Interview with Harrison C. White

This interview was prepared by the two authors and conducted by Jan Fuhse in March 2011 during a research visit to Tucson, Arizona. We addressed a few questions that had come up while writing a German language introduction to White’s theory. That book contains a translated and abridged version of the interview (Schmitt & Fuhse, 2015, pp. 177–186). The text here is again abridged and smoothed over, in a slightly different version for an international audience. We have added section titles and a few references to point readers to the works talked about.

Question: Let us first talk about the historical background of your work: You came to Harvard University when it was dominated in sociology by Talcott Parsons and his structural functionalism. Is your structuralist approach a direct response to his grand theory?

White: Well, it was actually both Parsons and Homans. They were both important and very different from one another. And I was very uncomfortable about Parsons, but I did know his grand theory. It just seemed so grandiloquent to me; it did not seem like science. One of my troubles was that I had been trained for 8 years at MIT in straight physics, and you cannot just give it up. I worked with Karl Deutsch, a political scientist who shaped my thinking a lot. And he is the one who got me to actually consider going into social science. And he knew Parsons, but he also had a modern and data-oriented, operational, pragmatic approach, which was much more sympathetic to mine.

Question: But Deutsch was not a network analyst. Do you remember what got you interested in formal network analysis?

White: I do not remember any transition. I think I took it for granted, more or less. Before I went to Harvard, I did a year in sociology, and then, I actually went into operations research. Apart from stochastics, I had to do some network analysis there. But I never really looked at myself as a network analyst. I was trying to solve problems, find out how social life worked, and use networks or stochastic modeling, or whatever was good.

A more important influence on me than Talcott Parsons was teaching an introductory course at Harvard called Social Relations 10 (White [1965] 2008). At that time, the departments of anthropology, sociology, and psychology were joined. I wanted to do it my own way and start from scratch. I actually used more Homans because a lot of Homans is very, very useful and very believable. He had done some important work on small groups (Homans, 1950). My first book was on Australian kinship. And that was both linguistics, but also networks, kind of institutional networks.

I knew Karl Deutsch’s work very well, but I never really thought of it as networks. The problem was to understand how the organization worked. So, I spent a lot of time teaching organization theory. And it seems to me a very healthy thing for sociologists to actually want to solve something. Find a problem. Solve it. Get something working. So, a lot of the way I do sociology came out of teaching, especially graduate students at Chicago and undergraduates and graduates at Harvard.

Question: Coming back to the contrast with Parsons, his theory centers on the social construction of order. In your work, change and deviation play a much larger role. Would you agree to this opposition of perspectives, stable order versus dynamics and messiness?

White: Well, I now agree to it. That was not the way I started off. I came into this through deviance and disorder. Life is a mess, and social life is particularly messy. While I was at Harvard, I spent a lot of time with the business school there because, after all, my first job had been in the business school, and I was particularly interested in organizations.

What I regard as the only real level of sociology: the meso level. The micro level is just too damn micro. And the grand theory level is not good. And so, how can you get somewhere? Get some purchase, get some results, and so on, yet still do justice to how messy it is, and how intricate. And one way you do that is by shifting gears. Like I worked seriously in sociology of art. I worked seriously in organization theory. You really got to do it in a detailed way. And that is one of the first things that tells you that life is a mess, and there is no one way to do it.

Question: Can we talk about James Coleman’s theory? Coleman demands the explanation of social processes on the level of individual action and a macro-micro-macro transformation. What is your stance on these positions?

White: I thought it was a waste of time. Jim Coleman was brilliant, an absolutely brilliant first-class sociologist. I just thought he was wasting a lot of time and effort with this ideological stuff: rational choice and micro-macro. That does not get you anywhere that I could see. He did a lot of very good specific studies. But then, his Foundations [of Social Theory] (Coleman, 1990) – I have never read it carefully because I just did not see the point. So, Coleman’s work is important, and he also is a mathematical sociologist like me.

Now I do not think of myself as a mathematician. I may use that for buzzword purposes and to explain to people that mathematics is gorgeous and beautiful, if you can make it work for you. But it does not do anything, you have to make it work for you. Coleman made it work for him, and he was one of the pioneers in stochastic modeling. Life is messy, so you need stochastic modeling.

Question: Now, we have a few questions about your work at Harvard from the 1970s. Your work on structural equivalence (Lorrain & White, 1971) and blockmodel analysis (Boorman & White, 1976; White et al., 1976) constitutes a crucial breakthrough in network analysis. Blockmodel analysis inductively identifies patterns and networks because we cannot know beforehand what these are. Is it possible to test hypotheses with blockmodel analysis? Or can we only explore and describe network structures?

