Doug White's collaborations in recent years at the Santa Fe Institute, with mathematicians and social, physical and biological scientists, are the outgrowth of a career-long interest in synthesis of the human and the hard sciences, now taking place in interdisciplinary fields of evolutionary complexity, network dynamics, and self-organizing systems. The early 1960s found him searching for potential sources for synthesis, starting with a split major at Brown University in physics and the humanities. Impelled by his undergraduate dean's refusal to grant a double major to pursue such a career goal, White decided on two specific disciplines in which this synthesis could be attempted at the University of Minnesota, where he did his BA and PhD. These were Anthropology, chosen for its universal breadth of coverage of the human sciences (archaeology, prehistory, world ethnography), and Mathematics, providing the theoretical and modeling tools for the sciences. A decade later, as assistant professor at Pittsburgh, he produced the first broad theoretical synthesis in the interdisciplinary field of Mathematical Anthropology for the 1974 Handbook of Social and Cultural Anthropology. In the intervening period he had finished a comparative historical dissertation on social networks in American Indian societies, founded the Societal Research Archives System, a coded qualitative database for cross-cultural research (published in the 1971 Handbook of Method in Cultural Anthropology and distributed online through Dartmouth College to universities and colleges in New England), taught at the NSF Summer Institutes in Quantitative Anthropology, co-founded and co-directed with G. P. Murdock the Cumulative Cross-Cultural Coding Center, and developed and published with Murdock the Standard Cross-Cultural Sample (SCCS) as a framework for scientific collaboration in comparative coding efforts (continued since 1985 in the World Cultures e-journal). His 1971 reconstruction of social networks among the Natchez people was a classic example of how processual and network modeling could solve analytical problems in the study of social structure, such as the famous "Natchez Paradox" of virtually every introductory text in Anthropology at that time. As a unique "traveling scholar" PhD student, before joining Murdock as Assistant Professor at Pittsburgh at the age of 25, he had also taken a year's work in Mathematical Psychology at Michigan (Coombs, Harary, Guttman, Boulding), six months of Mathematical Sociology at Columbia (Lazarsfeld, Henry and others at the BASR), and three months at Northwestern (Don Campbell and Raoul Naroll).
In 1969 White began an intensive study of social networks in a world sample of 296 societies, some of which are in the SCCS, finding a complex but structured dynamic by which role conflicts worked themselves out in widely distributed stereotypical behaviors such as joking relationships and avoidances, each of which occurred in over 50% of his sample. Presented at the 1975 Mathematical Social Science Board interdisciplinary conference in Social Networks, organized by Holland and Leinhardt, who had just completed an exhaustive study of triadic consistency models analyzed across 750 sociological datasets. A clash of paradigms was evident between triadic consistency models and White's complexity dynamic, which required a minimum role interlock models of tetrad INconsistencies, as proposed by S. F. Nadel. The contribution was dropped from the conference proceedings, but motivated White's continued research on networks and complexity from 1976 forward with a series of NSF funded large-scale network field-study grants (part of today's longitudinal social network field sites Linkages network). These sites produced a dozen massive multilevel and networked time-series databases on human populations that enjoy the support of the large-scale dynamical network analysis software (the Pajek software produced by Batagelj and Mrvar) for which White has been a consultant and concept developer since its early days.
In 1970 White was invited to join the Mathematical Social Science Board organizing conference in Anthropology, funded by NSF and hosted by Kim Romney and Roy D'Andrade. There he met Francois Lorraine who had just completed an MA thesis at the EHESS in Paris (with social science mathematician G.T. Guilbaud) on a category theory model of structural equivalence, and had gone to work on a PhD at Harvard with Harrison White. He and Lorrain agreed to take on the development of social networks through organizing several interdisciplinary MSSB conferences in the coming years. Soon after publishing (1971) with Harrison on structural equivalence, however, Lorrain left Harvard to work with his family's business in Montreal. White, meanwhile, took on a two-year research project in Ireland (1971-73) in which he and his wife (a sociolinguist) did extensive studies of the social networks of language use in relation to Irish. They presented that work at the first MSSB conference on Social Networks, organized by H. Russell Bernard (1974). In Ireland White developed and programmed the statistical entailment software still in use for analysis of multidimensional partial-order Guttman scaling.
