University of South Florida Scholar Commons

Anthropology Faculty Publications Anthropology

1-2011 Anthropologist View on Social Network Analysis and Data Mining Alvin W. Wolfe University of South Florida, [email protected]

Follow this and additional works at: https://scholarcommons.usf.edu/ant_facpub Part of the Anthropology Commons

Scholar Commons Citation Wolfe, Alvin W., "Anthropologist View on Social Network Analysis and Data Mining" (2011). Anthropology Faculty Publications. 8. https://scholarcommons.usf.edu/ant_facpub/8

This Article is brought to you for free and open access by the Anthropology at Scholar Commons. It has been accepted for inclusion in Anthropology Faculty Publications by an authorized administrator of Scholar Commons. For more information, please contact [email protected]. SOCNET (2011) 1:3–19 DOI 10.1007/s13278-010-0014-4

ORIGINAL ARTICLE

Anthropologist view of social network analysis and data mining

Alvin W. Wolfe

Received: 14 July 2010 / Accepted: 2 August 2010 / Published online: 6 October 2010 Springer-Verlag 2010

Abstract An anthropologist shares with the ‘‘SNA and systems. He concludes with hope that as improved methods data mining community’’ his own anthropological per- of data mining and network analysis are developed, other spective framed during more than five decades of network anthropologists and social scientists will be able to measure thinking about a broad range of anthropological problems. the evolution of supranational sociocultural systems that For 50 years he has viewed all people, things, and ideas in involve both states and multinational corporations. dynamic relationships. That perspective is a network per- spective and at the same time anthropological, combining ethnographic, historical, holistic, and comparative views. It 1 Introduction is valuable and beneficial to the community of scholars who use network analysis to try to understand what is When editor Reda Alhajj suggested that I prepare an article going on, what went on before, and what the future pros- for this first issue of SNAM journal, I thought first about an pects are. As an anthropologist, his interest is more in the academic history of social network analysis, something like wholes generated by network linkages—systems of I attempted to do for anthropology in the first issue of households, bands, lineages, communities, corporations, Social Networks (‘‘The Rise of Network Thinking in governments—than in the individual persons linked. Even Anthropology,’’ 1978) or as Jeff Johnson did, also for now, when personal network ‘‘communities’’ are getting so anthropology, in Advances in Social Network Analysis in much attention network analysis can clarify the more 1994, or something along the lines that Linton Freeman complex wholes such as multinational corporations and (2004) did for the entire field of network analysis in 2004 supranational systems. Those important entities and the in The Development of Social Network Analysis (2004). problems they represent should not be left to economists Ultimately, I have not done anything like those but rather and politicians. Concepts considered include system evo- have tried to share with the ‘‘SNA and data mining com- lution and increasing complexity; anthropological views of munity’’ my own anthropological perspective framed dur- transactions, relations, modes of transactions, spheres of ing more than six decades of network thinking about a transactions in multicentric economies, complexity across broad range of anthropological problems. the full range of embedded networks—material, biological, Since 1960 I have viewed all people, things, and ideas in and sociocultural. Social network analysis can help to dynamic relationships. That perspective is a network per- define systems at various levels of integration, both within spective and at the same time anthropological, combining communities and in the widest conceivable supranational ethnographic, historical, holistic, and comparative views. It level. Techniques such as regular equivalence and block- is valuable and beneficial to the community of scholars who modeling are useful in sorting the subsystems of complex use network analysis to try to understand what is going on, what went on before, and what the future prospects are. As an anthropologist, I have always been interested more & A. W. Wolfe ( ) in the wholes generated by network linkages—systems of Department of Anthropology, University of South Florida, 4202 E. Fowler Avenue, SOC107, Tampa, FL 33620-8100, USA households, bands, lineages, communities, corporations, e-mail: [email protected] governments—than in the individual persons linked. Even 123 4 A. W. Wolfe now, when personal network ‘‘communities’’ are getting so of the vastly changing world by the use of new advances in much attention I continue to try to use network analysis to mathematics and in the technology of electronic data pro- better understand the more complex wholes such as multi- cessing, anthropologists seem to have taken a different national corporations and supranational systems. Those direction that leaves network analysis to others. Just as we important entities and the problems they represent should anthropologists were poised to use network models to not be left to economists and politicians! describe and explain, after the fact, remarkable, dramatic, Believing that everything we observe has a network evolutionary changes such as the agricultural ‘‘revolution’’ structure—or can be better understood using a network per- (from bands to tribes to villages), the political evolution spective—I have found myself using ‘‘network perspective’’ from tribal organization to chiefdoms, the urban ‘‘revolu- or ‘‘network thinking’’ in all that I have surveyed during tion’’ from small settlements loosely organized in ‘‘aceph- 60 years of anthropology applied to all sorts of subjects alous’’ societies to state organization around urban centers. including ethnographic studies of hunters and gatherers in the A crucial aspect of the process of cultural and social Congo, studiesofneocolonialism in Africa, studiesofhousing, evolution has been the development of ever ‘‘higher’’ employment and urban migration in America, applications of levels of integration, what anthropologists tend to call, in anthropology to health and human services generally, and the order of their development, bands, villages, tribes, trying to understand the evolution of complex systems. chiefdoms, and states. ‘‘Network’’ tends to be in the title of whatever I write about. As these formations developed, beginning with aborig- Examples include Applications of network models to drug inal bands, the numbers of people organized and the abuse treatment programs (Wolfe 1980a, b); The Uses of complexity of their organization obviously increased. Network Models in Health and HumanServices (Wolfe 1981); Clearly that has had methodological implications. The Network models of the urban environment (Wolfe 1984); anthropologist studying people in bands could, in a few Network Thinking in Peace and Conflict Studies (Wolfe years, know personally every member of the society. 2004); Network Perspectives on Communities (Wolfe 2007). Obviously that is not the case for larger societies with Both Johnson in 1994 and Linton in 2004 recognized the urban populations. Understanding what is going on at those contribution of early anthropologists to the early origins of levels of integration is crucial to understanding not only the network analysis—its prehistory, as Linton says, and the process of evolution but the maintenance of some kind of metaphorical stage as Johnson dubbed it. Johnson (1994) regularity when the society is operating with a number of reported that prior to 1970 three of the top five scholars subsystems at the lower levels. perceived to be most influential in ‘‘the social network After a 100 years of over generalizing about evolution community’’ were anthropologists. Strikingly, by 1988, many anthropologists have given up trying to explain how according to Hummon and Carley (1993), only one these increasingly complex systems came into being. I anthropologist was in the list of the top ten social network think the use of network models and network analysis can scholars, while sociologists dominated the field. be of great help. Unfortunately, my favorite discipline of anthropology has More than 60 years ago Steward (1951) presented a not moved as rapidly in this direction as I had expected when clear structural and processual model of the relations I ventured the guess in 1978 that although network thinking among these levels of organization: had already shown a dramatic rise since the 1950s, it would ‘‘In the growth continuum of any culture, there is a really soar in the next quarter-century. I thought that would succession of organizational types which are not only be the case largely because much of the weighty burden of increasingly complex but which represent new data collection and data analysis was being lifted from us by emergent forms’’ (1951, p. 379). ever more rapid electronic data processing (Wolfe 1978, p. 61). While some of that burden was lifted, anthropologists ‘‘The concept is fairly similar to that of organiza- in great numbers still did not turn to the mathematics that is tional levels in biology. In culture, simple forms, such required to take advantage of those developments. Further- as those represented by the family or band, do not more, and this was important to me, the growing complexity wholly disappear when a more complex stage of of the evolving systems I was most interested in required development is reached, nor do they merely survive more and more data and more and more analysis. fossil-like, as the concepts of folkways and mores formerly assumed. They gradually become modified as specialized, dependent parts of new kinds of total 2 System evolution and increasing complexity configurations’’ (1955, p. 51). After setting the stage in our ‘‘prehistory’’ for some great Structurally oriented anthropologists working in Africa leaps of the kind that would have promoted understanding in the late 1940s and 1950s, mostly British, developed a

