone they are bedrock of any system - it obeys which then cause the stock to change. If goal. comes too close to a loudspeaker. the laws of conservation and accumula- you’re about to take a bath, you have a tion. You can understand its dynamics desired water level in mind (your goal). “... Very simple so far. Now let’s take readily, if you can understand a bathtub You plug the drain, turn on the faucet, into account that you have two taps, with some water in it (the stock, the and watch until the water rises to your a hot and a cold, and that you’re also state of the system) and an inflowing chosen level (until the discrepancy be- adjusting for another system state:

16 17 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Bringing the Challenge into Focus with “Systems Dynamics”

Example of the Product Awareness System + Sales + + Revenue R2

Donella Meadows’ explanation gives us a “While a firm controls its advertising R1 Share from Advertising clear picture of the basics of a system. Let and sales budgets, word of mouth and Industry Market Awareness us now bring these systems concepts more media attention are largely outside Demand Share + + from Ads + Advertising to life. Take the example of how consumers of their direct control ... As sales boost + + + R4 become aware of a seller’s products. MIT’s the installed base and the number of + Share from Direct Professor John D. Sterman examines this customers who have experience of the Sales E ort system in his book “Business Dynamics”. He product, favourable word of mouth Sales reasons: increases awareness, increasing total Capability demand (R5) and also persuading more Product “How do potential customers become people to purchase the products of the R3 Attractiveness aware of a firm’s products? There are 4 seller (R6). A hot product or company Awareness principal channels: advertising, direct will also attract media attention, which, from Sales E ort sales effort, word of mouth, and media if favourable, stimulates additional attention. Each of these channels cre- awareness and boosts market share still ates positive feedback [loops]. more (R7-9).”1 These are two views of the same product awareness system “In most firms their advertising budget ... There are 4 channels to consumers - ‘Advertising’ & ‘Direct Sales Capability’ (depicted above) grows roughly as the company revenue And ‘Favourable Word of Mouth’ & ‘Media Reports’ (shown below) grow. Larger advertising budgets have two effects: (1) more potential consum- + ers are made aware of the item and + Sales choose to enter the market (loop R1); (2) to the extent the advertising is effective, R5 + more of those who are aware and in R6 the market are likely to buy the product Awareness Installed from WOM Customer offered by the company (R2). Similarly, + Industry Market Share from + WOM Base Demand Share Sales or Firm the larger the revenue of the firm, the + + + greater the sales budget. The more sales FIGURE 5 (opposite page) - TWO VIEWS + + Growth Rate representatives, and the greater their OF THE SAME SYSTEM Favourable skill and experience, the more calls they ‘One view capturing the advertising & direct Word of + R9 can make, the more time they can spend sales channels, the other view the favour- Mouth with customers, and the more effective able word of mouth and media reports Hot Product R7 their calls will be, increasing both total channels’ Product E ect industry demand (R3) and the share of Source: Both figures adapted from “Business Attractiveness Awareness from R8 total demand won by the firm (R4). Media Reports Dynamics” by John Sterman Share from Media Reports + Media Perception of Hot Product Reports + +

18 19 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Bringing the Challenge into Focus with “Systems Dynamics”

FIGURE 6 - TWELVE LEVERAGE POINTS IN SYSTEMS IN ORDER OF EFFECTIVENESS Leverage Points in Systems “‘Paradigms’ are the most effective, ‘buffers” and ‘numbers’ the least”

Source: Carré & Strauss drawn up from “Thinking in Systems” by Donella Meadows

(6) REINFORCING FEEDBACK “Numbers, the size of flows, are dead last LOOPS: The strength of the gain on my list of powerful interventions. Did- of driving loops dling with the details, arranging the deck SCENDING PARA (7) INFORMATION FLOWS: The AND DIGM chairs on the Titanic. Probably 90 - no 95, TR PARADIGMS S no 99 percent - of our attention goes to structure of who does and does GOALS parameters, but there’s not a lot of leverage not have access to information 1 ORGANISAT in them.” SELF ION (8) RULES: Incentives, punishments, RULES constraints TION Donella Meadows ORMA FLOW INF S (9) SELF ORGANIZATION: The ING FEEDBAC RC K L power to add, change or evolve FO EEDB OO IN ING F ACK P system structure RE NC LO S LA DELAYS O A P B W S S Donella Meadows identifies 12 places to (10) GOALS: The purpose of the FLO TRU 2 & C intervene in a system. In increasing order system K FFER TU C BU S R of effectiveness they are: O E T MBERS S (11) PARADIGMS: The mindset S NU (1) NUMBERS: Constants and out of which the system – its MOST MOST LEAST EFFECTIVE parameters such as subsidies, goals, structures, rules, delays, EFFECTIVE EFFECTIVE taxes and standards parameters – arises INFORMATION & ‘PHYSICAL’ PARTS INFORMATION & (2) BUFFERS: The size of stabilizing (12) TRANDSCENDING ‘CONTROL’ PARTS OF SYSTEM ‘CONTROL’ PARTS stocks relative to their flows PARADIGMS: The ability to get OF SYSTEM OF SYSTEM at a ‘gut’ level that no paradigm (3) STOCK AND FLOW is ‘true’ – that every one of these STRUCTURES: Physical systems paradigms is a tremendously ‘LEVERAGE POINTS’ and their nodes of intersection limited understanding of reality 12 Places to (4) DELAYS: The length of time Figure 6 illustrates the hierarchical nature Intervene in a System relative to the rate of system of these leverage points. How changing an changes outer leverage point - one higher in the list - can condition all those within it. (5) BALANCING FEEDBACK LOOPS: The strength of the feedbacks relative to the impacts they are trying to correct

20 21 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Bringing the Challenge into Focus with “Systems Dynamics”

about them. What you read here is still Sen demonstrated that food production “The test of a first class mind is the ability work in progress; it’s not a recipe for was actually higher in the famine years.) to hold two opposing views in the head at finding leverage points. Rather, it’s an Technology can improve crop yields or the same time and still retain the ability to invitation to think more broadly about systems for storing and transporting function.” system change.”4 food; better responses by nations and nongovernmental organizations to F. Scott Fitzgerald on paradigms Elaborating on this she continues: emerging famines have reduced their number and severity. But famines will “I have come up with no quick or easy still occur because there will always be formulas for finding leverage points in bad governments.”6 complex and dynamic systems. Give me Meadows states that: a few months or years and I’ll figure it out. And I know from bitter experience “Paradigms are the sources of systems. that, because they are so counterintui- From them, from shared social agree- tive, when I do discover a system’s lever- ments about the nature of reality, come age points, hardly anyone will believe system goals and information flows, me.” 5 feedbacks, stocks, flows, and everything 3 else about systems.” Take the example of famines. Jason Pon- tin, editor, journalist and publisher at MIT Change the paradigm - the mindset from explains: which the system emanates - and everything else changes. “Until recently, famines were understood to be caused by failures in food sup- Meadows observes that the higher the le- ply (and therefore seemed addressable verage point, the more the system will resist by increasing the size and reliability of changing it. the supply, potentially through new agricultural or industrial technologies. She is also careful to caveat this list noting But Amartya Sen, a Nobel laureate that it is tentative and that there are excep- economist, has shown that famines tions to every item that can move it up or are political crises that catastrophi- down in order of leverage: cally affect food distribution. (Sen was influenced by his own experiences. As a “I offer this list to you with much humil- child he witnessed the Bengali famine ity and wanting to leave room for its of 1943: three million displaced farmers evolution... [it is] distilled from decades and poor urban dwellers died unneces- of rigorous analysis of many different sarily when wartime hoarding, price kinds of systems done by many smart gouging, and the colonial government’s people. But complex systems are, well price-controlled acquisitions for the complex. It’s dangerous to generalize British army made food too expensive.

