The Positivism of Harrison White

I was drawn to Harrison White’s theory of social networks, and to relational sociology around him from a rather Teutonic angle: My studies in political science left me disappointed about its ability to say much about the social basis, or the social embeddedness of politics, or even of the overall contours of the political realm. This pushed me into sociological theory, with first Jürgen Habermas, and later Niklas Luhmann, as guiding lights. Like Karl Marx and Max Weber before them, they were grand theorists: offering comprehensive and integrated theoretical mappings of the social world. This is what I was looking for: Theory that covers pretty much everything, with high levels of logical consistency and elegant conceptual arrangements. White’s theory is destined to disappoint on these criteria. Only I wasn’t able to see that from the start.

In the early 2000s, I started my doctoral research on Italian migrants in Germany. I wanted to apply Luhmann’s systems theory to collective identity phenomena like ethnic minority groups. However, systems theory could not help much: while elegant and cohesive, it remains mostly disconnected from empirical research. Basically, we are asked to believe Luhmann in his proclamations of the functional differentiation of society into subsystems, of the self-reproducing “autopoiesis” of these systems, and of the self-referential nature of communication. Similarly with Marx, Weber, and Habermas. So, to do empirical research, I had to look for other theoretical approaches. To supplant my systems theoretical starting point, or to complement it.

At that time, prominent systems theorists Dirk Baecker had come to praise White’s theory (1996). Like Luhmann, White does not start from individual actors, but what from the processes and social structures between them. And he offers an elaborate architecture of the social world. So, I dove into the first edition of Identity and Control (1992), only to drop it twice after the first 20 or 30 pages. I simply did not understand what was going on, or why I should bother with the rest of the book. However, I continued thinking about networks, their nature, and their place in the social world. And I suspected that the answers were somehow to be found in the book with the abstract Piet Mondrian-style art cover. Finally, I managed to get into White’s line of thinking, nudged and helped by Stephan Fuchs’s (2001) book and by an article of Ikegami (2000). After much deliberation, I decided that this was the way to go, and that I would henceforth identify as a network theorist.

In spite of my fascination with the works of White, and with the work around him by Peter Bearman, Ron Breiger, Kathleen Carley, Paul DiMaggio, Roger Gould, Eric Leifer, Paul McLean, Ann Mische, John Padgett, Margaret Somers, Charles Tilly, and many others, I felt that I was missing something. What I thought of as a consistent theory left me with a number of puzzles. Why did everybody else know how to do network theory, and how to apply it, and only I could not figure it out?

In 2005, I decided that I wanted to meet the masters. Gathering my courage, I contacted Harrison White and Charles (“Chuck”) Tilly and asked them if I could spend 1 month at Columbia, to learn more about network theory. Both of them reacted supporting my research visit, and, together with Peter Bearman, also my later application for a postdoctoral fellowship from the Humboldt Foundation at Columbia. I came to spend first a month (September 2005) and then a year (May 2007–April 2008) there. I was fortunate to spend generous time with Harrison, even co-authoring a paper with him (and with Larissa Buchholz and Matthias Thiemann; White et al., 2007), and to take part in a variety of department activities and connect to many others (Chuck, Peter, Emily Erikson, Frédéric Godart, Joscha Legewie, Tammy Smith, etc.). Shortly afterwards, Chuck died after an extensive battle with cancer, and Harrison pretty much retired from academic work, and moved to Tucson, Arizona. I consider myself fortunate to witness the sunset of the golden age of relational sociology at Columbia.

However, I did not find there what I was looking for: the clear-cut, logically integrated theory of relational sociology simply did not exist. White had certainly tried to set it out in the two editions of Identity and Control (1992; 2008). His ideas became hugely influential in stimulating theoretical arguments and inspiring empirical studies (Fuhse & Mische, 2024)—from Mustafa Emirbayer’s (1997; Emirbayer & Goodwin, 1994), Somers’s (1994), and Emily Erikson’s theoretical papers (2013) to the empirical studies by Roger Gould on the Paris commune (1995), by Bonnie Erickson on the role of culture and social networks for social inequality (1996), by David Gibson on turn-taking among managers (2005), by Eiko Ikegami on collaborative arts in Tokugawa Japan (2005), by Ann Mische on the role of movement leaders in the Brazilian student protests (2008), and many others.