Returning from Ireland in 1974, White sought out a former colleague at Pittsburgh (1967-1969) with whom he had shared thoughts about networks approaches to the social sciences. Lin Freeman had since moved to Nova Scocia and then on to Lehigh. They found they had both shifted their primary research emphases to networks, yet the people interested in networks approaches -- which they saw as an interdisciplinary field -- were scattered across different universities as well as different fields. To develop social networks as an interdisciplinary field, White obtained funds from HRAF for a 1974 planning committee meeting involving Freeman, Bill Lambert, Raoul Naroll and others. Freeman submitted an NSF grant in 1976, with White as an evaluator, to pursue the goal using the first trials of research-oriented computer conferencing on the internet -- then the ARPA net -- with EIES teleconferencing run at NJIT. Funded for the two years 1977-78, Freeman and White got network researchers from different disciplines to sign up, and succeeded with sociologists, anthropologists, historians, computer scientists, geographers, mathematicians and statisticians. In contrast to today's interdisciplinary collaborations, no physicists, biologists, chemists, or engineers came forward to participate. The teleconference helped to broaden interaction in the social networks community, breeding as well the SN journal edited by Lin (1979-). During the same period, Barry Wellman founded the International Network of Social Network Analysis (INSNA) in 1976, which loosely coordinated with the Annual "Sunbelt" meetings (the first organized by Russ Bernard and Al Wolfe) starting in 1981. White hosted the 2nd meeting in San Diego in 1982.
When White moved to Irvine in 1976, he retained the vision of an interdisciplinary field of social networks, which led to his founding the Focused Research Program in Social Networks in 1978 and chairing Irvine's faculty group in Social Networks (1979), which became in 1981 the first Academic PhD Program in Social Networks. In 1979 he developed a full-scale prototype in APL for network analysis (similar in concept to the late 1990s version of Pajek and developed prior to the UCInet package), and in the 1980s developed an APL prototype for network autocorrelation analysis. He organized with Lin Freeman the 1980 UCI conference on social networks (resulting in the 1989 Freeman White and Romney edited book on network methods), and the recruitment of Freeman to the faculty in 1980.
In 1984 White co-founded with systems physicist Arthur Iberall the Faculty Seminar on Hierarchical Physics and the Social Sciences that has continued to the present day, involving faculty from UCLA, UCI, USC, CUNY, and the University of Connecticut, in the fields of physics, chemistry, biology, medicine, anthropology, psychology, economics, political science and law, and includes several of the original contributors to the sciences of complexity as well as participants in the Santa Fe Institute. Curiously, no sociologist has ever joined this interdisciplinary group, which has had only one member in common with the field of social networks.
Out of a physics of complexity understanding of social dynamics, White expanded his longitudinal field site emphasis (the Linkages project) into a set of worldwide collaborative projects initially funded by the Wenner-Gren, then by NSF, and most recently by the Mellon Foundation. He also totally revamped the cross-cultural database for testing social theories with comparative ethnographic data by adding study-restudy data as well as time series coding of historical effects on local societies of the incursions or larger scale world civilization, and with a subproject with Mike Burton also funded by NSF. In the process he developed the prototype APL software package for the study of network effects across cultures, one of the early autocorrelation packages in the social sciences (1984-85). He and his students Malcolm Dow and Karl Reitz, along with Michael Burton, published a series of simulation studies showing the effects of network autocorrelation on regression results, as well as a series of studies using the method to efficiently estimate models of causality underlying the sexual division of labor cross-culturally, this on top of his earlier entailment studies of universal gender-specific implicational chains in the division of tasks.
In 1989 White was awarded the Alexander von Humboldt Distinguished U.S. Scientist Award and began a long series of collaborations in the mathematical social sciences in Germany and France. The year spent in France before and after his Humboldt Award period was one devoted to developing applications of Galois Lattices to entailment-type discrete structure analyses of data on bipartite graphs of relations between individuals and their attributes, their group memberships, and their material possessions. This series of papers was edited with V. Duquenne in the Social Networks special issue on discrete structure analysis. The approach was extended by his Humboldt host in Cologne, Tom Schweizer, in a series of publications in Current Anthropology.
After returning from several years in Europe using software graphics packages to study networks and discrete structure lattice representations, White reprogrammed the MAPTAB software that he had written earlier for cross-cultural data analysis and classroom use, in order to use extensive graphics capabilities for mapping and data representation. He also made an important generalization of techniques employed by French structural anthropologists in studying social networks: the p-graph system of coding genetic, kinship and marriage networks so as to analyze social structure by use of graph theoretic algorithms. These algorithms were applied to reduced graphs where nodes represented marriages and links were individuals, an inversion of the privileged position of individuals as nodes in most "methodologically individualist" social network studies. This structure proved to be the basis for a kind of fundamental "Feynman diagram" for starting to understand the nature of social networks in bisexual species. For humans, it proved key to understanding the dynamics of class and ethnic group formation through self-bounding endogamous groups.