123 Anthropologist view of social network analysis and data mining 5 similar view of the importance of distinguishing levels of In 1962, I described this network of corporations organization, but they tended not to put the view in evo- exploiting the non-ferrous metal industry of Southern lutionary perspective. Commenting later, Fortes (1969) Africa in an activist article entitled ‘‘The Team rules mining noted that among the most important results of their careful in Southern Africa’’ (Wolfe 1962). By 1963 I saw this comparative studies of many African societies had been network as an evolutionary development, and described it ‘‘the analytical separation of the politico-jural domain from as such in a paper, entitled ‘‘The African Mineral Industry: the familial, or domestic domain within the total social Evolution of a Supranational Level of Integration,’’ the first universe of what has been clumsily called kinship based publication in which the development of a supranational social systems’’ (p. 72). Even while recognizing the system is recognized as a major evolutionary saltation. importance of that distinction, those British anthropologists ‘‘I found the mineral extraction industry of southern were not interested in studying the processes of change Africa to be organized in an intricate social system which might be evolutionary: a ‘‘growth continuum’’ in based more on overlapping membership of a variety which simple forms do not wholly disappear when a more of groups than on a bureaucratic centralization of complex stage of development is reached (Steward 1955). administrative power. The network binds groups that To those anthropologists who visualized a structural are different both structurally and functionally, some perspective, those studies of relations among individuals business corporations, some states, some families, in based on kinship led easily to relations among categories a modern supranational structure that is more than and groups. Radcliffe-Brown (1957) published A Natural just international….The several hundred mining Science of Society in 1957 a series of lectures that he had companies operating in southern Africa are integrated presented at the University of Chicago 20 years earlier. He through a series of relationships that focus on some of called for better measurement of relations so that ‘‘rela- the larger among them…. Then, in a variety of ways, tional analysis,’’ a new kind of mathematics, would help these corporations are linked with governments’’ identify systems. (Wolfe 1963, pp. 153–54). By the time Radcliffe-Brown published those com- ments, some anthropologists were already beginning to My interpretation of the development of this complex think more about ‘‘complex societies’’ but the kinds of supranational network was that we were seeing Julian measurement Radcliffe-Brown thought of were not yet Steward’s evolutionary theory being played out at a level existent. Anthropologists spoke easily about cultural and of integration above that of states. As an assistant profes- social ‘‘systems’’ but they were slow to adopt mathematical sor, I was elated to receive an accolade from the senior formalizations of those concepts. The society-wide social professor whose theory I had found so useful. In a formations called segmentary lineage systems described December 1962 letter, Professor Steward wrote: for many African peoples are good examples of that— ‘‘I am more than gratified than I can tell you that my Evans-Pritchard (1940) for the Nuer, Fortes (1940) for the concepts have found such incisive application and Tallensi, and Bohannan and Bohannan (1953) for the Tiv. brilliant illustration in the paper that you so kindly A decade or so after those studies, I did ethnographic sent me. When first published in the Southwest work among the Ngombe peoples of the Congo whose social Journal, the idea of levels seemed almost too simple organization had that kind of structure, lineage segments and obvious to have great utility. In subsequent years stratified so that there was a clear hierarchy of systems in I have found it increasingly valuable, but I had never which functions or operations were so distributed that there imagined its applicability to the kind of supernational was no centralization of power or authority (Wolfe 1961). [stet] levels you discuss. You open up a vast area of Structurally, a network was the perfect model. possibilities. Pursuant to those observations I began to study the ‘‘You not only express the concept very correctly but modern mining industry in the Congo and the mineral-rich you go somewhat beyond what I have written and area south of the Congo. The multiple connections among relate the concept to evolution in what seems to me a the mining companies led me to see a highly influential very crucial way. That is, you connect the emergent whole network that made many other area-wide events and qualitatively new forms to qualitatively new much more understandable. processes (Steward personal correspondence 1962). Furthermore, the dynamic growth of the structure seemed to fit well those ideas of Julian Steward that I just Another anthropological giant, Mead (1967), found the quoted above: ‘‘In the growth continuum of any culture, idea of a network of corporations operating at a suprana- there is a succession of organizational types which are not tional level interesting for its possible application to world only increasingly complex but which represent new peace. I was quite interested to see her cite my analysis of emergent forms’’ (1951, p. 379). that supranational system as identifying an emerging form 123 6 A. W. Wolfe of acephalous control, against which rebellion or revolt is 3 Anthropological view of transactions, relations, structurally difficult. economic spheres ‘‘One of the principal contributions of anthropology Before going onto discuss further what I called ‘‘sorting the should be to distill from our available treasure house subsystems’’, permit me to identify some other anthropo- of small and unusual social models—many of them logical discoveries and interpretations that also fit into the outside the single narrow and steadily converging category of preparation for a better social network analysis, mainstream of ‘civilization’—new combinations and an SNA that would advance our understanding of the new forms that will release us from our historically human condition in its largest sense, pretty much as Boas limited imaginations’’ (Mead 1967: 225). would have put it: description and explanation of human Fellow anthropologists paid little attention, but my form and behavior and the varieties thereof. writings along this line upset American financial interests The elementary structure of any network is constituted and the United States government, both of which were at of elements in relation. For social networks, the elements that time obsessed with the ‘‘Cold War.’’ On April 4, 1963, are ordinarily people and the relationships can be seen as the New York Times reported that a leading American transactions or exchanges. It seems profitable to see them industrialist, Clarence Randall, had denounced me for what as such, but do not leap to the conclusion that the network he called ‘‘a scandalous attack’’ against the mining industry perspective entails a return to a Spencerian image of of Southern Africa and by implication the entire mining industrial society resting on a vast system of particular industry of the West. contracts linking free individuals (Spencer 1884). An ide- How does all this—my observation in the early 1960s of ology of individualism is neither the cause nor conse- the evolution of a system at a very high level of integration quence of adopting a network approach. Network linkages, in a complex network—relate to SNA and data mining? though they entail transactions, do not necessarily imply We were just then learning what might become possible exchanges in which maximization is a dominant principle, with computers–IBM card readers, basic programs, For- where giving is contingent only upon receiving. tran, COBOL. We are told by authors too numerous to cite that each Could one understand the system at the highest level, the individual tends to maximize his/her own satisfactions. On supranational level, before fully understanding the sub- the other hand, we are told by Mauss (1990) but by systems that operate within it? It was only later that I read numerous others since, that a gift creates an obligation for Simon (1977) on the architecture of complexity: its recipient. On the first premise has been built a socioeconomic ‘‘A complex system, made up of a large number of model that has tended to dominate thinking in the social parts that interact in a nonsimple way, will evolve sciences for more than a century, the model that sees from simple systems much more rapidly if there are integration based on the market principle. Price becomes stable intermediate forms, ‘sub-assemblies,’ than if the focus, directly related to demand and inversely related there are not, and the resulting complex form in the to supply. The market price, they say, tends to stabilize former case will be hierarchic’’ (1977, p. 209), near what Smith (1789) called the ‘‘natural price,’’ what is and also, more likely to be referred to today as ‘‘actual producer’s cost.’’ ‘‘In hierarchic systems we can distinguish between On the second premise, that a gift creates a reciprocal the interactions among subsystems, on the one hand, obligation, is built a reciprocative socioeconomic model and the interactions within subsystems—that is, that has not enjoyed anything like the success of the market among the parts of those subsystems—on the other. model, but has been elaborated in greater or lesser detail by The interactions at the different levels may be, and Malinowski (1922), Levi-Strauss (1949), Polanyi (1957), often will be, of different orders of magnitude’’ Bohannan and Dalton (1962), and others. (Simon 1977, p. 209). According to this view, a system of exchange relation- I was faced with enormous problems at every level of ships is started and maintained by persons recognizing their data collection and analysis. I kept looking for ways of mutual obligations of giving and receiving. The three forms ‘‘sorting the subsystems of complex societies.’’ Convinced of economic integration outlined by Polanyi (1957) may be as I was that network analysis would help us to differen- profitably viewed in this perspective. tiate the levels of integration in societies—whether bands, Systems integrated by the mode of market exchange are tribes, states or supranational systems—it was a difficult those in which the rates of exchange are near the floor thing to prove given the limited data available and given because the dominant, though not the only, tendency is to the analytical programs available during those times. get as much as one can for as little. Some go without what 123 Anthropologist view of social network analysis and data mining 7 they need or want. Correspondingly, systems integrated by the same as a human organism, and when that corporate the mode of reciprocity, while assuring that none go ‘‘person’’ is a vast conglomerate of some kind the situation is without, see rates of exchange above the floor that the even further removed from the ideal of economic theory. market alone might set. Further, systems integrated by the Anthropologists have for decades been studying all mode of redistribution are those where the rates of kinds of peoples organized in all varieties of societies with exchange are manipulated on other principles and ‘‘prices’’ all manners of ways of doing and thinking. What light have may be either high or low because neither tendency anthropological studies thrown on the nature of these dominates. actors, the nature of their resources, and the nature of their relations? Mauss put the question in the early part of the twentieth 4 Modes of transactions and spheres of transactions century: ‘‘What is the principle whereby the gift received in multicentric economies has to be repaid?’’ (Mauss 1964, p. 1). And we have only slowly developed the intellectual capacity to appreciate that I have always admired the way Emerson (1969, 1973, there are a number of such principles (Polanyi 1957; 1976) emphasized that if one wants to address phenomena Gouldner 1960; Bohannan and Dalton 1962; Meeker 1971). like social power, one must recognize that it is always It took even longer to recognize some of the varied ways embedded in an exceedingly complex structure of in which the actors could be conceived, even though there exchange relations among people and therefore social had been discussions of many forms of ‘‘corporations’’ relations rather than persons must be taken as the unit of when anthropology was just beginning. Maine Henry analysis. This leads one to recognize the exchange relation Sumner Sir (1917[1884], pp. 141–160), for example, spoke as the fundamental unit of analysis. This relation may be of the differentiation of ‘‘corporation sole’’ from ‘‘corpo- symbolized by A(x);B(y), representing the mutual contin- ration aggregate’’ in the nineteenth century, and yet we gency between the behavior of A and the behavior of B. have not fully assimilated that and other kinds of ‘‘corpo- In that formalism, A and B represent actors (persons or rations’’ in describing some societies where it would be other corporate social units), x and y represent resources helpful to our understanding, for example, the Nyakyusa of that they potentially exchange, and the ‘‘;’’ represents their Africa (Wilson 1951). relation. That relation can be quite complex, representing It was late in anthropological experience that some giving and receiving in various modes. began to appreciate that many transactions in Africa are not Considering the exchange relation as some kind of made on a person’s own account but by the person as minimal whole, we must also understand as well as we can representative of some other kind of actor A, a ‘‘lineage’’ in how that elementary structure is constituted. It is a little the Congo (Wolfe 1961), a ‘‘family estate’’ in East Africa risky to assume that we know about A and B, x and y, and; (Gray 1963; Gray and Gulliver 1964), etc. just because we are human beings who participate in To this day, anthropologists have not done as well as we exchange relations and are involved in those complex might have in recognizing the variety of categories of networks. Our particular views of what kinds of actors resources that people through their institutions may control, A and B are, what kinds of things x and y are, and what value and exchange. Anthropologists have not yet offered a kinds of relations the semi-colon ‘‘;’’ may represent are set of categories as reasonably comprehensive and useful most likely strongly colored and indeed sharply delimited as that of psychologists Uriel and Edna Foa (1971, 1974, by culture. 1993) whose six ‘‘categories of interpersonal resources’’ Deeply ingrained in Western cultures is the notion that (love, status, information, money, goods and services) individual human organisms have a significant degree of seem to have a regular relationship to one another in our autonomy, and that each acts to maximize his or her own cognitive structure. Nobody knows the extent to which satisfactions. Economics, which tends now to guide decision those categories are universal or whether the structure of makers at all levels, local, national, and international, is their relations is universal. based on the premise that the value of x and the value of y are Putting these several considerations together, and determined by all the As and Bs relating to one another on adopting a network perspective, leads to an argument that that basis, with the further assumptions that the wants of we can do a better job of understanding so-called ‘‘eco- A and B are insatiable and that the As and Bs are all about nomic’’ issues than we have up to now. The anthropolog- equally informed. It does not take much searching to find ical conception of spheres of exchange (Bohannan and situations that do not conform to that picture. Even in our Dalton 1962) is a more sophisticated development on the own society and economy, we sometimes define an actor A as much older, simpler, contrasts of ‘‘dual economy’’ (tradi- the union of two or more persons who have somehow joint tional subsistence vs. market) (Boeke 1953) and of ‘‘sub- control over a resource x. Such a ‘‘corporate’’ A is clearly not sistence’’ versus ‘‘prestige’’ economies (Herskovits 1952). 123 8 A. W. Wolfe