22 23 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Setting out a Methodology to Meet the Challenge

Smarter Intervention in Complex Systems MANY SYSTEMS ARE “NONLINEAR”

“Why are nonlinear systems so much harder to analyze than linear ones? The essential difference is that linear systems can be broken down into parts. Then each part can be GETTING TO GRIPS WITH ‘COMPLEXITY’ drawn with curves and wiggles, not solved separately and finally recombined to get the answer. This idea allows a fantastic with a straight line. If I put 100 pounds simplification of complex problems ... In this sense, a linear system is precisely equal to Before employing Meadow’s 12 leverage of fertilizer on, my yield will go up by 10 the sum of its parts. But many things in nature don’t act this way. Whenever parts of a points, we first need to get a better bushels; if I put on 200, my yield will not system interfere, or cooperate, or compete, there are nonlinear interactions going on. handle on ‘complexity’. Only then will we go up at all; if I put on 300, my yield will Most of everyday life is nonlinear, and the principle of superposition fails spectacularly.”5 have a better shot at achieving smarter go down. Why? I’ve damaged my soil intervention in complex systems. 1 with ‘too much of a good thing’.” Steven Strogatz Meadows draws attention to the fact Meadows also provides the example of although people deeply involved in a traffic flows: system often know intuitively where to find leverage points, more often than not they “As the flow of traffic on a highway push the change in the wrong direction. increases, car speed is affected only The equation states that: and others were searching for a general theory of motion that could explain slightly over a large range of car x +1=Rx (1-x ) She observes that this is because “complex” density. Eventually, however, small t t t and predict the dynamics of any object systems are inherently unpredictable. further increases in density produce a The specifics of the equation are or group of objects subject to physical ‘Nonlinearities’ in their architecture can rapid drop-off in speed. And when the unimportant. What matters is that the force, whether it be earthly or celestial. flip them from one mode of behaviour to number of cars on the highway builds components of the equation (x, 1, R, etc.) are Before Newton’s time, there were another. Take as an example, an economic up to a certain point, it can result in a readily comprehensible, but the resulting individual pieces of such a theory (e.g. system flipping from boom to bust. traffic jam, and car speed drops to zero.”2 behaviour of the equation is not. It the notions of ‘infinitesimal’, ‘derivative’, ‘integral’, etc. existed) but no one had Meadows explains that: produces erratic, chaotic and unpredictable Software can behave in complex nonlinear behaviour - see Figure 7. figured out how all these pieces fit “A ‘linear’ relationship between two ways. It is made up of lots of different together to produce a completely elements in a system can be drawn ‘if-then-else’ branches in the code and Put differently, the system’s behaviour general theory that explained a on a graph with a straight line. It’s a feedback loops. As stocks of data are is greater than the sum of its parts - it is huge range of phenomena that were relationship with constant proportions. processed changing combinations of these ‘emergent’. To date no comprehensive previously not unified. Similarly, today, If I put 10 pounds of fertilizer on my field, branches and feedback loops come to explanatory theory for complex systems we have many different pieces of theory my yield will go up 2 bushels. If I put dominate the system at different moments. exists. related to complex systems, but no one This can lead to unpredictable results. has yet determined how to put them on 20 pounds, my yield will go up by 4 Melanie Mitchell, Professor of Computer bushels. If I put on 30 pounds, I’ll get an That is why even after exhaustive testing all together to create something more Science at Portland State University, 4 increase in 6 bushels. software for any significant undertaking is general and unifying.” never entirely bug free. explains: Nevertheless, this has not prevented “A ‘nonlinear’ relationship is one in “The mathematician Steven Strogatz which the cause does not produce a A good illustration of nonlinearity is the Strogatz from offering up a top-level ‘Logistic Map’ - one of the most famous has termed this goal the quest for a framework for classifying systems on the proportional effect. The relationship “calculus of complexity”. The analogy between cause and effect can only be equations in the science of dynamical basis of their dynamics - see Figure 8. systems.3 is apt in some ways: Newton, Liebniz

24 25 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Setting out a Methodology to Meet the Challenge

There is an important role for prediction in FIGURE 7 - THE LOGISTIC MAP “You think that because you understand this approach - some ‘attempt at foresight’ - “An illustration of how complex nonlinear behaviour can result from a very simple starting point” ‘one’ that you must therefore understand but with Meadows major caveat that:8 ‘two’ because one and one makes two. But Source: Melanie Mitchell, “Complexity” you forget that you must also understand (A) MODELS: All our predictions are ‘and’”6 “models” of the world. None of these models is the real world. Meadows recounting R=2 R=4 (B) CONGRUENCE: Our models may a Sufi teaching story well have a strong congruence with the world and often do.

(C) LIMITATIONS: However conversely our models also fall A NEW INTERVENTION “MINDSET” short of representing the world In the absence of a ‘calculus of complexity’ fully. enabling us to perfectly ‘predict-and-control’ This led statistician George Box to proclaim reality we need some other mechanism 30 years ago: for effectively deploying Meadows’ twelve leverage points in a complex nonlinear “All models are wrong, but some are systems environment - for using her useful”9 ‘heuristics’ (i.e. ‘rules of thumb’) as a point of departure for systems intervention. It is when we couple ‘prediction’ with the illusion of perfect ‘control’ - that they are Melanie Mitchell elaborates on the ‘complex systems confound our expectations about Tim Harford offers some insights in his book reality - that we run into problems. “Adapt - Why Success Always Starts With nonlinear behaviour’ of ‘The Logistic Map’: the relationship between action and Failure”. As Meadows puts it: response - between input and output / “The logistic map is an extremely cause and effect. He sets out three ‘Palchinsky principles’,7 “Systems can’t be controlled, but they simple equation and is completely named after the Russian engineer who can be designed and redesigned. We deterministic: every xt maps onto one Change the value of ‘R’ in the logistic map formulated them: can’t surge forward with certainty into a and only one value of xt+1. And yet the from ‘2’ to ‘4’ and the system’s behaviour world of no surprises, but we can expect chaotic trajectories obtained from this changes radically - it flips to a new (1) VARIATION: Seek out new ideas surprises and learn from them and even map, at certain values of R, look very behavioural pattern (in fact this happens and try new things profit from them. We can’t impose our random - enough so that the logistic somewhere between R=3.4 and R=3.5). This will on a system. We can listen to what map has been used as a basis for system is complex, nonlinear, unpredictable. (2) SURVIVABILITY: When trying the system tells us, and discover how generating pseudo-random numbers on something new, do it on a scale A fuller simulation of its dynamics can be its properties and our values can work a computer. Thus apparent randomness where failure is survivable viewed at: together to bring forth something much can arise from very simple deterministic systems.”11 (3) SELECTION: Seek out feedback better than can ever be produced by our http://en.wikipedia.org/wiki/Logistic_map 10 and learn from your mistakes as will alone.” It is a simple illustration of how ‘complex’ you go along

26 27 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Setting out a Methodology to Meet the Challenge Nonlinear Linear Nonlinearity

FIGURE 8 - A FRAMEWORK FOR CLASSIFYING SYSTEMS for single species Logistic equation relaxational dynamics Overdamped systems, Bifurcations Fixed points decay Radioactive circuitRC Exponential growth equilibrium Growth, decay, or n = 1 ‘One axis tells us the number of variables needed to characterize the state of the system. The other axis tells us whether the system is linear or nonlinear.’

Source: Steven H. Strogatz, “Nonlinear Dynamics and Chaos” (van der Pol, Josephson) Nonlinear electronics Predator-prey cycles cells)(neurons, heart Biological oscillators Limit cycles oscillatorsAnharmonic Pendulum (Kepler, Newton) 2-body problem RLC circuit and spring Mass Linear oscillator Oscillations n = 2

Steven Strogatz proposes a framework for classifying systems:

“Admittedly, some aspects of the picture are debatable. You might think that

some topics should be added, or placed Quantum chaos ? Practical uses of chaos (Levinson, Smale) Forced nonlinear oscillators (Mandelbrot) Fractals Iterated maps (Feigenbaum) Chemical kinetics 3-body problem (Poincaré) (Lorenz) Strange attractors Chaos engineering Electrical structures Civil engineering, n ≥ 3 differently, or even that more axes are

needed - the point is to think about of variablesNumber classifying systems on the basis of their dynamics.