However, these do not so much form part of a coherent research enterprise, following a fixed set of theoretical ideas, laid out by White. Rather, relational sociology is united by a rough orientation toward similar research practices, and to common ideas of what to look for and study. Its studies share a few buzzwords—network, identity, story, and culture. But they frequently pick up on White’s theory only in passing, and very selectively. Partly, I would argue that this comes from its idiosyncrasy: Unlike most other theoretical enterprises, including Emirbayer’s, Erikson’s, Nick Crossley’s (2011), and my own (Fuhse, 2022) rendering of relational sociology, White develops a unique array of theoretical concepts that are very much his own, and not borrowed from earlier approaches: stories and identities, domains and netdoms, catnets, disciplines, switchings, and so forth. In the following, I discuss this uniqueness and idiosyncrasy of his theory: What is Harrison White’s mode of constructing sociological theory?

I argue that, probably unknowingly, White followed the playbook of logical positivism: devising a theory that (supposedly) springs inductively from empirical observations to systematize them. As a result, the theory is more geared toward empirical research than to theoretical coherence, and it deliberately breaks out of the trodden pathways of previous sociological approaches, ignoring familiar and coveted concepts like individual action, power, class, and culture. It shuns metaphysical speculation with little connection to empirical research, as well as unobservable constructs, normative judgments, and ideology. This innovative and iconoclastic mode of theory construction also leads to its difficult accessibility. Tilly (1993) compared Identity and Control with James Joyce’s obscure masterpiece Finnegan’s Wake, mirroring the difficulties in reading and understanding of many, including myself (see above).

I cannot substantiate these claims in full here. Instead, I sketch my position by pointing out the peculiarities of White’s position, and its adherence to positivist principles. The following two sections trace the development of White’s theory from catnets, structural equivalence, and blockmodels to the two editions of Identity and Control. The last section discusses in what sense White’s theory follows the model of logical positivism.

As a trained physicist, White moved into the social sciences with an interest in the mathematical modeling of social structures. This contrasted with the grand theory of Talcott Parsons, dominant at Harvard where White took up the position of Associate Professor in 1963. However, Harvard also housed small group researchers George Caspar Homans and Robert Freed Bales, with whom he pursued his interest in sociometry. This may sound rather empirical, but the lecture notes of his student Barry Schwartz from White’s undergraduate course Introduction to Social Relations display a clear interest in developing a theory of social structures (White [1965] 2008): There, White develops the notion of “catnets”, of network segments separated by social categories, and shaped by them. While overall structuralist, this piece already points to the interplay of networks and meaning that became central to White’s thinking in Identity and Control (1992). Nicholas Mullins, one of the participants in the course, saw Harvard structuralism with White as “intellectual leaders” as one of the “theory groups” in American sociology (1973).

With the connection to Homans, it would have made sense to term densely connected network partitions “groups”. But White insisted in coining his own term “catnet”, a neologism combining “cat” for category and “net” for network. This falls in line with White’s avoidance of theory texts in his seminars and in his general reading, focusing on empirical studies from history, anthropology, psychology, and organizational studies (Padgett, 2025). He did want to arrive at far-reaching general models of social structure, but without recourse to the established models in sociological theory from Marx to Parsons. Instead, theoretical notions and arguments are to arise out of the empirical observation of a wide range of social phenomena.

At about the same time, White worked on the formal-mathematic modeling of kinship relations (1963). This led to the concept of “structural equivalence”, advanced with mathematician François Lorrain (Lorrain & White, 1971). Kinship categories like “uncle” do not bound densely connected network segments, as in the catnet concept. Instead, they mark “structurally equivalent” positions in family relations: Uncles are united in their systematic relations to other kinship categories: aunts, parents, siblings, nephews and nieces, cousins, and grandparents. Again, White’s ideas center on the interplay between networks of social relationships and symbolic forms—kinship categories, in this case.

The notion of structural equivalence led to the method of blockmodel analysis with White’s Harvard students and colleagues, Scott Boorman and Ron Breiger. The famous blockmodel papers (Boorman & White, 1976; White et al., 1976) pick up on the concept of “roles”: The relations between structurally equivalent positions in networks are interpreted as role patterns. White and his co-authors here start from anthropologist Siegfried Nadel’s vision of social structures as systematic patterns of relations between roles. This has important parallels with the well-established concept of roles in sociological theory, as devised by Ralph Linton, Robert Merton, Talcott Parsons, Ralf Dahrendorf, and Ralph Turner. In principle, the two strands of literature can well be brought together (and combined with concept of “institution”; Fuhse, 2022, p. 133–163).