Within each sphere certain resources are exchanged along number of small establishments not characterized by the pathways entailed in relationship of various kinds. The ‘‘satellite or peripheral status’’ (1985, p. 507). It impresses values of whatever is being exchanged are not universal, me as another elaboration of the anthropological notion of but are rather established within those clusters of relations ‘‘spheres of exchange,’’ Pleased as I am that Granovetter’s associated with spheres of exchange. This is true regardless 1985 insight into this situation of embeddedness has been of the ‘‘type’’ of ‘‘economy’’ one is dealing with, subsis- given much attention, I still wonder why the anthropo- tence, market, peasant, prestige, formal or informal. In logical interpretations of different transactional and eco- every case there is a set of persons or other social actors nomic spheres of decades earlier were so often ignored. related in such a way that transactions occur with some They fit right into the social network approach. frequency among them. The following are some examples: (a) the set of Pacific Islanders exchanging arm bands and 5 Complexity across the full range of embedded necklaces and other things described by Malinowski networks: material, biological, sociocultural (1922); (b) the set of Nigerian Tiv exchanging cattle and ritual We understand that if we are to describe and explain the offices in what Bohannan (1962) called the Shagba structures generated by transactions and relations among sphere; modern corporations and states we must certainly think in (c) the set of Darfur agriculturalist maintaining the terms of billions of transactions and billions of relations. spheres of exchange of labor and beer described by This is why vast databases are required and very fast Barth (1967); computers to mine the data contained in those databases. It (d) the set of farmers exchanging labor and capital in the is not an impossible task, however, because these relations Northern Plains (Bennett 1969); are ordered and we will ultimately be able to find those (e) the set of employers and workers in the ‘‘dual labor orders. The embeddedness of these multicentric ‘‘econo- market’’ of the U.S. (Piore 1971) or in the informal mies’’ in various levels of integration is no doubt sector of urban employment in Ghana (Hart 1973); describable, but it will require lots of data collection and (f) the set of corporations and states that formed the very efficient programming. nonferrous metal industry network in southern Africa Invited to present the Keynote address at the 2003 (Wolfe 1963, 1977); INSNA meeting, I decided to talk about network aspects of (g) the set of corporations whose relationships establish the full range of evolution of systems. That address was the commercial biotechnology industry described by published in 2005 in Connections under the title ‘‘Con- Barley (1990); necting the Dots without forgetting the circles’’ (2005). Although Bohannan and Dalton’s original definition of The sub-fields of anthropology—physical/biological multicentric economy included the stipulation that anthropology, archaeology, linguistics, and social/cultural exchangeable goods fall into two or more mutually exclu- anthropology—involve systems, and thus the systemic sive spheres, the ‘‘mutually exclusive’’ criterion was whole is relevant. My view of the physical universe is a set quickly abandoned, for it was clear that conversions of nested subsystems down to molecules, corpuscles, (transactions between spheres) inevitably occur, and are, atoms, and quarks, and beyond that to strings or whatever. indeed, an important part of the dynamics of all systems. In the opposite direction are hierarchies of embedded The network perspective is important to our understanding systems such as planets, satellite systems, stellar clusters, of these dynamics. galaxies, metagalaxies, and the universe. We certainly agree with Granovetter (1985) when he There is no question that the material world is hierar- says that the problem of economic theory is not so much its chical. If you were a scientist fifty billion years ago this naive psychology as its neglect of social structure. Using material world is all that you could have known. That is what he calls ‘‘embeddedness analysis’’ in studying the all there was—anywhere. Only very recently did scientists situations of large and small firms in modern situations, and of our kind, the ‘‘modern’’ kind, start understanding it especially the persistence of small firms in a market setting ‘‘scientifically,’’ and that knowledge began with some where market principles alone would tend to predict their positive understandings about levels somewhere in the demise, he suggests that their persistence may be ‘‘because middle of its range, in the area between the crystalloid a dense network of social relations is overlaid on the aggregates and the planets. We have increased our business relations connecting such firms and reduces knowledge of material systems in both directions—up and pressures for integration’’ (Granovetter 1985, p. 507). down the hierarchy, using empirical methods wherever His ‘‘embeddedness account’’ is, without doubt, ‘‘more possible. The evidence suggests it is constantly expand- useful than other approaches’’ in explaining the large ing—evolving (Fig. 1). 123 Anthropologist view of social network analysis and data mining 9