White: It is possible to do hypothesis testing, to build theory, but it is not easy. You have to know where you have to start exploring. And then you hope you get lucky and get some kind of pattern that you can make sense of and then test with blockmodel analysis. But let me be very clear: blockmodeling was an interactive accomplishment of a number of people. The young people I was working with, Ron Breiger and others did just excellent work.

The important thing is: What is the phenomenon? What is the problem? How are you going to make sense of it? And of course, you have to spend part of your time looking for cases. I do not think I got very revealing results from networks with the Impressionists. If there were some place guaranteed to be successful for blockmodels and networks, it would have been the French Impressionists as a well-defined group of people. We know what they did ad their relations to each other, and still it did not work. But that is where I found out about institutional analysis (White and White, [1965] 1993). That is where institutional systems come from for me.

Question: According to the idea of structural equivalence, networks are patterned by role categories. Do these categories arise in the networks or can they also come from outside, such as gender, ethnic descent, or formal roles like professor and student?

White: Of course, they can come from outside. It is not a one or the other. That is why I love to read all kinds of studies in different fields and different scopes because that is what you need to do. You need to get new ideas and new perspectives and not get trapped in some stereotypical formula. But I did a lot of different things. Partly, it was because I got interested in a lot of things, and partly, it was deliberate because you cannot do sociology in some neat, tidy framework.

Question: Brint (1992) argued that blockmodel analysis already implies a focus on meaning. Networks are structured by role categories. Do you think that your book Identity and Control (1992) continues and completes the approach envisioned in your earlier work, or is it a sharp break with earlier structuralism?

White: First of all, it is not complete, and it would not be completed for a long time. But it certainly continues my earlier work. I never had a feeling of shifting gears. I think Brint is right.

Question: In what way was your “cultural turn” to meaning in networks influenced by Pierre Bourdieu?

White: I came across Bourdieu in teaching graduate students, and he just does damn good sociology. You do not need to think of him as having a theory. I never thought of myself as having a cultural turn. Maybe you are right, maybe there was a cultural turn in the sense that I was not as conscious about culture. But I would not say that there was a break from what had gone before. I think, culture and meaning were always latent there. And it was just a matter of being able to finally get a place.

Question: We have a few more questions on Identity and Control. The title implies that identity and control are the two basic building blocks of social structures. In what way are identity and control more or less foundational than other concepts, e.g., networks, stories, interaction, communication, switching, and meaning?

White: It is a judgment call, a matter of taste. I do not remember exactly when I came up with the title. I never thought that the network concept had sufficient theoretical potential. Networks are very important, but what is going on in the network? Identity and control are what is going on in the network. What is going on in institutions? Identity and control again. Stories and story sets are byproducts of networks, as identities and controls build a social reality.

It is ridiculous to have those two individual words do everything, but they seem to me a reasonable basis and they go together. Identity is not control, and control is not identity. Out of the two of them, you can build disciplines, and you can build networks.

Question: What are the main differences between the first and the second edition of Identity and Control (White, 2008)?

White: I started off with disciplines as the foundation [in the first edition]. And then in the second edition, I moved to networks. Or maybe I moved back to networks because I originally started with them. I was unhappy with the mechanistic way that the network concept was used. Often, it is kind of technical: you draw lines. Even if you put in movement, it is still very mechanistic. There has to be more than that: meaning and communication and things going on. And so, in 1992, the discipline concept seemed a more fitting foundation because it was bringing these other things in.

But then in the new edition, I think it makes sense to go with networks. Once networks have been loosened up and you realize that they are not these mechanistic things, the concept is capturing a lot of what you need. Among other things, it is the open-endedness. If you start with networks, then the boundary is a foundational problem. The disadvantage of disciplines, at least the way I conceptualized them, is: Although they claim they are not bounded, they give you that feeling of being molecular. The advantage of disciplines is that switching comes more easily out of disciplines than out of networks.

But the thing is, you want to get somewhere. You do not want to just write very complicated books. Switching is all very well to say, but it needs to be observed. That is a very complicated thing to do. And one of the troubles I have been having with language is: every time you want to observe linguistic matters, it gets very complicated. All kinds of things come together at once. And yet, to get results, you have to simplify them.

Question: In the theoretical developments from the first to the second edition, how important were the studies of Mische (2008) and Gibson (2003, 2005)?