... You’ll notice that the framework also The frontier contains a region forbiddingly marked ‘The frontier’. It’s like in those old maps

of the world, where the mapmakers Economics Ecosystems systemImmune Neural networks cell synchronization Heart Josephson arrays (semiconductors) Nonlinear solid-state physics mechanics statistical Nonequilibrium Lasers, nonlinear optics Coupled nonlinear oscillators mechanics Equilibrium statistical dynamics Molecular Solid-state physics Coupled oscillators harmonic phenomena Collective n >> 1 wrote ,’Here be dragons’ on the unexplored parts of the globe. These topics are not completely unexplored of course, but it is fair to say that they lie at the limits of current understanding. The problems are very hard, because they are both large and nonlinear. The resulting behaviour is typically complicated in both space and time, Spatio-temporal complexity Spatio-temporal Viscous uids Acoustics Heat and di usion (Schödinger, Heisenberg, Dirac) Quantum mechanics Electromagnetism (Maxwell) Wave equations Elasticity Waves patterns and Continuum Life Turbulent uids (Navier-Stokes) Epilepsy Fibrillation biological and chemical waves Reaction-di usion, Quantum eld theory General relativity (Einstein) Earthquakes Plasmas Nonlinear waves (shocks, solitons) as in the motion of a turbulent fluid or the patterns of electrical activity in a fibrillating heart.”12

Note that systems such as ‘economics’ and ‘ecosystems’ fall beyond this frontier.

28 29 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Setting out a Methodology to Meet the Challenge

Frame the question. Provide the right tools FIGURE 9 - UNDERSTANDING MODELS “I can live with doubt, and uncertainty, and for it to yield the answer. Try to release ‘Models often have a strong congruence with reality but not always’ not knowing. I think it’s much more inter- beneficial system behaviours. Correctly esting to live not knowing than to have structured, a system will arrive at the Source: Carré & Strauss drawn up from “Thinking in Systems” by Donella Meadows14 answers which might be wrong.” solution.

Richard Feynman Put simply, we need a fundamentally new paradigm for problem solving. We need to adopt a new ‘Complex Systems’ mindset Congruent Incongruent based on three tenets:

(1) EMBRACE UNCERTAINTY MODEL MODEL In other words, we need to embrace a far REALITY more ‘iterative’ approach to problem solving (2) USE ‘TRIAL-AND-ERROR’ if we are to learn to ‘dance’ with increasingly REALITY complex systems in the 21st Century. (3) PROVIDE TOOLS

This means taking small steps, predicting, Recall that changing the ‘paradigm’ is the but monitoring reality constantly and being second most effective place to intervene in Everything we think willing to change the course as we find out a system - point eleven in Meadows’ twelve we know about the ... BUT NOT ALWAYS which is more about where it is leading. point list. For this is the mindset out of which the entire system arises and the basis world is a model. Our why we encounter surprises. We have to abandon the illusion of ‘predict- for regulating it. models have a strong For our models fall far short and-control’ and adopt a new ‘trial-and- error’ approach. It has concrete implications. For example, congruence with the of representing the real individuals and organisations are held liable world. world fully. Think of the It is a different approach to problem solving. in courts of law for system behaviours based recent failure to understand Meadows again: on the presumption that they can ‘predict- and-control’ these complex nonlinear ‘risk modeling’ in nance. “The trick, as with all the behavioural systems. This can result in multi-million possibilities of complex systems, is to euro fines for those accused and found recognize what structures contain which guilty. Think of the litigation surrounding latent behaviors, and what conditions counterfeit items. Author Robert Pirsig captures the that produced that government are left release those behaviors - and, where importance of prevailing paradigms when intact, then those patterns of thought possible, to arrange the structure and Similarly, Information Society Service he states: will repeat themselves... There’s so much conditions to reduce the probability of Providers (ISSPs) may use probabilistic talk about the system. And so little destructive behaviors and to encourage techniques, such as ‘rankings’, to try “If a factory is torn down but the understanding.”15 the possibility of beneficial ones.”13 and help consumers with foresight in a rationality which produced it is left complex retail environment. They are not standing, then that rationality will The longer we cling to a 19th Century In other words, try to coax the system into providing assurances that these are perfect simply produce another factory. If a reductionist view of a clockwork providing the solution - try to ‘evolve’ the predictions. Yet we are seeing lawsuits revolution destroys a government, but Newtonian universe that is ticking along solution, not ‘determine’ it. against ISSPs based on claims to the latter. the systematic patterns of thought a perfectly predictable path, the more

30 31 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Setting out a Methodology to Meet the Challenge

trouble we are going to get into - be that not control the physiological system. in our environmental, economic, political We do not dictate what happens. We and social systems or our system of try to dance with the system. We add globalisation. inputs to the system. We monitor. We make adjustments. But we do not have Disciplines such as physics have already had the final say over what happens. It’s to undergo this change in mindset - albeit a nonlinear way of dealing with the that they began 100 years ago. system. Not a linear one. I can look at It was in Parc Leopold adjacent to the two identical cases. Treat them exactly European Parliament that Einstein and the same and have two different 16 Bohr debated this change in mindset at the results.” famous Solvay Physics Conference of 1927. This approach fits with Strogatz’s framework Playing out at the beginning of the 20th for classifying systems. Recall that he Century and known as the ‘Einstein-Bohr places ‘life’ beyond the ‘frontier’ of current debates’, the arguments centred over understanding in figure 8. whether the universe evolves in an arbitrary Indeed the entire modern ‘scientific method’ or deterministic way. is a constant process of falsification - of ‘trial- Einstein argued that “God does not play and-error’. dice”, Bohr the opposite, on one occassion The social sciences mindset needs to catch answering, “Einstein, stop telling God what to up with the natural sciences mindset. do”. It is a case of C.P. Snow’s ‘Two Cultures’. Today’s prevailing scientific consensus In his book “The Two Cultures and the is that ‘God’ (i.e. nature) does play dice - Scientific Revolution”, Snow contends that FIGURE 10 - A CENTURY AGO PHYSICISTS WERE ALREADY CHANGING THEIR MINDSET perfect prediction à la Laplace is impossible. the breakdown of communication between ‘The prevailing scientific consensus is that perfect prediction turns out to be impossible’ And they act in this knowledge daily. the ‘Two Cultures’ of modern society – the sciences and the humanities – is a major Source: Institut International de Physique de Solvay, Parc Leopold Chris Pollard, Member of the Royal College hindrance to solving the world’s problems. of Veterinary Surgeons, cites the example of As Meadows puts it, our challenge is one of: administering anaesthetics: Note: 17 “With anaesthetic drugs we know what “Linear minds in a nonlinear world” 1) For those interested, the physicist Stephen Hawking expands on the debate at: they can do, what side effects they may have in certain situations, which http://www.hawking.org.uk/index.php/lectures/64 situations they may be most effective in. However, for the most part, the 2) The picture above was taken on the steps of what is today the Lycée Emile Jacqmain in Parc most elusive element is determining Leopold, Brussels exactly how they really work. We do

32 33 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Setting out a Methodology to Meet the Challenge

A NEW INTERVENTION ‘MECHANISM’ training, etc. FIGURE 11 - CYCLE THE “OODA LOOP” ‘A practical intervention mechanism for complex nonlinear systems’ How then do interventionists put this new ... Now that the aircraft is in radar ‘Complex Systems’ mindset into concrete contact the speed, size, and Source: Carré & Strauss expanding on John Boyd’s concept of the OODA Loop action in a system? manoeuvrability of the enemy plane has become available. I Here it is instructive to draw on the work of have some real information to go John Boyd, a military strategist, US Airforce on and unfolding circumstances fighter pilot and Colonel. Boyd helped can take priority over my earlier design and champion the F-16 fighter jet. predictions. Translate 18 these decisions O He formulated the ‘OODA Loop’ to ... The aircraft’s intentions appear B Gather sensory enable fighter pilots to quickly assess and into action T S hostile.“ C E inputs from the adapt to complex and rapidly changing A R V

environments - ones that cannot be (3) DECIDE: Use this new environment E controlled. In which poor decision-making ‘knowledge’ as the basis for

can be a matter of life and death. Today the decisions.

loop is written into US Air Force doctrine.

“... Time for evasive action.