Fig. 1 A Hierarchy of Materials Systems. [Note: This figure and similar succeeding ones were created using the network drawing program, Krackplot (Krackhardt et al. 1994)] [Reprinted from Wolfe 2005]

Somewhere in the middle of the hierarchy of physical networks, and some network generalizations apply to it at systems, something quite different ‘‘evolved’’ and there is every level. no doubting that it has a directionality to it—at one point One principle that applies in these cultural circles is there was no life on earth, but life emerged from the ‘‘cultural relativism,’’ affirming that judgments are based interactions of colloidal components, and then took off. on experience, and experience is interpreted by each of us The original system generated new forms and new sys- in terms of what we have learned in our enculturation tems at hierarchically organized levels (Gould 1981, (Herskovits 1964, p. 49). This is a principle quite analo- 2002). gous to the general principle of relativity applicable to the At each level there are different systems, in some cases physical universe. It does not tell you what you should or billions of systems, and each of those systems is a network. should not do. It is a statement of how things relate to each And the whole is also a network, a network of networks. other. Mass, gravity, and movement are interrelated in the Having emerged out of the physical systems, the biological one case; judgment, experience, and learning, in the other. ‘‘life systems’’ form their own hierarchy: cells, organs, Relativism does not mean an end to scientific activity, organisms, species, clades, etc. (Fig. 2). rather it changes the way we conduct science. What followed that was just as startling. After some Theories of ‘‘cultural evolution’’ are not popular in billions of years of biological development the interactions anthropology these days, but empirically there can be no of components that were already there generated another question that there has been a general trend represented in extremely interesting new systemic pathway. We call it the systems that are associated with culture. Greater culture, or the sociocultural system. It was new in the sense numbers of people can organize themselves, and there has that there was a time when it did not exist, several million been a tendency toward more complexity because of the years ago. Then, when the interaction of elements in the numbers of levels at which people do organize themselves. biological world reached one of those equilibrium-punc- For millennia, human beings lived in small local groups tuating events, it appeared. Now, a million years later, or bands of multi-family local groups that followed some socio-cultural systems in a variety of forms, are very learned, cultural, systems of mating, getting food, educat- prominent on our planet, so prominent they are even ing their young, procreation, and so on. At some point, causing climate change (Fig. 3). some of these thousands of autonomous bands organized Human culture is different from other cultures you multi-band systems that were more successful in achieving might know or might believe to exist. This new pathway— goals. So, thousands of years after they already could have along which human socio-cultural systems are develop- done so, we find evidence on all continents that many of ing—may be as different from the biological as the these populations developed what anthropologists have biological was from the material. But it is still made up of tended to called ‘‘tribal’’ organization.

123 10 A. W. Wolfe

Fig. 2 Biological Systems Hierarchy. [Reprinted from Wolfe 2005]

Fig. 3 Hierarchical Arrangement of Material, Biological, and Sociocultural Systems. [Reprinted from Wolfe 2005]

6 Social network analysis to define systems at various I concluded a paper entitled ‘‘Sociocultural integration level of integration above the level of the state’’ with optimism: Data on the relations among corporations, nation-states, cities, families As I said earlier, the embeddedness of multicentric and individuals can now be analyzed with the help of ‘‘economies’’ in various levels of integration is no doubt network models from graph theory and dynamic nonlinear describable, but it will require efficient programming and models of generative processes, toward the end that we will lots of data collection, data mining, and so is it with these be able to appreciate the evolutionary significance of the other hierarchical systems, they are describable. In 1982, development of a supranational system (Wolfe 1982).