White: They are very important. Style has greater importance in my thinking now and in the new book. And I think a lot of that is due to Ann Mische. I think of her advances in network switching as a modus operandi, which is that social things get accomplished. A lot of that comes out of work with Ann and David. I am nervous about making individual comparisons because all these people have been great and helpful. But Mische and Gibson were certainly important and important to this particular phase and change. Now, who are the others who are important? Well, I think, capital city journalism is for me.

Question: Mützel’ s (2002) work?

White: Yes. It was a significant piece of work. It helps you see things. And there is Frédéric Godart, important in several ways. First of all, he is a beautiful writer and very clear thinker. And so, he was great in an editorial role. But he also has ideas of his own and his thesis, on fashion as an actual functioning stylistic area. All of these people were contributing ideas. One of the things that makes it tricky is it is often in a group context. That was also true back in the old days of networks and block models. I always get uncomfortable when people say it is Harrison White’ s thing. It was always done either in a team or a group of people.

Question: We now have a number of questions about switching and communication because they are an important research interest of ours. Let us turn to the basic social processes in networks. In Identity & Control and in the recent articles with Frédéric Godart, you call them interaction. And in the systems theory paper from 2007, we pick up on the concept of communication of Luhmann (White et al., 2007). We also point to switching as the communicative processes leading to the emergence of meaning. Could you relate these three concepts: interaction, communication, and switching?

White: To me, interaction would be the least interesting. It is kind of a word out of the dictionary. To me, the crucial ones are switching and communication, and they are different sides of the same thing. Communication is switching. If you have this idea of a flow going through a pipe, that is not communication. Now, that is why I do not think I am that far off Luhmann.

Question: Would you say that switching is a special instance of communication?

White: No, I would say communication more likely constitutes a special kind of switching. Switching is a more universal concept. Indeed, I have been playing with the idea of stitching instead of switching, I could use stitching. The main thing to get clear is that there is a rapid, you know, unexpected, unpredicted, uncontrolled change. And it is a change in both the perspective of who is seeing what and of context and what is involved. And so, switching has got to be at the very heart of everything. ⁽¹⁾

And that is why style in this new version is now, the macro or at least a larger scale version of switching. And indeed, I argue that a style, in effect an identity can be from switching, you know, that is if you like capital city journalism. Though I am not sure whether that is how she was thinking, but the way I read it is that I am showing my style as a sign of vanity, switching and communication.

It It always makes me nervous [when people] bring in cognition. I want to get away as far [away from that] as possible. It is not that I am against cognition, but I do not do it. At least I recognize cognition. Luhmann brackets it, which means that he makes it believe like bullshit. What are these communications that are keeping themselves sorted out in the economy and the political code and so on? He has no mechanism. He has no one doing anything. I think he had great intuition to make communication the central idea. But to me, communication really is switching and things like style and so on.

Question: You define switching as the change in conversation between different netdoms (White, 1995). That could mean a switch between networks, as when a node or a relationship, or a node or relationship, skip from one network to another. Or it could refer to a switch in domain, as when the conversation changes from professional to private topics. Would both of these be instances of switching?

White: Yes. Switchings and stitching are a mix of these. They are not purely this or that.

Question: Would you translate the concept of switching into Goffman’ s theory of framing interactions as the switch from one frame to another?

White: I think so. Goffman’ s concept of frame would include both network and domain aspects, so it sounds reasonable.

Question: So, are there non-linguistic ways of switching between netdoms, for example, when somebody enters a room, or when a phone call is interrupted by a broken connection?

White: Sure. You mentioned two of the main ones.

Question: And how exactly does switching produce new meaning?

White: They do not so much produce new meaning. Rather, they embody, instantiate, and are new meaning.

Question: So how does switching embody new meaning?

White: Well, it is by definition. If you are switching both the domain and the partners, that is something is going on. It is not the same.

Question: Why cannot communication, for example, in friendship, produce new meaning without switching between netdoms?

White: Show me an example. Any actual example you look at, it will turn out. It would not be like that. For one thing, in real life, we are embedded in a context. We are looking at the windows as someone comes in and another thought enters our mind.

Question: Well, two partners would be talking about a conversation from the day before, and then, they would share ideas and come up with new ways of thinking about it. Would that involve switching?

White: Or as I sometimes think I should have said, twitching is even better than switching. The truth of the matter is, one partner will have her sister or her brother in mind, or they will have other figures, even if they are not explicitly there.

Question: So these other relationships form part of the context of what is being instantiated as meaning. Would you claim that switching induces change, whereas communication that does not switch tends to reproduce existing sociocultural structures?