INTERVENTION

Taken at its simplest level, Boyd contends E MINDSET

... I am going to get into the sun R

that to prevail up in the air a fighter D I

pilot needs to be more effective than his so that my opponent cannot get a C N

Use this ‘new’ E Make sense

clear visual on me.” D

opponent at continually cycling through

knowledge as of this four processes: (4) ACT: Translate this into action. the basis for situational (1) OBSERVE: Gather sensory inputs “... Pull back on the joystick and let decisions reality from the environment - ‘monitor’ my aircraft climb.” and ‘intelligence gather’. Back to the process of ‘observation’: “Radio chatter informs me that there is an unidentified aircraft in “Is the other plane reacting to my my airspace.” change of altitude?” assumptions? Intelligence. (2) ORIENT: Make sense of this Then to ‘orient’: Are they congruent with the Indeed in software development it is akin to observational data by creating a ‘Agile Programming’. Agile Programming is “Is he reacting characteristically complex reality that I am mental picture - a ‘model’ - of the a methodology for quickly and effectively - predictably - or perhaps instead observing?” situational reality. cycling through software development. acting like a noncombatant? And so the cycle continues. “....The aircraft is not yet in range It supersedes an earlier linear design approach called the ‘Waterfall Model’. but it would already be useful to Is his plane exhibiting better-than- We find similar decision making loops in expected performance? try to predict the likely identity of other disciplines. For example, the ‘Sense- Importantly the 3 Palchinsky Principles of the pilot, his nationality, level of Do I need to revisit my Model-Plan-Act’ Loop in the field of Artificial ‘Variation’, ‘Survivability’ and ‘Selection’ can

34 35 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Setting out a Methodology to Meet the Challenge

FIGURE 12 (opposite page) - THE SIMPLIFIED ‘INTERVENTION MIX’ IN 2D ‘Interaction can emanate from average people, experts or computers and can be configured in AGENT TYPE different forms, either in the form of individuals or of wisdom of crowds’ Non-Expert Expert

Source: Carré & Strauss - the human decision making components of this diagram are assem- People have ‘common sense’ Experts come into their bled principally from Len Fisher’s explanation in “The Perfect Swarm”19 - in more formal terms, own in the area between ‘sensitivity to context’ - ‘rote rule following’ and computers do not. This is also ‘probabilistic prediction’ the domain of ‘personal choice’ - an area in which a be integrated into the OODA Loop. ‘Intervention Agents’. Then we can decide combination of knowledge ‘who’ should intervene. and initiative is required As we cycle through the loop we can Individual seek out new ideas and try new things Should individuals be the interventionist? (Variation). The ‘wisdom of the crowd’? Providing we cycle through the loop iteratively, then when we do attempt Experts or non-experts? To gure out the value of Groups of experts will something new we can act on a scale where Computer algorithms? something it’s best to take outperform most, if not all, failure is survivable (Survivability). an average of a group’s individual experts A combination of all the above - an And as we act the OODA Loop is designed answers. For choosing the ‘Intervention Mix’? right answer from a small

to seek out feedback and learn from our CONFIGURATION AGENT number of possible mistakes as we go along (Selection). In fact we live with intervention mixes Crowd alternatives majority opinion everyday. Think of when you apply the serves us better. The right In other words the OODA Loop is a practical brakes whilst driving your car. It is your foot ‘trial-and-error’ mechanism for dancing with conditions need to be in place pressing the brake pedal, but it is also the to take advantage of either a complex systems. car’s anti-lock braking system at work - a method Therefore if we wish to optimise computer. It is the combination of these systems intervention in for example, the two agents that brings a fast moving car to consumption system, we need to know how a smooth stop at the traffic lights. A mix of Algorithms to cycle most effectively through the OODA man and machine. Loop. Again these are not abstract issues. People can be replaced by computers when it comes to At this very moment we are trying to EACH OF THESE making certain types of ‘INTERVENTION AGENTS’ & ‘THE determine how far we can automate High SQUARES IS A ‘rule-based decisions’ if the INTERVENTION MIX’ Frequency Trading? Can we leave the POTENTIAL requisite hardware, software software algorithms to their own devices or and data exists and it makes To successfully address this question do we need human intervention to avoid ‘INTERVENTION AGENT’ economic sense to do so we need to examine the strengths and events such as the May2010 Flash Crash? weaknesses of those that have the potential to cycle through the loop, namely the Similarly a debate ensues over whether

36 37 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Setting out a Methodology to Meet the Challenge

PAGE’S THEOREM ON ‘STATE ESTIMATION’ CONDORCET’S JURY THEOREM ON MAJORITY OPINION

collective error = average individual error - prediction diversity “The theorem in its simplest form says that if each member of a group has a better than 50:50 chance of getting the right answer to a question that has just two possible answers, “The calculations are slightly tricky because statisticians use squared values of errors. then the chance of a majority verdict being correct rapidly becomes closer to 100 percent But the message is simple ... the theorem shows that our collective error as a group has as the size of the group increases. Even if each individual has only a 60 percent chance of to be smaller than our average individual error because of the diversity of answers. The being right, the chance of the majority being right goes up to 80 percent for a group of 17 20 crowd performs better than most of the people in it.” and to 90 percent for a group of forty-five.”22

Len Fisher, “The Perfect Swarm” Len Fisher, “The Perfect Swarm”

ISSPs can completely automate counterfeit problems that involve choosing the right (1) STATE ESTIMATION: “When then a rigorous mathematical items off of their platforms? answer from among a small number of answering a state estimation formulation [Condorcet’s Jury possible alternatives, a group’s majority question, the group as a whole Theorem] proves that the answer As we will see, no intervention agent is the opinion serves us better. To take the best will always outperform most of its of the majority has a better than panacea - expert, ‘wisdom of crowds’ or advantage of either, we need to fulfil just three individual members.”23 University 99 percent chance of being the computer algorithm. Each has its strengths conditions: of Michigan complexity theorist, correct one. and weaknesses depending on the specific Scott Page, proves this with his circumstance. For example, we are all too (1) DIVERSITY: “The people in the theorem that demonstrates that (3) VALIDITY: “Even when only a familiar with the ‘madness of crowds’ and group must be willing and able “our collective error as a group has few people in the group are well- with ’mob mentality’ in certain situations. to think for themselves and reach to be smaller than our average informed, this is usually sufficient diverse, independent conclusions. individual error because of the for the majority opinion to be the 26 24 right one.” (2) VERIFICATION: “The question diversity of our answers”. As THE ‘WISDOM OF CROWDS’ must have a definite answer that Page puts it, “being different is as 25 can ultimately be checked against important as being good.” What follows on the ‘wisdom of crowds’ ‘EXPERTS’ AND ‘GROUPS OF EXPERTS’ versus that of ‘experts’ is in large part reality. (2) MAJORITY OPINION: “If drawn from the explanation provided by (3) ALIGNMENT: “Everyone in the most of the group members Fisher cautions that, “There is just one caveat, Len Fisher, a Physicist at the University of group must be answering the are moderately well-informed which is to remember that Page’s theorem Bristol, in his book “The Perfect Swarm - The same question. (This may seem about the facts surrounding proves only that the group outperforms 21 Science of Complexity in Everyday Life”. obvious, but it is often possible a question to which there are ‘most’ of its members when it comes to state several possible answers (but only estimation problems. It does not necessarily “For problems that involve figuring out the for people to interpret the same question in very different ways.) one correct one), the majority outperform ‘all’ of them. If there is an value of something (like a compass bearing opinion is almost always bound identifiable expert in the group, it may be that or the classic case of the number of beans in “When these 3 conditions are fulfilled, the to be right. If each member of a they will do better than the group average ... a jar) the best way to tackle the problem is mathematics of complexity leads to three group of one hundred people has That’s not to say that experts always do better to [employ group decision making and to] astounding conclusions: a 60 percent chance of getting than the average.” take an average of all the answers. Scientists the right answer, for example, call these ‘state estimation’ problems. For

38 39 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Setting out a Methodology to Meet the Challenge

FIGURE 13 (opposite page) - THE RISE OF THE MACHINES ‘Algorithmic trading has rapidly taken off’