However, White and his co-authors simply ignore the sociological literature here. This leads, among other things, to a lack of reflection on whether role patterns emerge out of the interaction in network contexts, as among novices in a monastery. Or whether role patterns can also result from institutionalized cultural categories, as in kinship relations, or from the formal organization of the positions of, say, professors and students, and the relations between them, in a university. The empirical phenomena are bracketed from their wider social and cultural contexts; and the role concept denotes that which can be observed—without consideration of origins (and consequences).

During the 1980s, White became dissatisfied with the prevailing structuralist understanding of social networks (Mische, 2011, p. 82). Following graph theory, these were regarded as patterns of ties, without worrying too much about the nature of these ties, or about the differences between various “types of ties”. The blockmodel papers exhibit this basic stance of social network analysis:

The cultural and social-psychological meanings of actual ties are largely bypassed in the development. We focus instead on interpreting the patterns among types of tie found in blockmodels. Our sole assumption here is that all ties of a given observed type share a common signification (whatever their content may be).

The first edition of Identity and Control (1992) is White’s attempt to fill this void, to reflect on the meaning of (different kinds of) social relationships and network patterns, and on the processes in them. There he lays out a grand scheme of ideal types of social structures, including ties and networks, catnets again, “disciplines” as different kinds of social organization, institutions, and styles. Again, he draws on a wide range of empirical studies from anthropology and organization science to his own studies of markets and of the field of arts. Also, like before, he eschews sociological theory. The only theorist picked up on a number of times is Erving Goffman—with concepts like “footing”, but also with his empirical studies of everyday situations. Like Goffman, White puts his trust in empirical observations, rather than traditional theory and concepts.

However, White does pick up on abstract concepts in Identity and Control, and in the articles (published and unpublished) from the 1990s and 2000s. He takes most of them from other disciplines. For instance, he writes of social networks resembling the long molecules in “polymer gels” that “reptate through messy, inhomogeneous environments” (Baecker, 2025; White, 1992, p. 70), and of management shake-ups as “annealing” (1992, p. 281ff), picking up on physical chemistry. He conceptualizes social networks as composed of “stories” that link “identities” (1992, p. 5ff; 66ff), with terms from linguistics. The fascination with linguistics also leads to the processes in networks called “switchings” (Fontdevila, 2025; White, 1995). These new terms are often little more than suggestive metaphors, rather than well-defined concepts. For this, they would need clear definitions and their relations to other terms fleshed out. For instance, we do not quite learn from White what exactly a story is, or how it makes for the ties and types of ties observed in network research. Also, it is not clear what kinds of interaction constitute “switching”, and which ones do not, if any. ⁽¹⁾

The result is an idiosyncratic combination of different strands of literature, a “creative conceptual mess”, which opened up new ways of looking at social structures and the processes in them. White broke up old ways of thinking, pointed to conceptual gaps in sociology and network research, and offered fresh concepts and research questions. However, this also left many of us bewildered, sometimes even taken aback by this conceptual jumble—fascinating, irritating, and incomprehensible. It was impossible to fully grasp this theory bomb. Most of White’s disciples and students, including myself, adopted his framework only in parts, and often enough more implicitly than explicitly.

There were certainly alternatives for this “creative mess”: Symbolic interactionists had by then offered conceptualizations of social relationships and networks as patterns of meaning (Fine & Kleinman, 1983; McCall et al., 1970), and of identities as projection points arising in interaction (McCall and Simmons [1966] 1978). There was a budding research literature on social relationships (Duck, 1990), mostly hailing from psychology and offering both conceptualization and empirical studies (White’s treatise curiously missed empirical work on relationships). Exchange theory offered a plausible account of social networks based on the idea that exchange is based on obligations, and gives rise to them (Cook & Whitmeyer, 1992).