123 Anthropologist view of social network analysis and data mining 11

The optimism I expressed in 1982 kept me motivated to social networks and not neighborhood solidarities. He work toward finding programs that would help us sort out continues: ‘‘The social network approach enables the subsystems at various levels of integration, including a authors in this book (Wellman 1999) to study community system at the level that most interested me, the suprana- without necessarily assuming that all communities are local tional level that I believe to be just now evolving. Unable solidarities. They do so by defining community as personal to prove much at that high level, it seemed necessary to test community, a person’s set of ties with friends and relatives, the ideas first at lower levels. neighbors and workmates (Wellman 1999:xiv–xv)’’. Because local communities seem to be universal, that Defining community in that way is not helpful for my level seemed a good prospect. Ralph Linton, in his 1936 purposes because it distracts from the vision of a com- masterpiece ‘‘The Study of Man,’’ argued that there are munity as a measurable level of organization somewhere only two social units that appear to be as old as the human between a household/family level, and some higher level species itself—the ‘‘basic family group’’ and the ‘‘local like tribe, state, nation, etc. above which is developing now group, an aggregation of families.’’ The latter ‘‘served as one or more supranational (above-state) levels. the starting point for the development of all the current Somewhere among those levels we should be able to types of combined political and territorial units such as identify a structure—even a loose cluster or set of nodes, a tribes and nations’’ (1936, p. 209). set of interlocking circles, a set of equivalent nodes—that My project of trying to use network analysis to define is doing what Linton said there would always be a need for, the subsystem that is almost universally recognized as making that connection between the immediate biological community proved more difficult than I had imagined. As I realities of humanity and the longer term historical conti- reported in ‘‘Network Perspectives on Communities’’ nuity of human institutions. That complex whole, the (2007), ‘‘People who write about ‘‘community’’ nowadays structured set of phenomena, is what I would call com- use the term in a wide variety of ways. The ‘‘little com- munity. The individuals and their apparently egocentric munities’’ that Redfield (1960) wrote about have very little relations are obviously crucial to the relations at all levels, in common with the present-day ‘‘community of nations, but there is more to the relations among the higher-level the community of Jamaica Plain, the gay community, the formations than just the interpersonal egocentric relations IBM community, the Catholic Community, the Yale of individuals. Thus, the egocentric sets that sociologists Community, the African American community, the ‘vir- and psychologists now make so much over, while neces- tual’ community of cyberspace,’’ all mentioned by Putnam sary to a community as well as to the individuals, are not in (2000). themselves the community. The editors of the Encyclopedia of Community: From It is a kind of reductionism to put so much emphasis on the Village to the Virtual World (Christiansen and Levin- individual interpersonal relations when one is trying to son 2003) tell us that Robert Putnam’s book, Bowling understand a community, a society, or other macro-system. Alone (2000) was by far the work most cited by the hun- We do not come to understand the biological human being dreds of authors of articles in their four-volume encyclo- by studying only the relations among the cells. Rather, we pedia. Carrying the subtitle ‘‘The Collapse and Revival of study the relations among cells in various structures, such American Community,’’ Bowling Alone (2000) should be a as organs. The organs are variously related to one another good source in which a curious student might look to find a even though those depend also on inter-cell relations. We definition of community. will not understand human societies or the social systems Before I could get on with my effort to use SNA in of our entire species if we focus on the interpersonal describing systems at the community level of integration, I relations among six billion individuals without taking into spent lots of effort looking at the history of community account the various structures at intermediate and even studies in anthropology and the social sciences generally. I global levels. became convinced that the level of integration that I would call community includes a wide range of social formations, generally local systems of fairly densely connected persons 7 Network perspective on communities in households and organizations, systems on a scale somewhere between those domestic households themselves To see a community as a network, a first step might well be and the wider society—the state or nation. Within this to collect data on egocentric networks and collate them. range, several authors have identified diagnostic features of The next step should be to recognize the structuring pro- different types of communities (Arensberg 1955; Redfield vided when several individuals belong to the same groups. 1960; Wellman et al. 1988). A network perspective on communities—or on structures In 1999, in the preface to Networks in the Global Vil- relating to communities—includes seeing those groups lage, Barry Wellman wrote that communities are far-flung both as networks of the individuals composing them and as 123 12 A. W. Wolfe nodes related to each other through their common mem- analysis, ‘‘local’’ does not have to mean geographical pro- bers. Such ‘‘affiliation networks’’ are a little more complex pinquity. But as one observes the current scene, that is still than being just the sum of the personal networks. the most likely scenario. Obviously, people or groups who The nodes of an affiliation network are not all of one are physically adjacent have a common environment and kind; they include both persons and groups. This in itself is some degree of common experience, both of which con- complicated enough, but it gets more complex when we tribute to our conception of community. recognize that the groups are themselves quite varied both Burt goes onto concern himself with the redundancy of structurally (internal relationships among their members ties that the condition of structural equivalence implies. and relationships with other groups and with other kinds of ‘‘Structural hole’’ is the term Burt uses to refer to structural nodes) and functionally (what they do for their members locations where there are few ties between denser segments and what they do with respect to persons and groups of a network. Such places would be what I have above external to themselves). Laumann et al. (1978) discussed called seams or gaps. For me those terms, seams or gaps, such complications in ‘‘Community Structure as Interor- evoke a more realistic image of the situation than the term ganizational Linkages.’’ ‘‘hole.’’ In any event, Burt (2001) makes the point that We know these human social networks are seldom while the conventional wisdom seems to have it that dense clearly bounded. At the margins of every egocentric net- networks with high redundancy characteristic of cohesive work, relations tend to shade off by degrees rather than or structurally equivalent sets indicate high social capital, a being definitely on or off. Adding groups into the sets of good argument can be made that the ‘‘structural holes’’— relations we are considering does not make the matter of areas of sparser connections—actually increase social definition any easier. But we can set criteria for recognizing capital by providing the opportunity for competitive ‘‘boundaries,’’ and there may be some regularities that will advantage. He brings the two seemingly opposing propo- be discovered as ‘‘natural’’ seams or tears or gaps between sitions together in a ‘‘productive’’ way, saying that ‘‘while segments of the whole seen in network perspective. brokerage across structural holes is the source of added Historically, some ties between different communities value, closure can be critical to realizing the value buried in have been getting stronger over the years, due to increases the structural holes’’ (2001, p. 52). in velocity of travel and increases in telecommunications, If it is not obvious to the reader that structural equiva- etc., blurring the boundaries between such network seg- lence and structural holes are crucial to understanding the ments and blurring the boundaries between communities. network aspects of communities, that may be in part Each community can be seen as a fairly complex cluster because community is not a concept that Burt uses, except in a larger network. When the gap or seam between two quite metaphorically. For example: ‘‘Groups can be dis- communities is bridged with numerous ties, the two clus- tinguished on many criteria. I have in mind the two network ters will, at some point be effectively merged. What had criteria that define information redundancy (cohesion and been two communities will become one. For measuring structural equivalence), but it is just as well to have in mind such processes, I recommend analyses something like those a more routine group: a family, a team, a neighborhood, or that Freeman developed in ‘‘On Measuring Systematic some broader community such as an industry’’ (2001, Integration’’ (1978), or perhaps something like those used p. 47). I cannot find an instance where Burt talks about a in cultural consensus analysis (Romney et al. 1986). whole community in the sense that we have been thinking of it. Whatever his conception of community, I believe Burt, like me, would see it as a network, a complex network 8 Seams between segments or clusters within the larger identifiable through application of network criteria. network

Another structural analytic procedure that is applicable to 9 Regular equivalence our topic of network perspective on community. Burt (1982, 1992, 2001) deals with this sort of situation using Beyond structural equivalence, ‘‘regular equivalence’’ is the concepts of structural equivalence and structural holes. perhaps even more important to understanding the network Within a community seen as a network, there are sets of structure of communities (White and Reitz 1983; Borgatti nodes (whether the nodes be persons or groups) that have and Everett 1992; Doreian 1999). Regular equivalence is the same or similar relations with others. even less dependent on physical or geographical propin- The nodes of such a set, whether they are tied to each quity, but is, in my opinion, crucial to understanding other or not, are said to have structural equivalence. This is community as a complex network. an important thing to know, whether the ‘‘local’’ ties of a set The more complex the system of actors, the less does of persons and groups are equivalent. Recall that in network structural equivalence alone, with its local focus, tells us 123 Anthropologist view of social network analysis and data mining 13 about the whole. Borgatti and Everett (1992) put it this Regular equivalence coefficients were calculated for all way: ‘‘The concepts of structural, automorphic, and regular pairs. Then, multi-dimensional scaling coefficients were equivalence are listed in order of increasing generalization: calculated for the matrix of REGE coefficients. Applying Any pair of nodes that is structurally equivalent is also hierarchical cluster analysis to the MDS coefficients, the necessarily automorphically and regularly equivalent, and nodes were sorted into cluster/blocks, paying attention to any pair of automorphically equivalent nodes is also reg- the proportion of ties within each block and the proportion ularly equivalent’’ (1992, p. 4). Regular equivalencies of ties between each pair of blocks, etc. reveal more general structures beyond the ‘‘local’’ ones Figure 4 shows the distribution of those 577 nodes found with structural equivalence. As Patrick Doreian says, plotted with each node being labeled with its cluster ‘‘At a conceptual level, regular equivalence may be more number. A first division into three sets is clearly visible useful than structural equivalence in representing roles and with the naked eye. There is a top set (a cluster of nodes) role structures. For each equivalence, a position is occupied all bearing the label 6, a middle set of clusters of nodes by equivalent actors’’ (1999, p. 7). Where the networks are bearing the labels 1, 2, 3, and 4, and a bottom set of clusters complex—and in most communities, networks are com- of nodes bearing the labels 5, 7, and 8. plex—regular equivalence can reveal structures within and The distribution is, as I predicted, indicative of differ- between communities. ential participation in subsystems at different levels of Even beyond that, analysis of the patterns of relation- integration. The organizations in Cluster 6 may operate ships among persons or corporations or other nodes in a predominantly in a wider-scale subsystem in terms of both large complex network can tell us the degree to which that area and function. The others, while still participating in network has a hierarchical structure even if this is not the whole, may operate predominantly in subsystems that apparent to the participants or to outside observers. are narrower in range and lower on the scale of levels of Concepts such as structural equivalence and regular integration. equivalence are helping enormously to provide definition Hierarchical cluster analysis (HCA) permitted us to of subsets of networks such as will help us define com- identify clusters within clusters at different levels of hier- munities within larger networks. archy. Figures 5 and 6 show the clusters as nodes. Figure 5 clearly revealing the three levels of integration, and Fig. 5 showing the distribution of cluster/blocks in terms of their 10 Sorting the subsystems: still average distance from one another. Figure 6 shows a view of the structure of the 577-node In ‘‘Network Perspectives on Communities’’ (Wolfe 2007) set of organizations serving children and families in the I described the use of these techniques to try to find the Tampa Bay Area, seen as 16 clusters/blocks, distributed structure of a network of 600 agencies and organizations, according to the regular equivalence coefficients of the some public and some private, that serve children and individual nodes that are included in each of the regular families in a multi-county area The question for me was equivalence sets. Remember that the clusters are based, not whether analysis of their patterns of relationships could on cohesiveness, but on regular equivalence (having sim- identify subsets of organizations that are assignable to ilar relations with others that are themselves equivalent). different levels of integration, community level being The findings thus far permit me to retain some optimism among them. Would the analysis of the relationships about my hypothesis that regular equivalence will sort out among hundreds of organizations show that some operated the subsystems at various levels of integration. Even primarily at the high levels—state or national—others though the hypothesis has not been tested formally, I think primarily at the lowest level—perhaps ‘‘neighborhood’’— we have come a long way from seeing these 577 organi- with others operating in identifiable subsystems somewhere zations randomly related in an almost chaotic network of between—in what might be community levels? relations to seeing considerable order in 16, 8, or 3 blocks I had collected data on the relations among hundreds of made up of clusters of somewhat equivalent nodes. organizations serving children and families in the several Looking at these figures as representing a view of the counties of the Tampa Bay area, administrative relations, structure of communities from a network perspective, one relations based on common clientele, and fiscal relations. must remember that the 577 organizations represented by The analysis I described was based on one 577 9 577 these 16 nodes are actually composed of thousands of adjacency matrix that summarized those complex data. persons carrying out the missions of the hundreds of Geodesic distances were calculated for those 166,176 pairs, organizations—and, of course, that each node itself repre- as were centralities, both closeness and betweenness for sents a set of organizations. Within each cluster are many each of the 577 nodes. connections that are not shown as lines. The lines that do