White: I mean, there are two answers: that first is a plausible every day kind of statement. The second is: What is the point of it? I do not believe it in any significant way.

Question: So, there would not be any communication that does not do switching?

White: And why would I be interested in asking that question?

Question: Maybe we also want to account for stability and not only for the change of social structures.

White: Oh, the stability will account for itself. People are busy doing it all the time.

Question: So, how is it possible to discern switching empirically? How should we study them?

White: I think your best bet is to get on the email, to get on the Internet, and to study it there. I mean, it is just such an obvious thing to do. For example, at Columbia University, people are doing that because you have just the right kind of context to do it (Kossinets & Watts, 2009). There is communication, and other communications follow from it. It is a way of doing what David Gibson did without so much hard work. You do not have to sit in and you do not have to take detailed notes. And that is one of the reasons Gibson’ s study was important. He had the bravado to actually sit down and do it.

Question: Our next questions are about the story and the tie and about the network and the domain. You define domain as the array of symbolic forms in the network and claim that both are only analytically distinguishable. How does the notion of domain differ from that of culture?

White: Well, they are in the same ballpark. But a domain does not have as much freight attached to it.

Question: So, networks are interwoven with domains, just as ties are interwoven with stories.

White: It is wrong to think of ties or networks as separate from domains. That is why I called it netdoms.

Question: But they are analytically distinguishable.

White: Yes, analytically distinguishable. But if you want to make any kind of empirical statement, you have to do them both at once. Now, the very word culture makes me nervous because it is been used so heavily in so many ways, so many complexities. To the extent you want to use culture in a pretty straightforward sense, I do not see a problem.

The main difficulty, of course, is that you cannot possibly have a culture in a domain. Culture is only activated and becomes real when there is both domains and networks interacting and moving on. And this is the classic example of that because you cannot talk about the story and the tie and the network and the story domain without putting them in a situation. There would not be anything without a situation. So, you have to start with the situation. Then, when you do, that is not just a tie and a story, it is a context.

Question: If we understand you correctly, stories can be dyadic, for example, the story of a love relationship. But they can also relate multiple identities and, therefore, connect them meaningfully in networks rather than single dyads. What would be an example of such a multiplex story? A story that links, like a story of relating the members of the family to each other, or the members of a department.

White: I think it would be more useful to go back to small group research and look at casual pick up groups, which are perfectly real, and they are important to people. But they are not engrossed in kinship and all that complexity of culture and internal family dynamics.

I just think it is very important to try to find transparent, either very transparent or very obscure, places to do your empirical work. If it is in between, you get lost. But if it is either the web, you know it will be very clear, or if it is like what is going on in North Africa at the moment [the Arab Spring]. And of course, it is up in the air who the nodes are and all that kind of stuff.

Question: Can we also talk about the identities of a tie? Does a friendship, for example, also have an identity with the five senses of identity that you mentioned?

White: Well, the difficulty is that you are trying to wash out the history. I mean, the actual tie is a historical development or achievement.

Question: But so is an identity, isn’t it?

White: Yeah, but they are not the same thing. Then, an identity is moving back and forth with other identities in a way that ties are not because ties are confined by analytic definition to their own kind.

I mean, what we need is usable, may not be perfect, but usable ways of framing, asking, and answering questions. And I do not see that most of these questions are questions you really want to ask. When you are asking them, you know, like a schoolteacher, you are saying, well, you know, if we are going to do this and we must know the boundaries of the following category, you do not have to know that.

Question: Well, in our book [on White’s theory], we try to clarify the categories and their relationships to each other.

White: But is that clarifying categories?

Question: Well, the attempt of relating categories meaningfully to each other, to a coherent framework – that is what we call theory, isn’t it?

White: No, not at all. A theory is an actual account of what is transpiring and what is going on. That gives one a sense, maybe not completely correct, one can predict what is going to happen, or one understands what is happened, and it does not require at all that you understand these boundaries. In fact, it might hurt you to understand these boundaries and to clearly understand the boundaries. Additionally, that gets in the way with your ability to just work with the material.

Question: That is an interesting point. I would disagree, but we will leave it at that. Is it possible to separate domains from institutions and control regimes, or do these concepts overlap?

White: Well, they overlap from domain to control regimes. I would say the institution is furthest away. What you want to look at is discipline, control regime, and domain. Now you cannot have a discipline except as an interaction of entities struggling for control in their interaction. And my claim is that the three kinds of disciplines are fundamental. And so, if you look beneath what is going on in organized things, there is clearly networks.