All of these approaches embrace that social networks consist of patterns of meaning, without White’s unwieldy concepts. They could have been combined with White’s ideas, as the next generations of relational sociologists did (Crossley, 2011; Emirbayer, 1997; Emirbayer & Goodwin, 1994; Erikson, 2013; Fuhse, 2022). Many of them resorted to the notion of “expectations” as key for relationships—with “stories” about events feeding into expectations, but with events bringing about expectations between ego and alter even without stories told about them. This makes for a more accessible conceptualization of social relationships that resonates with everyday experience. It also curiously ties network research back to the work of Parsons, who (building on Weber) conceptualizes social structures as patterns of expectations (Parsons et al. [1951] 1959, p. 19f). However, expectations remain fundamentally unobservable in empirical research—we can only infer them from observed behavior. Stories, in contrast, are told in communication. Here, as elsewhere, White prefers concepts to denote something observable, rather than mere theoretical constructions. This might also be the reason why he rejects subjective meaning and motivations as theoretical constructs, against the grain of the action theory, pragmatism, interactionism, and critical realism.

Things changed a bit with the papers published in the late 2000s and early 2010s, and with the second edition of Identity and Control (White, 2008). Partly through the influence of colleagues, partly through his own changing interests, White now engaged more with other strands of sociological theory. In particular, he and his co-authors picked up on Niklas Luhmann’s notions of meaning and communication and on his theory of functional subsystems, and on Pierre Bourdieu’s theory of social fields (Fontdevila et al., 2011; Godart & White, 2010; White, 2008; White et al., 2007). He also filled some of the gaps in his theory, quoting Charles Tilly at length on “standard stories” (White, 2008, p. 29f), and offering a list of four possible meanings of “identity” (2008, p. 10). In addition, he picked up extensively on the empirical work carried out by students and colleagues to illustrate his theoretical concepts. The theory thus becomes more accessible and acquires more depth. But it also branches out in the areas of macro-sociology charted by Bourdieu and Luhmann.

The above discussion already pointed to a few peculiarities of White’s way of theory-construction: the shunning of previous sociological theory (for a long time), the strong connect to empirical observations, with theoretical terms based on these observations, and the avoidance of unobservable theoretical constructs. Together, this follows the idea that theory springs from empirical work, to systematize and integrate it, and to guide future empirical studies. This may sound reasonable, but it is a textbook case of an epistemological approach most social scientists reject by now. My claim is that Harrison White’s theory, probably unknowingly, is a prime example of logical positivism. To my knowledge, he never systematically engaged with philosophy of science, or outlined his own position. ⁽²⁾ So I will have to argue based on the development of his theory as sketched above.

This approach resembles the “positivism” Turner (1985) advocates, building on Auguste Comte: The task of social science is to assemble positive knowledge about the social world, based on empirical observations and their generalization. But the prime reference point is the philosophy of science school of logical positivism, springing from the Vienna Circle around Moritz Schlick, Rudolf Carnap, Herbert Feigl, Carl Hempel, and others. However, the writings of the original Vienna Circle are more complex and less united than typically presented (Friedman, 1991). The well-known position of logical positivism results at least as much from the sympathetic, but highly selective recasting by Ayer ([1936] 1990), and from the sustained criticisms by Karl Popper, Ian Hacking, and a diverse group of “post-positivists” as from the writings of the Vienna Circle. As in the theory of White, the identity “logical positivism” derives from the story-telling in academic discourse. Here are the main points of this position:

1. Logical positivism is thoroughly empiricist. That is, scientific knowledge is measured against empirical evidence, and nothing else (Carnap, 1956). Knowledge is seen as a logically consistent structure of symbols that represents the world in the form of abstract propositions (Blumberg & Feigl, 1931; Wittgenstein, 1922). These propositions are to be tested in rigorous empirical tests, and either verified or discarded. This leads to positive knowledge about the world, as Comte envisioned, even if it always remains fallible and provisional (see 6.).

White’s theory is empiricist throughout, with its many examples from empirical studies illustrating the concepts and theoretical arguments. He rejects “contributions [to science that] could not be overturned by any conceivable data” (White, 1980, p. 201). At the same time, he argues that “persuasion about definite features of the natural world is a distinct goal” of science (White, 1980, p. 199). Identity and Control and the papers from 1990 onwards entail many empirically testable conjectures without highlighting them. The first edition lists 27 conjectures in Appendix 1, suggesting that the theory is designed to formulate them. Some of these conjectures look hard to test, but their listing and the term “conjecture” leave little doubt that White intended them to be testable. The subsequent 19 speculations and 54 questions aim at elaborating and extending the empirical scope of the theory.