Source: Aite Group diagrams reproduced by The Economist, February 25th 2012

“A collection of experts can also outperform Back to Melanie Mitchell who notes that: most, if not, all of the individual experts. Page gives the example of a group of football “Computer controlled vehicles can journalists forecasting the top dozen picks now drive by themselves across in the 2005 NHL draft. Not one of them rugged desert terrain. Computer performed nearly as well as the average of all programs can beat human doctors at of them ... diagnosing certain diseases, human mathematicians at solving complex “According to Michael Mauboussin, a equations, and human grand masters Professor of Finance at Columbia Business at chess. These are only a few examples School, experts come into their own in of a surge of recent successes in artificial the area between rote rule following and intelligence (AI) that have brought a probabilistic prediction - an area in which a new sense of optimism in the field.”31 combination of knowledge and initiative is required.”27 But she also draws attention to the fact that formally in computer science terms, hard to replicate in computers - because, despite this optimism we need to remain ‘sensitivity to context’.33 in contrast with theoretical knowledge, That is why we continue to have experts grounded and not get ahead of ourselves: it requires a relatively large number of in policymaking, experts in software Duncan Watts, Principal Research Scientist rules to deal with even a small number development, experts in the enforcement of “There are a few ‘human-level’ things at Yahoo, provides a definition of ‘common of special cases.”35 trademark rights etc. computers still can’t do, such as sense’ in his book “Everything Is Obvious* - understand human language, describe *Once You Know The Answer”: Mitchell again: “When we don’t have an expert available, we the content of a photograph, and more must fall back on the diversity of the group.”28 generally use common sense... Marvin “Roughly speaking, it is the loosely “My computer supposedly has a state- Minsky, a founder in the field of artificial organized set of facts, observations, of-the-art spam filter, but sometimes intelligence, concisely described this experiences, insights, and pieces it can’t figure out that a message with paradox of AI as, ‘Easy things are hard.’ of received wisdom that each of us a ‘word’ such as V!a&®@ is likely to be COMPUTERS VERSUS ‘COMMON SENSE’ 36 Computers can do many things that accumulates over a lifetime, in the spam.” course of encountering, dealing Fisher observes that, “Experts are also we humans consider to require high with, and learning from, everyday Most people however see V!a&®@ for what being replaced in many areas by computer level intelligence, but at the same time situations.”34 it is instantly. That is why most spam filters algorithms when it comes to making ‘rule- they are unable to perform tasks that draw extensively on pooling information based’ decisions”.29 An algorithm is a any three-year-old child could do with He states that: feedback from real users. If 1,000 people 32 series of steps that transforms an input ease.” click on the ‘This is Spam’ button when they into an output - in other words, a ‘definite “It is largely for this reason, in fact, that encounter V!a&®@ then the computer can procedure’ or a ‘recipe’.30 We frequently experience the fact that commonsense knowledge has proven so computers lack ‘common sense’, or more catalogue it as likely to be spam.

40 41 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Setting out a Methodology to Meet the Challenge

FIGURE 14 (below) - THE LIMITS OF COMPUTATION ‘Computers are unable to perform some tasks that humans find trivial’ THE ‘NP-COMPLETE’ PROBLEM

Source: Drupal.org “If it is easy to check that a solution to a problem is correct, is it also easy to solve the problem? This is the essence of the P vs NP question. Typical of the NP problems is that of the Hamiltonian Path Problem: given N cities to visit, how can one do this without visiting a city twice? If you give me a solution, I can easily check that it is correct. But I cannot so This lack of sensitivity to context is also the It is for this reason, as well as for legal, easily find a solution.”38 reason why web users are on occasion asked ethical and moral considerations, that to retype a code - see Figure 14 below. military drones are remote controlled and Clay Mathematics Institute This code cannot be read by a computer are not fully autonomous. At least for now because it lacks the common sense - the humans still make the decision to fire. intuition - to tell that it says MY5N5. The aim in this ‘CAPTCHA’ code example is to Johann Borenstein, head of the Mobile prevent automated programmes (i.e. spam Robotics Lab at the University of Michigan observes that: decide its truth or falsity. He (2) HARDWARE AND SOFTWARE in this instance) from requesting access demonstrated this fact via the LIMITATIONS: Theoretical to some piece of information - it is just for “Robots don’t have common sense and ‘Halting problem’. In short, there limitations aside, there are human consumption - those “agents” that won’t have common sense in the next 50 are certain classes of problem hardware and software can discern the text. years, or however long one might want that cannot be computed.39 limitations to consider. Yes, to guess.”37 Moreover even when problems computer processing power can be computed it may be is increasing exponentially Consequently he believes that human skills very time consuming to do and becoming less and will remain critical in battle far into the so. For instance, to date no less expensive (a process future. efficient algorithm is known for encapsulated by Moore’s Law). an important class of problems However as virtual reality And there are additional limits to known as ‘NP-complete’. These pioneer Jaron Lanier notes in his computing power, namely: search problems include book “You Are Not A Gadget”, (1) THEORETICAL LIMITATIONS: In scheduling, map colouring, “software development doesn’t the 1930s Kurt Gödel and Alan protein folding, theorem necessarily speed up in sync with Turing quashed the hope of the proving, packing, puzzles, the improvements in hardware. It unlimited power of mathematics travelling salesman and many often instead slows down as and computing. Gödel proved more such problems. In fact computers get bigger because that mathematics is either the Clay Mathematics Institute there are more opportunities inconsistent or incomplete. classifies the NP-complete for errors in bigger programs. Building on this discovery Turing problem as one of the seven Development becomes slower and proved that the answer to the most important questions to more conservative when there is ‘Entscheidungsproblem’ (‘decision be answered this century and is more at stake, and that is what is problem’) is ‘no’ - i.e. not every offering $1,000,000 dollars to the happening.”40 mathematical statement has first person that does so. a definite procedure that can

42 43 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Setting out a Methodology to Meet the Challenge

(3) DATA LIMITATIONS: Eric As a result, leading artificial intelligence FIGURE 15 - THE FULL ‘INTERVENTION MIX’ IN 3D Schmidt, Executive Chairman of researchers, such as Professor Hans Moravec ‘As well as there being different intervention types and configurations, intervention can also Google, points out that, “From from the Robotics Institute at Carnegie originate from the private and public sectors’ the dawn of civilization until Mellon University, acknowledge that 2003, humankind generated “Computers have far to go to match human Source: Carré & Strauss five exabytes of data. Now we strengths”.44 produce five exabytes every two days... and the pace is Lanier recalls that “Before the crash, bankers accelerating”.41 As a result believed in supposedly intelligent algorithms AGENT AGENT there is a lot excitement that could calculate credit risks before making 45 CONFIGURATION surrounding the promise of bad loans.” TYPE what is being termed, ‘Big Expert Individual Indeed Lanier is careful to draw the Non-Expert Crowd Data’ - the ability to cycle the distinction between ideal and real Algorithms OODA loop in unprecedented computers.46 ways in almost every walk Private of life due to advances in Nevertheless this is not to say that “common Sector technology. Nevertheless there sense” approaches do not have their own are significant challenges to shortcomings. Public be addressed, not least those Sector pertaining to privacy. Simon Watts identifies three limitations in common ORIGIN AGENT Szykman, Chief Information sense reasoning:47 Officer of the US Commerce (1) INDIVIDUAL BEHAVIOUR: Our EACH OF THESE Department, has outlined his mental models systematically CUBES IS A POTENTIAL top nine Big Data challenges.42 fall short of capturing the These are: data acquisition; “INTERVENTION AGENT” complexity of what drives storage; processing; data individual behaviour. transport and dissemination; data management and curation; (2) COLLECTIVE BEHAVIOUR: Our 48 archiving; security; workforce mental models of collective Watts concludes that: PUBLIC VERSUS PRIVATE SECTOR with specialized skills, and; the INTERVENTION behaviour fail to take account of “Commonsense reasoning, therefore, cost of all of the above. the fact that collective behaviour does not suffer from a single We have covered the strengths and is greater than the sum of its (4) ECONOMIC LIMITATIONS: overriding limitation but rather from weaknesses of the various ‘Intervention parts. Sometimes it simply makes a combination of limitations, all of Agents’. We now need to introduce the third no financial sense to try to (3) LEARNING FROM HISTORY: which reinforce and even disguise one dimension to the ‘Intervention Mix’, namely develop complex software We learn less from the past another. The net result is that common the division between public and private algorithms to solve a problem than we think we do and this sense is wonderful at making sense sector intervention. that a human can solve instantly of the world, but not necessarily at misperception in turn skews our For both sectors can deploy all of these and effortlessly. Lanier notes understanding it.” perception of the future. agents. Think of general elections and that until recently, computers referendums (‘wisdom of crowds’), couldn’t even recognize a person’s smile.43

44 45 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Setting out a Methodology to Meet the Challenge

expert committees and groups of ‘wise A generic debate over the merits of more FIGURE 16 - SMARTER INTERVENTION men’ (‘groups of experts’), as well as or less government intervention - often in ‘A new intervention mindset, mechanism & principle - in short a new paradigm for intervention’ representatives elected to political office the form of regulations, subsidies or fiscal (often ‘non-experts’) and online tax forms or incentives - will serve no purpose. Source: Carré & Strauss speed cameras (‘computer algorithms’). There is no ‘off-the-shelf’ answer. Taken to its logical conclusion we can and INTERVENTION should extend this approach to include So let us “try” to answer this question when PRINCIPLE other sectors, such as civil society (i.e. NGOs, we seek to intervene in a specific system: Trade Unions, Think Tanks, Academia, Media, “How do we use intervention agents to STOP? Citizens, etc.). However for the sake of time cycle most effectively through the OODA and simplicity we will stick here to just two Loop?” sectors.