Through the work on self-bounding kinship systems (structural endogamy, first defined by White in 1997 and used in Brudner and White 1997), White began to make some fundamental scientific discoveries. A 1999 NSF grant on "Longitudinal Network Studies and Predictive Social Cohesion Theory" supported his research across 15 longitudinal field and historical sites on the hypothesis that self-bounding and hierarchical levels of k-connectedness (maximal subgroups in networks with k independent paths between all pairs of nodes) are what support the relational component of social solidarity, for which he created the term structural cohesion. The measurable variable of structural cohesion was found to have downstream effects on scores of social structural and dynamical variables in social networks. The problem of computing k-connectedness is NP complete, which can be a problem for large networks. White and Santa Fe Institute physicist Mark Newman programmed an approximation algorithm that runs in at a subquadratic level of algorithmic complexity. Because k-components are contained in k-cores (maximal subgraphs in which every nodes has at least degree k) and often equivalent to them, the Newman-White (2001) algorithm provided a feasible way to identify k-components within the k-cores of large networks.
Between 1993 and 2005, twenty-one of White's 38 articles (1996a,b,c, 1997a,b, 1998a,b,c,d, 1999a,b and 1999(web), 2001, 2002a,b,c,d, 2004,a,b, 2005a,b) and his 2005 book with Johansen dealt with reconstructing of the field of kinship and marriage from a network perspective, based on intensive analysis of thirty or so diverse case study materials -- from South and Meso- America, Polynesia, Australia, Indonesia, Sri Lanka, the Middle East, Europe, North Africa. Eight of these are his own (2), his students' (1), colleagues' (3) or colleagues' students' (2) field studies. These 30 case studies provided an empirical basis to develop a new 'practice theory' of the variety of kinship and marriage networks and to redefine complexity in terms of network structure and dynamics. This new and generalizable 'practice theory' of kinship and marriage supersedes those of Levi-Strauss, Bourdieu and a number of other major authors in the field of kinship.
White's theoretical framework evolved over the course of these 22 publications and 30 case studies by gradual increments in understanding empirical kinship networks as a new object of study previously unexplored in social anthropology. White's studies, with various colleagues, begin from what we can learn and enumerate concerning the diverse and changing patterns of individual behavior encoded in longitudinal network data, on the one hand, and changing structural characteristic of the larger network in which these behavioral are embedded, one the other. As these studies unfold, the focus moves to the dynamic and reciprocal relationship between behaviors at the individual level and emergent structures in the network as object of study. White develops methods and statistical tests for detecting the signature patterns for the organization of diverse preferences in behavior, on the one hand, and of the organization of diversity in the network structure, on the other. In anthropology this dual relation was traditionally encoded as that between organization and structure, but White moves the theory and analysis to a whole new level that allows longitudinal data (the implicit time-coding of genealogical networks) to be assessed within each case study to allow for analysis of the dynamic relationship between organizational and structural change. His approach, then, places the theory and analysis in the context of complexity theory (see his 2004 article in the journal Cybernetics and Systems).
The key to White's theoretical breakthrough for kinship networks was to show how the preference gradients for diverse types of marriages that are exhibited in each human population contain the spectral signatures for defining complex self-organizing behavior in a social network. Rather than the use made of scaling methods by physicists, aimed as showing invariance in structure in the self-organization of physical phenomena, White focused these same scaling methods on the characteristics of preference gradients that self-organizing in the sense that (a) they depart from random models of interaction (b) they contain implicit biases for behavioral dynamics, (c) these biases are mathematically related to the emergence and change in the structural forms that we see in social networks, and (d) changes in these structures form a significant part of what alters the context of further behaviors, and hence closes the loop on self-organization.
These approaches lead to new Tools for Marriage Network Analysis developed with a team of French anthropologists. Together with French historians this team obtained a grant from the French Agence Nationale de la Recherche (equivalent to the NSF) on the Informatic Treatment of Kinship Phenomena in Anthropology and History. The grant runs from 2006-2009.
Abstracts of 27 recent publications, since 1997, are found at http://eclectic.ss.uci.edu/~drwhite/abstracts98.html
pdf publications are at http://eclectic.ss.uci.edu/~drwhite/links2pdf.htm
vita is at http://eclectic.ss.uci.edu/~drwhite/6wwwvita.html
teaching is at http://eclectic.ss.uci.edu/~drwhite/seminars.html
software written is at http://eclectic.ss.uci.edu/~drwhite/softwdrw.html
on-line programs are at http://eclectic.ss.uci.edu/~drwhite/programs.html
the World Cultures journal is at http://www.worldcultures.org/~drwhite/worldcul/world.htm
Linkages field sites are at http://eclectic.ss.uci.edu/~drwhite/Java/Linkages/Linkages.html
and other links at http://eclectic.ss.uci.edu/~drwhite
and http://eclectic.ss.uci.edu/