2. Positivism is logical insofar as theory construction builds on the non-empirical sciences of mathematics and logics (Blumberg & Feigl, 1931, p. 282–285; Carnap [1928] 1967). It is probably impossible to construct a fully logical consistent sociological theory of the extensive scope of Identity and Control. But White obviously builds on his formal training in theoretical physics in his venture, deriving surprising theoretical ideas from the combination of arguments. His theory contains less mathematical formulas than most works in rational choice. However, he strongly identified as a mathematical sociologist (personal communication), and emphasizes the usefulness of mathematics for sociological theory (White, 1997).

3. The flipside of positivism’s empiricism is a vehement rejection of metaphysics, that is, of theoretical speculations not grounded in empirically observable facts (Ayer (1936) 1990). For instance, the grand theoretical schemes of Marx, Weber, Parsons, and Luhmann, of the Frankfurt School, also Durkheim’s considerations on the division of labor and organical/mechanical solidarity defy empirical tests. Positivism holds them to be “meaningless”—the meaning of a theoretical proposition consists of the conditions under which it is true or false.

Again, this mirrors White’s stance, which starts from the rejection of Parsons’s structural-functionalism. He contrasts science with “rhetorics” as “its chief competitor” (White, 1980, p. 199). We can interpret rhetorics here in terms of the metaphysics derided by logical positivism: abstract verbose arguments that are not rooted in empirical observations: “rhetoric is the fashionable way of selling one’s nostrums” (White, 1980, p. 202). In the 2000s, White picks up on the theories of Bourdieu and Luhmann, with the latter containing a fair amount of metaphysics. It is not clear to what extent this changes White’s overall stance.

4. The logical positivist strict orientation to empirical research and rejection of metaphysics corresponds with an opposition to normative and ideological arguments. Science, including the social sciences, aim at finding out facts, rather than reason on what could or should be. This conforms to Weber’s postulate of value-neutrality of sociology ([1919] 2009). White’s theory clearly adopts a neutral, disinterested standpoint, shunning normative arguments and evaluations. In 1980, he rails against “ideological cooptation” of academia (1980, p. 201). He rejects the feminist idea that women hold a particular standpoint giving us important insights, venturing that differences between men and women in their epistemic orientations are bound to disappear. His formulations are far from naïve though, and rather interestingly phrased:

How are views of science evolving on the part of that very large body of persons known as women? [...] As persons who are women become known as persons of other sorts, the question will/should evaporate.

First, he challenges that women should unquestionably and primordially be regarded as “women”, rather than as persons in general. Second, White argues that “the female question” at least with regard to the belief in science should “evaporate” as more women become “persons of other sorts”, presumably entering formerly male dominions like politics, academia, industry, medicine, and commerce to be seen as politicians, professors, engineers, medical doctors, managers, and bankers. White here clearly advocates the universalism of logical positivism, as opposed to the particularity of feminist standpoint theory (Harding, 1992).

5. The empiricism and rejection of metaphysics leads logical positivism to reject purely theoretical entities (Hacking, 1983, p. 48f; van Fraassen, 1980, p. 56–59). Instead, scientific concepts are to focus on what is directly observable, or what needs to be inferred to make sense of relations between observable entities. Theoretical constructs like “unconscious”, “culture”, and “habitus” come under scrutiny: Are they really necessary inferences from observations, or should they be rejected as “metaphysical” or “rhetorical” (in line with Occam’s razor)?

White’s theory displays the same disdain for unobservables: Most concepts in the first edition of Identity and Control (1992) stand for directly observable aspects of the social world: ties, stories, and identities; or they are summary terms for observables, for example, domain (of cultural forms), network, style, rhetorics, and career. White adds inferred terms to make sense of regularities of observed behavior (or, one could argue, they summarize observed behavior): institutions, coupling/decoupling, and the three types of disciplines. Finally, there are very few concepts further removed from direct observation: uncertainty (ambiguity, ambivalence, ambage) and control. I suspect that White held them to be observable, too, but we cannot know for sure. All of this marks a positivist focus on empirical observations as the only source for theoretical concepts, and as their chief targets. Again, things change a bit in the second edition, with White picking up on Bourdieu’s social fields and Luhmann’s functional subsystems with his concept of control regimes. At the same time, he adds more observables like switching. Overall, White shows a clear positivism in his direct connection of theoretical terms to empirical observations.