So can we delineate between whether we INTERVENTION need: A NEW “INTERVENTION PRINCIPLE” O MECHANISM (A) Public sector intervention And when answering this question let us B ‘try’ to invoke a new ‘Intervention Principle’ T S (B) Private sector intervention C E to guide us: A R (C) A mix of public and private V “An intervention agent is to intervene sector intervention only if, and in so far as, it is reasonably E

foreseeable that the objectives of the To even begin to make this call we need to

proposed intervention cannot better be

return to the OODA Loop and ask:

achieved by the system running itself or

“Knowing the strengths and weaknesses in default of this by another agent”

of the intervention agents that we have INTERVENTION

at our disposal - the various types and The objective should be as little E MINDSET

configurations ... intervention as possible. Referring back to D I

Donella Meadows: I E

And aware that we can deploy either E

T “Aid and encourage the forces and D

public or private sector intervention or

both ... structures that help the system run itself. Notice how many of those forces How do we use intervention agents to and structures are at the bottom of cycle most effectively through the OODA the hierarchy. Don’t be an unthinking Loop?” intervenor and destroy the system’s own self-maintenance capabilities. Before And we need to frame this question in the you charge in to make things better, pay context of a specific issue. attention to the value of what’s already there.“50 ‘SMARTER INTERVENTION’

46 47 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS Setting out a Methodology to Meet the Challenge

As she points out: (1) TEMPO: Can the Intervention ‘automation bias’.54 She defines Agent cycle through the OODA this as the tendency to trust “Any system, biological, economic Loop at the right tempo? an automated system, in spite or social that gets so encrusted that Intervening too fast or too slow of evidence that the system it cannot self-evolve, a system that can be detrimental, leading is unreliable, or wrong in a systematically scorns experimentation to oscillations in the system’s particular case. and wipes out the raw material of behaviour. To paraphrase the innovation, is doomed over the long words of Donella Meadows, 51 term on this highly variable planet.” intervention needs to fit with 53 For: “the beat of the system”.

“The ability to self-organize is the (2) ACCOUNTABILITY: What will strongest form of system resilience. A ensure that the Intervention system that can evolve can survive Agent can be held accountable almost any change, by changing itself.”52 as it cycles through the OODA Loop? Will it always be possible Note the Intervention Principle’s holistic to hold Agents accountable approach to intervention. It applies to any in systems whose behaviours ‘Intervention Agent’, regardless of its type, are greater than the sum of configuration or the sector from which it their parts? This has significant originates. implications for our current liability regimes. Also observe that the Intervention Agent has to be able to make a ‘Reasonable’ case (3) CONSISTENCY: How do we for why they are intervening. Why are they avoid ‘silo approaches’ to not leaving the system to self-organise intervention whereby one - to run itself? And if intervention must Intervention Agent’s action take place why not defer to the actions of undermines those of another? another Intervention Agent? Instead all of these actions need to reinforce one another. They The principle employs the word need to act in concert with each ‘Foreseeable’ because we have some other - to ‘interface’ successfully pointers to help us to intervene - we with one another. In other have the OODA Loop and we know the words, the various Intervention strengths and weaknesses of the various Agents need to cycle around the Intervention Agents, their different types loop as a team. and configurations - but nothing is certain in a complex nonlinear systems world. (4) BIAS: Mary Cummings, Director of the MIT Humans Finally, when we do intervene we need to and Automation Lab, draws our take into consideration: attention to the phenomenon of

48 49 COPYRIGHT ACKNOWLEDGEMENTS PRINTING & PRINTED COPIES SOURCES & FURTHER READING OTHER IDEAS TO WATCH See creativecommons.org Commons attribution 3.0 unported licence. This chart is issued under a Creative Thanks to Charlie @ Plum Creative [email protected] ready printed copies can be obtained from: High resolution digital files for this table and and www.nowandnext.com See www.futuretrendsbook.com 25. 24. 23. 22. 21. 20. 19. 18. 17. 16. 15. 14. 13. 12. 11. 10. 9. 8. 7. 6. 5. 4. 3. 2. 1. Voluntary simplicity Value redefinition Ultra-ecient solar Shift from products to experiences Slow education Smart infrastructure Self-tracking Reverse migration Quantum computing Peer-to-peer lending/giving Peak water Mobile money Local living Increasing complexity Holographic telepresence Gene hacking Facial recognition on mobile phones Electrification of transport Decline of voice communication Diminishing use of email Contextual deficit Comfort eating Clean coal Biomimicry Avatar assistants Creative Commons LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS by Richard Watson Need to Know 50 Ideas You Really The Future: TABLE OF TRENDS & TECHNOLOGIES FOR THE WORLD IN 2020 domestic policy E-government connectivity Xenophobia governance resurgence Regulatory Idealogical Erosion of Gg Volatility regional & Re Focus on Eg Xe change Global Er Fr failure Hyper trust H Ir Conclusion V nationalism Micro-grids conventional Clean fuels generation Scarcity of Biological resources Resource terrorism terrorism Mg volatility & micro- Ne Rn Nuclear Pv Nt reserves Bt Cf Sr Price Non Omg smart objects incrementalism disintegration Resurgence Robotics & Secularism of religion European materials KEY European No Ed Rb Na Eu E S involve varying configurations of experts, I Nano dentity ociety “The future is already here - it’s just not non-experts, crowds and individuals. This conomy personalisation Unsustainable Personalised

very evenly distributed.” in turn means correctly employing all of Hedonism Synthetic Op medicine Md poulation Up Predictive Oil price He Pd Sb biology growth spikes High probability Global risk T these intervention agents - playing to E echnology William Gibson their strengths, whilst mitigating their mployment protectionism Nationalism & Mm immediacy computing Culture of genomics migration managed Semantic Np Sw Personal Context Pg Ca Poorly Ci aware aware weaknesses - ensuring the right tempo, web Low probability Global risk E accountability and consistency, as well as P nergy opulation Holographics transparency Autonomous Explosion of North Korea Sws information systems & shortages & 3D web Personal Hg Cd As devices Kr worker clouds

avoiding automation bias. Invoking these Skilled Ti Total agents when necessary, but leaving the E P Uncategorised Augmented

system be when it is not. nvironment Food price

‘FAST FORWARD TO THE FUTURE’ olitics Internet of portability Csf Mobility & Real-time Mo analytics volatility Fp & virtual systems Au Ra Critical worlds data & failure things It When one looks at many of the different And increasingly many of these innovations wellfare state Automation recognition imbalances Collapse of Artiﬁcially intelligent Ws MEGATRENDS Gestural Ge Globalisation Sharing devices are going to challenge our preconceptions deregulation predictions that are being made about the At Ai Fi Fiscal S Gd

future - be they right or wrong - ‘Smarter of what constitutes: & communication Uub information and energy Collapse of Battery life Intervention in Complex Systems’ will Ubiquitous Nfc Urbanisation Too much sensors & pandemic Tm Near-ﬁeld Pakistan Gp tracking Bq Pk storage Um migration (1) COMPUTATION: How far are we Global

become an ever greater imperative. & going to be able to programme Cyber viruses Dematerialisation Desert based sustainability Provenance technology Dm Cw India war change & and data Db Pakistan our environment? After all Ha Climate