I would say the control regime is up levels. Since a discipline, by its nature, covers two layers in a context, a control regime can be any number of layers. And that is the reason I never claim the number of different types of control regimes. There can be many different kinds. I give, I think, good examples, interesting examples of different ones, but it is not like disciplines. You do not have any kind of molecular feeling at all [in control regimes] that I can see.

Question: We have a few questions about the style concept. Styles are signals for the observation of identities, but they are also profiles of switching. How do these two sides of the style concept relate?

White: They go with each other. When you see a style, this is going on. I mean, a style comes close to being the dynamic profile of an identity. It is more like people composing fine music, you know, advanced music. Now, when you talk about what musical style is, I am not greatly expert at that, but there are clearly a lot of them, almost an indefinite number. And how you tell them apart is not so obvious. What is obvious, though, is that, at least to me, the heart of music. You know, if you do not have a style, you do not have music, you know you have some noise or some notes on paper or something else.

Question: So, what can we actually explain with the concept of style?

White: We can explain what is self-consistent, able to keep itself going, and reestablish itself. And both of them and the other styles have some sense that they fit this style. You may not like baroque music, but you will recognize that that is a style, and it is really all it is about. Style is so important because it is there the whole time.

I really want to keep the concept tied to things that one would then go out and study, which is not easy. It is one hell of a lot of work to study any of these things. That is the reason we make these concepts simple and clear for our needs as analysts.

Question: We are now turning to the future or to the present of relational sociology. What do you see as the most important tasks for the theoretical foundations of sociological network research?

White: I would not know until I see it. The most important task is to read what the young people are doing. And until you find someone like a young professor who has come up with new ideas like Gibson.

Question: You mentioned the unfinished project of a book on language and networks or on meaning and networks. How can a sound theory of language and meaning contribute to sociological network research?

White: Good question. I have written a couple of papers trying to give examples of how it might happen, things like indefinite pronouns, and then Luhmann himself has done a lot with it. He has not explicitly applied it, but it could be quite suitable for that. But that is too hypothetical for me. I want to get some young men or women excited to do something. There are a lot more people out there than we know. Some of these people will start using it, and they will get results.

Question: In what way should we change our prevailing methods in social network analysis? Should we turn more toward qualitative interpretive methods?

White: I do not see why. I think it is based on a misunderstanding: you cannot get anywhere with one method. Whatever you do, you will want to do more than qualitative methods. You know, this whole idea of qualitative and quantitative is rather arbitrary and ad hoc. Qualitative just means that it has not been done. And so, I do not think they are that much opposed to each other. You want qualitative methods for nuance and shading of context, and you can do that quantitatively, too.

Question: Would you insist that qualitative data be coded in order to be analyzed quantitatively, as in the work of David Gibson?

White: I think it is prudent. If you have got qualitative data, which means that you have got some stories there or some collections of words, what are you going to do with it? Well, there are all kinds of things to do with it. But certainly, one reasonable way to try is to see if you can get some kind of metric account.

Question: You frequently insist on contextualizing context. But in sociological network research, we can only observe the immediate social context by bracketing the outer world and, in a way, artificially draw a boundary between the network to observe and its wider context. Is it possible or necessary to bring this wider context back in? In theory and in research practice?

White: I think so. It is happening now with Douglas White at Irvine. In addition, John Padgett and some of his colleagues are doing that partly because they are enabled now by the power of new levels of programming and software. It is amazing what you can do. If you can run 100,000 iterations, you can get a lot of things pinned down.

Question: Would you see a value in working with the grand theory of society as the one developed by Luhmann, and also, in a way, by Bourdieu and his field theory, in order to locate networks within their larger social settings?

White: I think that makes a lot of sense. I have certainly tried to do that myself. I had not explicitly put it within their theories, but they are certainly very much in my mind.

Question: In what way do you want to provide a similar grand theory with the concepts of disciplines of control, regime, and institutions? A grand theory potentially closer to empirical research by providing testable or at least empirically observable theoretical expectations?

White: You put it in a way that I find a little funny. It is not that I do it. It is reality that does it. Networks are embroiled in this larger thing. And I think having constructs like these are helpful in coming to terms with that because you cannot do everything at once.

Question: What do you see as the most promising projects in relation to sociology? Who will push relational sociology forward and how?

White: I will be eagerly awaiting to see who it is, but I do not know. And I am not up to date enough, or working hard enough, or involved enough to be a good judge. In fact, I try to stay out of judgments now.