123 14 A. W. Wolfe

Fig. 4 Distribution of 577 organizations based on regular equiva- -0.3 is an input error, that node should have been removed prior to lence coefficients. Node labels represent membership in one of eight the analysis, and there should have been only 576 nodes. [From Wolfe sets based on regular equivalence. [The node labeled 9 at -4.9 and 2007] show represent the fact that there are some relations among in the local communities of the Tampa Bay Area. Then the organizations in clusters at the different levels. eight yellow nodes represent eight sets of organizations How does this structural view relate to the real com- whose staff help to organize and support the organizations munity? Think of the lower block or set of six green nodes that directly serve the communities. Finally, the two red in Fig. 5 as representing the community-based organiza- nodes at the top of Fig. 5 represent the sets of agencies tions most directly serving the thousands of families living whose staff members represent the larger society, the State 123 Anthropologist view of social network analysis and data mining 15

Fig. 5 Distribution of 16 clusters/blocks roughly according to regular equivalence coefficients of the nodes that make up the sets. [Reprinted from Wolfe 2007]

Fig. 6 Network of the 16 Clusters/Blocks of 576 Agencies in the Tampa Bay Area. The lines represent whether there are linkages between organizations in the different clusters. [Reprinted from Wolfe 2007]

of Florida, national child welfare agencies, the ‘‘health middle blocks acting in their communities with connections system,’’ and so forth. In terms of the ‘‘local versus cos- both upward in the cosmopolitan direction and downward in mopolitan’’ dichotomy used by Merton (1957) and the local direction. Gouldner (1957) many years back, the organizations Among the organizations in the top block are the involved in the bottom blocks are like locals, whereas those several District Offices of the State Department of Health in the top blocks tend to be cosmopolitan, and those in the and Rehabilitative Services, the predominant arms of the

123 16 A. W. Wolfe

State reaching into the local communities, and other 3. Recalculate the betweenness for all edges affected by organizations that have a wider reach. In the middle range the removal. of clusters or blocks we find the large majority of agen- 4. Repeat from step 2 until no edges remain. cies that provide direct services to children and families Girvan and Newman are encouraged by the success in the several communities within the Tampa Bay Area. of their method in identifying hierarchically ordered In the bottom range of clusters or blocks we find orga- communities in relative simple data sets such as that in nizations less focused on children, such as local hospitals, Zachary’s (1975, 1977) Karate Club study and in the net- zoos, etc. work formed by American college football competition. My interpretation of the results is that there are three They admit, however, that it is not yet feasible to use on reasonably recognizable regular equivalence sets that networks of greater scale. ‘‘Perhaps,’’ they state, ‘‘the basic appear to represent roughly three different levels of inte- principles of our approach—focusing on the boundaries of gration. Figure 4, in which each node represents a set of communities rather than their cores, and making use of ‘‘equivalent’’ organizations, shows that when you plot them edge betweenness—can be incorporated into a modified roughly according to the cohesive distances between method that scales more favorably with network size’’ blocks the network takes on the shape of roughly concen- (p. 7826). tric circles around a ‘‘core’’ if you will. Another approach, that I also mentioned in 2007, is Both of these visualizations are important to seeing evidenced in the work of Douglas R. White and several ‘‘community’’ and its social environment in a network colleagues using a graph theoretical basis that was laid out perspective. clearly by White and Harary in 2001 under the title ‘‘The I recognize that my approach to the analysis of the Cohesiveness of Blocks in Social Networks: Node Con- hierarchy of nested sub-networks has been fairly clumsy, nectivity and Conditional Density’’ (White and Harary and so far inadequate to the task that I set for myself some 2001). 50 years ago, to demonstrate the evolution of a new, higher This latter approach seems to me to be more rooted in level of integration, a system at a supranational level. anthropology. Perhaps that is only because White is a Still, I am optimistic that better approaches are being fellow anthropologist who has over the years been very developed by scholars studying complex systems. I men- helpful to me, including advice on my 2007 article ‘‘Net- tioned two of those in my 2007 publication, and they are work Perspectives on Communities’’ (2007). In any event, worth mentioning to this audience, the readers of Social Douglas R. White and his coauthor James Moody, in an Network Analysis and Mining. article entitled ‘‘Structural Cohesion and Embeddedness: A Girvan and Newman (2002) seem to be thinking about Hierarchical Concept of Social Groups,’’ link social complex networks in both biological and social hierarchies cohesion and social embeddedness ‘‘by developing a con- as I do, and that cohesive clusters with relatively strong ties cept of structural cohesion based on network node con- within each are connected by weaker ties to form wider nectivity’’ (White 2003; Moody and White 2003; White systems in a hierarchical structure, and that the identifica- et al. 2004). The structural dimension of embeddedness is tion of levels in such a system is facilitated by removing defined through the hierarchical nesting of these cohesive nodes successively to discover the structure. structures. Girvan and Newman clearly are thinking of these While neither approach has been demonstrated on any complex networks in both social and biological hierarchies system quite so complex as the kinds of social systems with as I tend to do. which I am concerned—‘‘whole’’ communities embedded ‘‘It is a matter of common experience,’’ they say, ‘‘that in ‘‘whole’’ societies—or ultimately supranational systems such networks seem to have communities in them: subsets of corporations and states integrated above the level of of vertices within which vertex–vertex connections are nation-states—there is no doubt that they are making dense, but between which connections are less dense’’ enormous advances technically and theoretically. (2002, p. 7821). They elaborate parenthetically, ‘‘Certainly Only within the past 10,000 years have we seen the it is possible that the communities themselves also join generation of very dense urban populations and evidence of together to form metacommunities, and that those meta- control over bounded territory that is associated with communities are themselves joined together, and so on in a nation-states. The reader is referred back to Fig. 3 where hierarchical fashion’’ (7821). that further development in sociocultural evolution is Their proposed method for detecting such ‘‘community’’ illustrated as a circle labeled ‘‘suprasys’’ building upon all is stated in four steps: those subsystems that were developed earlier. 1. Calculate the betweenness for all edges in the network. The pattern of development of these kinds of systems 2. Remove the edge with the highest betweenness. can be seen as creating a hierarchy not unlike the materials