6. Logical positivism by and large adheres to the principle of induction: Theoretical concepts and arguments are supposed to spring from concrete empirical observations by way of tentative classification and generalization. This requires a certain trust in empirical observations, which are supposed to be true to the phenomena and not biased from theoretical preconceptions (Carnap, 1956). The basic, raw observations are then translated into theoretical languages, which always remain fallible. In this vein, White’s theory is inductively built up from empirical observations, with a mistrust of the previously available concepts and theoretical approaches in sociology (though selectively drawing on concepts from other disciplines).

The issue of induction, or of the method of constructing theory in relation to empirical observations, has been extensively discussed in the philosophy of science. Karl Popper argues that building scientific theories through induction is impossible, not least because there are no pre-theoretical observations (1963, p. 59–72). Observations are always selective, they build on preconceptions of what to look for, and how to conduct them. Instead of going bottom-up from observations to theory, Popper suggests going top-down: We logically deduce propositions from theoretical assumptions, and then submit them to empirical tests. If empirical observations match these propositions, they are not verified (as in logical positivism)—that would be impossible. Instead, they are “not falsified” and provisionally accepted as true. The assumptions themselves are untestable. Rational choice follows this model of the scientific enterprise: Human beings are assumed to strive for the maximization of their subjective utility with their actions, and this leads to concrete propositions for how individual actors behave in particular situations.

In principle, I agree with Paul Feyerabend that we can arrive at good theory by any method (Feyerabend [1975] 2010). However, both induction and deduction suffer from the idea that either theoretical reasoning or empirical observation could proceed without the other. This idea has been refuted in successive waves, but decisively in the 1950s (Hanson [1958] 1965; Sellars, 1956; Quine, 1951). Quine points out that all scientific sentences carry both “analytic” (theoretical) and “synthetic” (empirical) implications. It is not possible to observe pre-theoretically, pace Popper. But theoretical reasoning is similarly always already informed by empirical observations. As a consequence, the strict separation between assumptions and conjectures does not hold: Assumptions start from empirical considerations (like the rational choice idea that human actors try to maximize their well-being), and may become subject to empirical tests with the right tools.

Also, following this principle, “metaphysics” cannot be discarded and rejected straightaway, as logical positivists and Popper would have it. All metaphysical speculation somehow rests on empirical observations, even if they are not methodically controlled (and also subject to theoretical pre-conceptions). Certainly, not all theory connects to empirical research to the same extent. However, even abstract theoretical schemes like those of Parsons and Luhmann are not wholly disconnected from the world of observations, and may yet become useful for empirical research.

What do these last deliberations mean for White’s theory? First, his disdain of previous sociological theory is misplaced. While White believed to work inductively entirely from observable facts, these are already shaped by theoretical preconceptions. The idea of theory-abstention, heralded by Geertz ([1973] 1993, p. 24–28), fails to realize: All observation starts from theoretical ideas, from concepts that researchers hold about their subject area. In White’s case, the theory builds on observations guided by the “structural intuition” of social network research (Freeman, 2004, p. 3): That social phenomena can and should be observed with regard to patterns of social relations, and that these networks are key for social behavior. Also, the concepts brought in from other disciplines do not come without theory: roles (from anthropology), stories, narratives, identities, etc.

As I argue above, there would have been alternatives from sociology—from symbolic interactionism, exchange theory, cultural sociology, role theory, and the sociology of organizations, even Parsons and Weber. These would have connected White’s theory to previous research strands, and made it a lot more accessible and acceptable in the discipline. The more recent generation of network theorists pursues this path, including myself (Crossley, 2011; Emirbayer & Goodwin, 1994; Erikson, 2013; Fuhse, 2022; Lin, 2002; Martin, 2009). On the upside, White gave theoretical reasoning on networks a fresh start: He brought in unique concepts and ideas building on other disciplines, and on his curiosity about a wide range of research on social structures. Key concepts like identities and stories, the intertwining of networks and cultural domains, and the role of linguistic forms and communicative process (“switching”) in networks are bound to shape our thinking about networks over the coming decades. As they make their way into more accessible accounts (by Emirbayer, Crossley, Erikson, and others), they lose some of their iconoclasm and idiosyncrasy, but they are still irritating and disruptive in the best possible sense: Here are some new ways to think about social structures, which lead to new ways of carrying out network research. That is no small feat, and we have Harrison to thank for it.