These are but a few of the exciting advances Cs Haptic Pr Pi solar theft that are likely to be grounded in complex ‘computation’ is not only done Vs convergence Population food & water re-charging & lifespan by silicon based computers. Technology nonlinear systems: augmented reality 3D printing Collapse of Search for happiness Lol growth Wr access to Wireless Cn Ua D Uneven Tc P and virtual worlds, gestural recognition, It is widespread in the natural China 3

holographics and the 3D web, 3D printing, world. For instance, we find re-regulation Open-source Gamiﬁcation Localism treat obesity connectivity discovery & Ubiquitous Lr inequality networks invention Failure to epidemic Oa Ga Os Sn income Severe Social & Si nano materials, real time data and analytics, it taking place in cells, tissues, U robotics and smart objects, the semantic plants and the immune system. climate change Age Urban living Population Population employee Failure to Extreme

Consider this statement by the Rc weather Cc Po web, synthetic biology, ubiquitous sensors Fertility adapt to Fe Ex decline growth ageing Rogue events and trackers, and quantum computing. As US geneticist and ‘biohacker’, Li

Craig Venter, in May 2010: Sure-volcano Paw Fragmented bio-diversity middle class acidiﬁcation a result the ability to cycle the ‘Observe- Mega-quake Sph family units households Expansion mega-city Vol Changing Mq attention of global Mc eruption Ac person Loss of Ocean Single B F

Orient-Decide-Act’ loop effectively will “We’re here today to announce in be a critical success factor. Doing so will the first synthetic cell, a cell manufacturing Ethnic shifts Resurgence Am Low probability High probability reduction Aquifer Carbon Ar Es Rise of pricing Africa of US A C Individualism Shift to the agriculture capitalism Precision Me Ag creation Top soil erosion Sv Sc Shared Te Br BRICs value State FIGURE 17 (opposite page) - TABLE OF TRENDS & TECHNOLOGIES FOR THE WORLD IN 2020 Chart maker: Richard Watson Normalisation enhancement real & virtual

‘One futurist’s recommended set of ideas to watch’ Uuh entitlement Blurring of of obesity Focus on Cosmetic Sense of Ce Se the self worlds N I Source: Richard Watson1 consolidation Atm competition outsourcing Intensifying Atomisation Workforce homing of Pm part time Mobile & meaning Purpose working Bh Industry ageing W Pt In Ic Back &

50 51 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS

ones instilled with empathy and compassion? As military forces ‘OTHER IDEAS TO WATCH’ around the world increasingly “Avatar assistants; Biomimicry; Clean coal; Comfort eating; Contextual deficit; deploy drones they are having Diminishing use of email; Decline of voice communication; Electrification of transport; to grapple with this issue. Can Facial recognition on mobile phones; Gene hacking; Holographic telepresence; Increasing or should armed robots be given complexity; Local living; Mobile money; Peak water; Peer-to-peer lending/giving; fully autonomous capabilities Quantum computing; Reverse migration; Self-tracking; Smart infrastructure; Slow in combat? Various approaches education; Shift from products to experiences; Ultra-efficient solar; Value redefinition; have been proposed, such Voluntary simplicity.”2 as Ronald Arkin’s ‘Ethical Governor’.5 Arkin is a roboticist Richard Watson at the Georgia Institute of Technology and he suggests a 2-step decision procedure before pulling the trigger- i.e. an algorithm. However a report made by starting with the digital new software for the cell. He coauthored by Human Rights code in the computer, building believes that “If we can really get Watch and Harvard Law School the chromosome from four cells to do the production that contends that, “fully autonomous bottles of chemicals, assembling we want, they could help wean weapons would be incapable that chromosome in yeast, us off oil and reverse some of the of meeting international transplanting it into a recipient damage to the environment by humanitarian law standards, bacterial cell and transforming capturing carbon dioxide.”4 including the rules of distinction, that cell into a new bacterial proportionality, and military (2) MORALITY: To what extent, species. So this is the first self- necessity. These robots would are we going to be able encode replicating species that we’ve had lack human qualities, such as the morality into our algorithms? on the planet whose parent is a ability to relate to other humans Or are we still going to need to computer. It also is the first species and to apply human judgement mix human intervention into to have its own website encoded that are necessary to comply in its genetic code.”3 Venter has the OODA loop to guarantee likened the advance to making ethical decisions and actions -

FIGURE 18 (opposite page) - CRAIG VENTER’S SYNTHETIC CELL ‘The synthetic cell looks identical to the ‘wild type’ and extends our concept of ‘computation’’

Source: BBC Online News sourcing the picture from Science, 20 May 2010

FIGURE 19 (above) - ‘MORALS AND THE MACHINE’ ‘Do humans need to remain in the OODA Loop to ensure ethical decision making?’

Source: The Economist, June 2nd-8th 2012

52 53 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS

be without it. I can hold the FIGURES 20 and 21 - ‘THE INCREASING CONVERGENCE OF MAN AND MACHINE’ “We are made wise not by the recollection phone, shake hands and wash ‘Nigel Ackland using his ‘bebionic3’ prosthetic hand made by RSLSteeper (below). Ray Kurweil’s of our past, but by the responsibility of our my left hand normally, which controversial vision of ‘The Singularity’ (bottom)’ future.” I haven’t been able to for five years!”8 Ray Kurzweil, Director Sources: YouTube video of Nigel Ackland, RSLSteeper technical information manual for bebionic3, George Bernard Shaw of Engineering at Google, and “The Singularity” by Ray Kurzweil argues that this convergence between man and machine is passing through six epochs culminating in a “Technological with the law.”6 As a result, Steve Singularity” - a moment when Goose, Director, Arms Division, “humans transcend biology”.9 Human Rights Watch, argues Critics such Stephen Pinker, that, “... you must have meaningful Psychology Professor at Harvard human control over key battlefield University, counter that “There is decisions of who lives and who not the slightest reason to believe dies. That should not be left up in a coming singularity. The fact to the weapons system itself”.7 In that you can visualize a future in other words, when it comes to your imagination is not evidence the question of armed drones, that it is likely or even possible. then humans need to remain in Look at domed cities, jet-pack the OODA loop. An ‘Intervention commuting, underwater cities, Mix’ is required. mile-high buildings, and nuclear- powered automobiles - all staples (3) HUMANITY: Rapid advances of futuristic fantasies when I was in technologies, such as a child that have never arrived. prosthetics, are increasingly Sheer processing power is not a blurring the line between pixie dust that magically solves all man and machine. Take the your problems.“10 example of Nigel Ackland and his “bebionic3” prosthetic hand. So are we are entering a brave new world or It operates by responding to one full of exciting technological promise? Nigel’s muscle twitches. He configures its functionality Much will depend on how successful we - grips, thresholds and radio are in pursuing “Smarter Intervention in frequency - by using software Complex Systems”. that runs on Microsoft Windows. Adopting a new intervention mindset, Nigel states that, “Having a mechanism and principle will improve bebionic hand is like being human our chances of building and securing a again, psychologically I wouldn’t sustainable future.

54 55 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS

Endnotes

INTRODUCTION BY HANNE MELIN Pages/04Smart.pdf Places to Intervene in a System”, pp. 4-5 . SMARTER INTERVENTION IN COMPLEX See: http://www.sustainer.org/pubs/Lever- SYSTEMS 1 Richards, Neil M. and Smart, William, 7 Speech by Neelie Kroes, 8 May 2012, Brus- age_Points.pdf “How should the law think about robots?” sels. Available at: http://europa.eu/rapid/ 1 Donella Meadows, “Thinking in Systems”, (May 10, 2013). Available at: http://ssrn. press-release_SPEECH-12-334_en.htm p. 91 com/abstract=2263363 or http://dx.doi. EXAMPLE OF THE PRODUCT AWARENESS org/10.2139/ssrn.2263363 SYSTEM 2 Ibid., p. 92 A MANIFESTO FOR CHANGE 2 Mandelkern Group on Better Regulation, 1 John Sterman, “Business Dynamics”, pp. 3 Melanie Mitchell, “Complexity”, pp. 27-33 13 November 2001. Available at http:// 1 Pankaj Ghemawat, “World 3.0”, p. 3 365-366 ec.europa.eu/governance/better_regula- 4 Ubiquity, “An Interview with Melanie tion/documents/mandelkern_report.pdf Mitchell on Complexity”. See: http://ubiq- THE GLOBAL SYSTEM DEFINED LEVERAGE POINTS IN SYSTEMS uity.acm.org/article.cfm?id=1967047 3 Hans-Bredow-Institut for Media Research at the University of Hamburg, Study on 1 Donella Meadows, Jorgen Randers, Dennis 1 Donella Meadows, “Thinking in Systems”, p. 5 Steven Strogatz, “Nonlinear Dynamics and co-regulation measures in the media sec- Meadows, “The Limits of Growth”, p. 53 148 Chaos”, pp. 8-9 tor, June 2006. Available at http://ec.europa. eu/avpolicy/docs/library/studies/coregul/ 2 Donella Meadows, “Leverage Points - Plac- 6 Donella Meadows, “Thinking in Systems”, final_rep_en.pdf SYSTEMS DYNAMICS INTRODUCED es to Intervene in a System”, p. 3 p. 12