123 Anthropologist view of social network analysis and data mining 17 hierarchy and not unlike the biological systems hierarchy. lavished on institutions of family and kinship and com- Certainly, these hierarchies can be seen as networks. There munity. Now, when it is critical that we understand them are the smaller networks nested within the larger ones at all and their relations, we seem to be accepting the wisdom of levels. Components of the subsumed networks have some conventional political scientists and economists. We have direct connections with components of the broader net- not subjected these concepts—business firm, corporation, works, but most connections are indirect. Hierarchical state—to analysis in the light of our own comparative and clustering expresses the general structure, but of course it is emic/etic perspectives. more complicated than any representation can show. In this twenty-first century, one should not talk about the How many socio-cultural levels are there? People world economy without taking into account the actions and everywhere learn to see some things as being natural, fixed, transactions of multinational firms and enterprises. Many and real while other things are believed to be merely multinational corporations are engaged in transactions of probable or even only possible. European cultures, from greater dollar value than the entire trade of many of the which most American ideas derive, have tended to see nation-states studied. The argument has been made that market transactions (rational exchange, maximization of every firm is included in one or another nation-state. While returns, getting the most you can for what you give) as there is a certain legal truth in that view, there are also natural, while altruism, reciprocity, and other modes of good reasons to view the situation differently (Wolfe transaction are seen as unnatural. These latter must be 2006). We are talking here about control over resources explained when they occur, whereas the former do not and control over persons. Of course, every corporation is require explanation. That narrow vision of human inter- registered in one or more states, and many transactions of action has led to the fairly specialized development of multinational corporations are included in the statistics for economics as the study of the consequences of transactions countries or states, but if you really want to know about the of the first type, and it has led to European and American world economy, you must also attempt to trace the deci- over-reliance on economics in all aspects of life. Is not it sions major corporations make about the disposition of the strange that it took generations of work by anthropologists goods and services under their control. Multinational cor- and sociologists to get some recognition of the fact that porations make a variety of arrangements to assure that market transactions are embedded in a much more complex transactions do not appear as transactions in order to avoid network of transactions? Anthropologists, at least, had been duties, taxes, imposts, publicity, etc. This kind of network talking about that for generations (Malinowski 1922; analysis will require very sophisticated data mining. Bohannan and Dalton 1962). The political state or nation-state is another construct that Europeans and Americans have come to see as a nat- 11 Concluding with hope that others will be able ural phenomenon. It seems to be treated by scholars and the to measure the evolution of a supranational system public alike as the inevitable outcome of thousands of years of evolution. Scholars have somehow got it into their heads I stated with great certainty in 1963 that that the State is the highest level of integration, something ‘‘I found the mineral extraction industry of southern natural and permanent. Africa to be organized in an intricate social system Common concepts are imbued with cultural meanings based more on overlapping membership of a variety that have been fixed in our languages and institutional of groups than on a bureaucratic centralization of memories. We put states in a categorical box, and we put administrative power. The network binds groups that business firms in a completely separate box, making it are different both structurally and functionally, some difficult to see that their interactions are generating a sys- business corporations, some states, some families, in tem at a still higher level of integration. Although many a modern supranational structure that is more than speak of globalization as a process, few have seen that just international. …The several hundred mining process as a network development process leading to a companies operating in southern Africa are integrated genuinely new social form. That new formation is at a through a series of relationships that focus on some of ‘‘supranational’’ level, above the level of any given nation the larger among them. … Then, in a variety of ways, or set of nations (Wolfe 1977). these corporations are linked with governments’’ While states and business firms have been around for (Wolfe 1963, pp. 153–54). thousands of years, in the perspective of millions of years of evolution these are both relatively recent emergents, I recognize 47 years later that evolution is a difficult having been constructed through the processes of adapta- process to prove, and its products, being increasingly tion that generate all social formations. Anthropologists complex systems, are even hard to describe. As I think have not given these forms the kind of attention we have I have demonstrated, progress is being made very slowly. 123 18 A. W. Wolfe

One problem is that research of this kind requires tons of Doreian P (1999) An intuitive introduction to blockmodeling with data that are not readily available. When I started studying examples. BMS Bulletin de Methodologie Sociologique 61:5–34 Emerson RM (1969) Operant psychology and exchange theory. In: the transactions and relations among the mining companies Burgess RL, Bushell D (eds) Behavioral sociology. Columbia there was a publication entitled Mining Yearbook published U. Press, New York by Walter E. Skinner in London, Moody’s Banking and Emerson RM (1973) Social exchange: from micro to macro theory. Finance Manual and Industrial Manual that contained lots Lecture presented at the annual convention of the american sociological association. (mimeo) of important data, but I had neither electronic data ware- Emerson RM (1976) Social exchange theory. Ann Rev Sociol houses nor data mining techniques that would have been 2:335–362 helpful. Nor were there at that time the network analysis Evans-Pritchard EE (1940) The Nuer, a description of the modes of algorithms that would have been helpful. livelihood and political institutions of a Nilotic people. Oxford University Press, New York Another problem has been that research at this level— Foa UG (1971) Interpersonal and economic resources. Science international, multinational, tends not to be fostered unless 177(3969):344–351 there is financial profit in its findings. Foa UG, Edna F (1974) Societal structures of the mind. Charles Another problem has been conceptual. I had spoken of C. Thomas, Springfield Foa UG et al (eds) (1993) Resource theory: explorations and the interaction of corporations and states developing a applications. Academic Press, San Diego ‘‘supranational’’ system well before economists and orga- Fortes M (1940) The Tallensi. In: Fortes M, Evans-Pritchard EE (eds) nizational scholars began to speak even of ‘‘multinational’’ African political systems. International African Institute, London corporations. Economists and ‘‘World System’’ scholars Fortes M (1969) Kinship and the social order: the legacy of Lewis Henry Morgan. Aldine, Chicago seem always to restrict their thinking as if corporate firms Freeman LC (1978) On measuring systematic integration. Connec- always represent states in an international system. Realis- tions 2(1):13–14 tically, states are influencing firms and firms are influenc- Freeman LC (2004) The development of network analysis ing the states in a supranational system (Wolfe 2006). Girvan M, Newman MEJ (2002) Community structure in social and biological networks. Proc Natl Acad Sci USA 99(12):8271–8276 Analytic separation of states and firms as if they are in Gould SJ (1981) The mismeasure of man. Norton, New York separate networks or systems, economy, and polity, is Gould SJ (2002) The structure of evolutionary theory. Belknap Press formalist folly. The proper network model must include of Harvard University Press, Cambridge both companies and states. Gouldner A (1957) Cosmopolitans and locals: toward an analysis of latent social roles. Adm Sci Q 2(3):281–306 Gouldner A (1960) The norm of reciprocity: a preliminary statement. Am Sociol Rev 25(2):161–178 Granovetter M (1985) Economic action and social structure: the References problem of embeddedness. Am J Sociol 91(3):481 Gray RF (1963) The Sonjo of Tanganyika. Oxford, London Arensberg CM (1955) American communities. Am Anthropol Gray RF, Gulliver P (1964) The family estate in Africa. NYU Press, 57(6):1143–1152 New York Barley SR (1990) Niche and network: strategic alliances in commer- Hart K (1973) Informal income opportunities and urban employment cial biotechnology. Paper presented at the international sunbelt in Ghana. J Mod Afr Stud 11:61–89 social network conference, February, San Diego, California Herskovits MJ (1952) Economic anthropology. Knopf, New York Barth F (1967) Economic spheres in Darfur. In: Firth R (ed) Themes Herskovits MJ (1964) Cultural dynamics. A. A. Knopf, New York in economic anthropology. Tavistock, London Hummon NP, Carley KM (1993) Social networks as normal science. Bennett JW (1969) Northern plainsmen: adaptive strategy and Social Netw 15:71–106 agrarian life. Aldine, Chicago Johnson JC (1994) Anthropological contributions to the study of Boeke JH (1953) Economics and economic policy of dual societies. social networks: a review. In: Wasserman S, Galaskiewicz J Institute of Pacific Relations, New York (eds) Advances in social network analysis Bohannan P, Bohannan L (1953) The Tiv of central Nigeria. Krackhardt D, Blythe J, McGrath C (1994) Krackplot 3.0: an International African Institute, London improved network drawing program. Connections 17(2):53–55 Bohannan PJ, Dalton G (1962) Markets in Africa. Northwestern Laumann EO, Galaskiewicz J, Marsden P (1978) Community University Press, Evanston structure as interorganizational linkages. Ann Rev Sociol Borgatti SP, Everett MG (1992) Notions of position in social network 4:455–484 analysis. In: Marsden P (ed) Sociological methodology, pp 1–35 Levi-Strauss C (1949) Les Structures elementaires de la parente. Burt RS (1982) Toward a structural theory of action: Network models of Presses Universitaires de France, Paris (trans. Published in 1969, social structure perception and action. Academic Press, New York Boston: Beacon) Burt RS (1992) Structural holes: the social structure of competition. Linton R (1936) The study of man. Appleton-Century-Crofts, New Harvard University Press, Cambridge York Burt RS (2001) Structural holes versus network closure as social Maine Henry Sumner Sir (1884) Ancient law, 5th edn. London capital. In: Lin N et al (eds) Social capital: theory and research. Maine Henry Sumner Sir (1905) Ancient Law, 4th American from Aldine De Gruyter, New York 10th London edition. H. Holt & Co, New York Christiansen K, Levinson D (eds) (2003) Encyclopedia of commu- Maine Henry Sumner Sir (1917) Ancient law. E. P. Dutton, New York nity: from the village to the virtual world. Sage Publications, Malinowski B (1922) Argonauts of the Western Pacific. Routledge Thousand Oaks and Kegan Paul, London

123 Anthropologist view of social network analysis and data mining 19

Mauss M (1964) The gift: the form and reason for exchange in archaic White D, Reitz K (1983) Graph and semigroup homomorphisms on societies. W. W. Norton, New York networks of relations. Soc Netw 5(2):193–234 Mead M (1967) Alternatives to war. In: Fried M, Harris M, Murphy R White DR, Owen-Smith J, Moody J, Powell WW (2004) Networks, (eds) War: the anthropology of armed conflict and aggression. fields and organizations: micro-dynamics, scale and cohesive Natural History Press, Garden City, pp 215–228 embeddings. In: Computational & mathematical organization Meeker BF (1971) Decisions and exchange. Am Sociol Rev theory, vol 10, no. 1, pp 95–117 36:485–495 Wilson M (1951) Good company. Oxford University Press, London Merton RK (1957) Social theory and social structure. Free Press, Wolfe AW (1961) In the Ngombe tradition: continuity and change in Glencoe the Congo. Northwestern University Press, Evanston Moody J, White DR (2003) Structural cohesion and embeddedness: a Wolfe AW (1962) The team rules mining in southern Africa. Toward hierarchical concept of social groups. Am Sociol Rev Freedom 2(1). ‘‘The Team Rules Mining in Southern Africa’’ 68(1):103–127 Wolfe AW (1963) The African mineral industry: evolution of a Piore M (1971) The dual labor market: theory and implications. In: supranational level of integration. Soc Probl 11(2):153–164 Gordon DM (ed) Problems in political economy. D.C. Heath, Wolfe AW (1977) The supranational organization of production. Curr Lexington Anthropol 18(4):615–636 Polanyi K (1957) The economy as instituted process. In: Polanyi K, Wolfe AW (1978) The rise of network thinking in anthropology. Soc Arensberg CW, Pearson HW (eds) Trade and market in the early Netw 1(1978):53–64 empires. The Free Press, Chicago Wolfe AW (1980a) Applications of network models to drug abuse Radcliffe-Brown AR (1957) A natural science of society. Free Press, treatment programs: a brief review paper. Connections Glencoe 3(2):28–29 Redfield R (1960) The little community and peasant society and Wolfe AW (1980b) Multinational enterprise and urbanism. In: Collins culture. University of Chicago Press, Chicago TW (ed) Cities in a larger context. University of Georgia Press, Robert P (2000) Bowling alone: the collapse and revival of american Athens, pp 76–96 community. Simon and Schuster, New York Wolfe AW (1981) The uses of network models in health and human Romney AK, Weller SC, Batchelder WH (1986) Culture as consen- services. Paper presented at the Sun Belt Social Network sus: a theory of culture and informant accuracy. Am Anthropol Conference, Tampa, Florida, 20 February 1981 88(2):313–338 Wolfe AW (1982) Sociocultural integration above the level of the Simon H (1977) The architecture of complexity. In: Models of state. Cult Futur Res 7(1):9–22 discovery and other topics in the methods of science. D. Reidel Wolfe AW (1984) Network models of the urban environment. Paper Publishing Co, Boston presented at the 83rd Annual Meeting of the American Smith A (1789) An inquiry into the nature and causes of the wealth of Anthropological Association, Denver, November 1984 nations. A. Strahan and T. Cadell, London Wolfe AW (2004) Network thinking in peace and conflict studies. Spencer H (1884) The man versus the state. Liberty Classics, Peace Confl Stud 11(1):69–75 Indianapolis Wolfe AW (2005) Connecting the dots without forgetting the circles. Steward J (1951) Levels of sociocultural integration: an operational Connections 26(2):107–119 concept. Southwestern J Anthropol 7(4):374–390 Wolfe AW (2006) Supranational networks: states and firms. Peace Steward J (1955) Theory of culture change: the methodology of Conf Stud 13(1):68–80 multilinear evolution. University of Illinois Press, Urbana Wolfe AW (2007) Network perspectives on communities. Structure Steward J (1962) Personal correspondence and dynamics. J Anthropol Relat Sci 1(4):Article 2 (electronic) Wellman B (ed) (1999) Networks in the global village: life in [this article is forthcoming as a chapter in book, Social contemporary communities. Westview Press, Boulder Networking and Community Behavior Modeling, edited by Wellman B, Carrington PJ, Hall A (1988) Networks as personal Maytham Safar, published by IGI Global] communities. In: Wellman B, Berkowitz SD (eds) Social Zachary WW (1975) The cybernetics of conflict in a small group: an structures: a network approach. JAI Press, Greenwich (Ablex information flow model. MA Thesis, Temple University, Edition) Department of Anthropology White DR (2003) Ties weak and strong. In: Christensen K, Levinson Zachary WW (1977) An information flow model of conflict and D (eds) Encyclopedia of community, vol 4. Sage Reference, fission in small groups. J Anthropol Res 33(4):452–473 Thousand Oaks, pp 1376–1379 White DR, Harary F (2001) The cohesiveness of blocks in social networks: node connectivity and conditional density. Sociol Methodol 31:305–359

123