4 “The European Parliament 2025: Preparing 1 Donella Meadows, “Thinking in Systems” 3 Ibid., p. 18 7 Tim Harford, “Adapt - Why Success Always for Complexity, Brussels, European Parlia- Starts with Failure”, p. 25 ment, January 2012” (PE479.851/BUR). Avail- 2 David Cliff, University of Bristol, “The Fu- 4 Donella Meadows, “Thinking in Systems”, able at http://www.europarl.europa.eu/pdf/ ture of Computer Trading in the Financial pp. 146-147 8 Donella Meadows, “Thinking in Systems”, SG/documents/EP2025-Preparing_for_Com- Markets: Working Paper”. See the abstract: p. 87 plexity_EN_FINAL.PDF http://www.bris.ac.uk/engineering/people/ 5 Ibid., p. 146 dave-t-cliff/pub/2827197 9 Melanie Mitchell, “Complexity”, p. 222 5 Blog of Neelie Kroes, Vice-President of the 6 Jason Pontin, MIT Technology Review, European Commission, 25 April 2012: http:// 3 Donella Meadows, “Thinking in Systems”, p. “Why We Can’t Solve Big Problems”. See: 10 Donella Meadows, “Thinking in Systems”, blogs.ec.europa.eu/neelie-kroes/dg-con- 152 http://www.technologyreview.com/fea- pp. 169-170 nect-launch/ turedstory/429690/why-we-cant-solve-big- 4 Donella Meadows, “Thinking in Systems”, p. problems/ 11 Melanie Mitchell, “Complexity”, pp. 31-33 6 Interview with Chan Lai Fung. Available 175 at: http://www.cscollege.gov.sg/Knowl- 12 Steven Strogatz, “Nonlinear Dynamics and edge/Ethos/Issue%201%20Apr%202006/ 5 Donella Meadows, “Leverage Points - Chaos”, pp. 11

56 57 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS

13 Donella Meadows, “Thinking in Systems”, 28 Ibid. 42 Mark Hoover, Washington Technology, p. 72 May 01, 2013, ”Do you know big data’s top 51 Ibid., p. 160 29 Ibid. 9 challenges?“. See: http://washingtontech- 14 52 See endnote 8 30 Melanie Mitchell, “Complexity”, p. 129 nology.com/articles/2013/02/28/big-data- Ibid., p. 159 challenges.aspx 15 53 Donella Meadows, “Thinking in Systems”. 31 Ibid., p. 187 Ibid., p. 170-172 Meadows opens her book with this quote 43 Jaron Lanier, “You Are Not A Gadget”, p. 54 from Robert Pirsig 32 Ibid. 159 Mary Cummings idea of “Automation Bias” is examined by Peter Asaro in “Modeling the 16 44 Discussion between the author and Chris 33 Ibid., p. 186 Hans Moravec, Journal of Evolution and Moral User”, IEEE Technology and Society Pollard, Member of the Royal College of Technology, “When will computer hardware Magazine, p. 22 Veterinary Surgeons 34 Duncan J. Watts, “Everything Is Obvious* match the human brain?” See: ftp://io.usp. br/los/IOF257/moravec.pdf 17 Donella Meadows, “Thinking in Systems”, - * Once You Know The Answer - How Com- mon Sense Fails”, p. 8 CONCLUSION - “FAST FORWARD TO THE p. 91 45 Jaron Lanier, “You Are Not A Gadget”, p. 32 FUTURE”

35 18 Ibid., p. 10 Fast Company, “The Strategy of the Fight- 46 Ibid., p 7 1 Richard Watson, “Table Of Trends & Tech- er Pilot”. See: http://www.fastcompany. 36 Melanie Mitchell, “Complexity”, p. 186 nologies For The World In 2020”. See: http:// com/44983/strategy-fighter-pilot 47 Duncan J. Watts, “Everything Is Obvious* www.nowandnext.com/PDF/Trends%20

37 - * Once You Know The Answer - How Com- &%20Technologies%20for%20the%20 19 Peter Finn quoting Johann Borenstein Len Fisher, “The Perfect Swarm - The Sci- mon Sense Fails”, pp. 25-29 World%20in%202020.pdf ence of Complexity in Everyday Life”, drawn in the Washington Post, September 19, 2011, “A future for drones: Automated kill- from Chapter 5 48 2 ing”. See: http://articles.washingtonpost. Ibid., pp. 27-28 Ibid. 20 com/2011-09-19/national/35273383_1_ Ibid., pp. 73-74 49 3 drones-human-target-military-base/2 Article 5.3 of the Treaty on European TED Transcript of “Craig Venter unveils ‘syn- Union was the source of inspiration for me thetic life’”. See: http://www.ted.com/talks/ 21 See endnote 19 38 Clay Mathematics Institute. See: http:// to propose a new “Intervention Principle”. craig_venter_unveils_synthetic_life.html This article states that: “Under the principle Also see: http://www.sciencemag.org/con- 22 www.claymath.org/ Len Fisher, “The Perfect Swarm - The Sci- of subsidiarity, in areas which do not fall tent/329/5987/52.full ence of Complexity in Everyday Life”, p. 76 39 Melanie Mitchell, “Complexity”, see Chap- within its exclusive competence, the Union shall act only if and in so far as the objectives 4 BBC Online News quoting Craig Venter, 23 ter 4 Ibid., p. 68-69 of the proposed action cannot be sufficiently “’Artificial life’ breakthrough announced

40 achieved by the Member States, either at cen- by scientists”. See: http://www.bbc.co.uk/ 24 Ibid., p. 73 Jaron Lanier, “You Are Not A Gadget”, p. 181 tral level or at regional and local level, but can news/10132762 rather, by reason of the scale or effects of the 25 Ibid., p 74 5 41 Rick Smolan and Jennifer Erwitt, “The Hu- proposed action, be better achieved at Union Arkin’s ‘Ethical Governor’ is examined by level.” Peter Asaro in “Modeling the Moral User”, 26 Ibid., p 68-69 man Face of Big Data”, Eric Schmidt being quoted on the page before the foreword IEEE Technology and Society Magazine, pp. 50 Donella Meadows, “Thinking in Systems”, 21-22 27 Ibid., pp. 74-75 p. 178

58 59 LIGHTS | CAMERA | ACTION ! MANIFESTO FOR SMARTER INTERVENTION IN COMPLEX SYSTEMS

6 Summary of Losing Humanity - “The Case Against Killer Robots”, Report by Human Rights Watch together with the Interna- tional Human Rights Clinic at Harvard Law School. See: http://www.hrw.org/sites/default/files/reports/arms1112_ForUpload.pdf

7 Daily Mail article quoting Steve Goose in “Top Gun in the 21st century: New U.S. stealth drones launched from aircraft carrier by REMOTE CONTROL”. See: http://www. dailymail.co.uk/news/article-2324571/U-S- Navys-X-47B-stealth-drone-launches-aircraft-carrier-time--critics-warn-heralds-rise- killer-robots.html

8 RSLSteeper’s “bebionic3 - Patient Stories”. See: http://bebionic.com/the_hand/pa- tient_stories/nigel_ackland

9 Ray Kurzweil, “The Singularity is Near”, p. 15

10 “Tech Luminaries Address Singularity”, IEEE Spectrum. See: http://spectrum.ieee. org/computing/hardware/tech-luminaries- address-singularity

60 61 www.carre-strauss.com