Jerrry Kim (Columbia University), winner of the 2007 Lou Pondy Best Paper from a Dissertation Award, interviews Ron Burt (University of Chicago), the recipient of the 2007 OMT Distinguished Scholar Award.

 

JK: Your presentation was extremely stimulating. In many ways, what was surprising was that it represents a pretty significant shift from your early work, especially in the mechanisms of brokerage. I was curious, what led to this shift? Was it a gradual evolution over time, or was there a sudden a-ha moment?

RB: The initial work was an opportunity that Jim Coleman presented, and it was a purely structural play. Relations that contradict one another free you, in a Simmelian sense. No process play. That was through the 1980s and applied to industries. Structural Holes was an exploration of the network idea taken from industries to people. The application to people didn’t sell copy at first. Then it started bubbling up at the Academy (much more than Sociology) around managers doing well. I was giving consulting gigs about this, and perhaps because my background is social psychology, it was natural to start thinking about people. But I knew bupkes about process, and so I just started tracking process more carefully. Especially when I was in Raytheon, just watching people try to be network brokers, watching them fail, watching what they struggled with, it was interesting to me to try to nail down the agency question. I started looking for ways to get an elegant solution to process. It seems inelegant to add personality variables to performance-prediction equations variables, which will get you some traction if you guess the right personality variables, but it doesn’t give you a general solution on how much a person matters. Then I tried this neighbor network approach. I didn’t want to see how little the social matters beyond the immediate network. But the result is stable. The result could be methodological—typically when a social scientist gets results this stable, it’s a fluke of their method, so that’s possible here. But I tried different ways of measuring and estimating, and the result is stable. People will go out and test it saying “No, no, that can’t be.” Already there are papers here [at the meetings] where they say “We get indirect effects in this way or that way”, but I’ve tested those, and I’m not sure it will be general. We’ll see how it plays out. The bottom line answer to your question is that it wasn’t deliberate. Rather, I was just trying to determine the process by which social networks constitute social capital.

JK: It’s ironic that a lot of people are building on your work—evidenced by all the network sessions here at the Academy—at the same time that you’re shifting your view. A lot of the studies seem to build on the global processes mechanism you mention, when in fact you see that as a wrong mechanism.

RB: Not wrong; for closure it’s right. Closure is about controlling other people, so it makes sense that the indirect and third-order ties would matter. Brokerage is really about seeing things differently, so it makes sense that friends of friends of friends don’t matter how you see things. What matters is your direct contacts. You’re doing the same thing with status I think. You’re disentangling the mechanism, by which a relatively crude calculation is shown to have effects. Applications of network power and centrality models was dormant until Joel [Podolny] breathed life into it. In the past it was “who in this community is powerful?”, “Who in this interlock network is powerful?” Jim Coleman has his elaborate model giving rise to the eigenvector measure of status, which is exactly the Bonacich measure. But the model was used as a measurement model, not a theoretical model. It wasn’t until Joel said “Well, this could be about signaling, let’s take it over to the banks” that applications of network prominence became so rich. It’s one of these interesting things where software is offered, and the popularity is often a function of the killer app. Joel brought the killer app. The very next phase is why network prominence, now network status, has the effect it does? I would expect a lot of work around this. And as you clearly show, it’s going to be a really interesting area. In the same way with me on brokerage, the model itself is routine, but why does it work, how does this happen? That’s where we in OMT get our great traction, because we don’t have a platform. We don’t own strategy, we don’t own the personnel process. We devise engines by which organizational context and market context give rise to competitive advantage in one way or another. And then these engines put into models of strategy and personnel bring the models to life.

JK: I thought that comment about engines was really interesting because conventional wisdom has it that OMT theory should start to incorporate culture, different contexts, and make the models more contextualized. But it seems like you’re flipping it on its head and saying that OMT should go in and look at established theories, and we can provide new insights into their phenomena.

RB: Imagine two ways of doing research. One is, I have a phenomenon that I want to explain; What x variables do I need to add to the equation? Much of the research done by contextual people—sociologists, labor economists, a lot of the Academy—is of the order of “here are my explanatory variables, and I added the China variable, the Japanese variable, etc.” I’ve been much more impressed with the medical science model, which is, state what you know clearly, and pull it out of the data, and see what’s left. This Mill’s method of residues, where you confront where you’re ignorant, in theory could be seen as the same thing. But in fact, what happens by graphing the data you don’t understand is that pattern recognition starts to become important. If you have a model that states what you believe, and look at what you don’t know, you’re forced to be clear on something. That’s a real virtue that we don’t have right now. If you’re always adding a qualifier, it leads to sentences like “This might be important” or “This can sometimes be important” It’s not knowledge. It’s a weak conversation. You should be able to put a stake in the ground and say that “this is how it works”, except in these circumstances. And all the models I know that are robust, that build a platform for other people to work on, are models that put a clear stake in the ground, and are clearly wrong in some places. That’s a paper for someone to publish. For example, when Raytheon, which has a massive missile share in the American armaments, deployed missiles in Saudi Arabia, they hadn’t anticipated the extent to which clutter (sand and things) would get in the way. They didn’t realize how much it would screw up their algorithms. The algorithms were right. But when you’re doing it in sand, you do it differently. In the same way, once we get the models right, it’s a causal pressure that should always be there. And then you just bring in adaptations to the situations.  There’s this lovely analogy between how science makes progress and how the humanities make progress. In business and science, you formalize in order to add value with good people who aren’t geniuses. They’re smart, they can see things, and what the organization does is give you a process so that you add more value than you would as an individual. Science, with theory and basic models, you get to create things you wouldn’t be able to create if you were on by your own. The humanities are more about exposing people to perspectives they hadn’t seen before—frame-breaking. We don’t do well with the humanities model in terms of making progress. But I think a lot of our work ends up looking like that because we are a little shy about saying “This is how I think it works”. We are so afraid of being wrong that we don’t make clear statements, and ultimately that limits our contribution.

JK: I take that as a general statement about the field in general, and in the field of networks as well. There’s been a lot of conversation about the state of the field. Some people are very anxious that the field is not heading in a good direction, it’s too methodological, where’s the theory criticisms. I’m curious about your take…

RB: There was a nice blog written after the last Sunbelt discussing this. It’s clear that network analysis is an image that speaks to our constituency, managers and MBAs, and underlies many of our theoretical engines. In the 70s and 80s there was a lot of methodological work in network analysis without purpose. But what’s happened—I think productively in interaction with the Academy—is that network analysis has winnowed out things that don’t contribute to outcome variables of interest. Where’s the value in this network measure? Just having a performance metric to calibrate why a network variable should matter is important. Look at what it added to the eigenvector measures of network prominence, status. Network measures of prominence had been there forever, but they were used to predict themselves, i.e. central people are more powerful. Joel connected network status to performance. That’s a great disciplinary force that leads us to start winnowing out purely methodological measures. On the other hand, I’d hate to see the exploratory methodology disappear, because that’s where the innovations come from. It’s not unlike Silicon Valley, where there’s no real good reason – at the moment -- for a lot of the stuff that’s going on, but then you discover something you can do because this person has worked on this specialty or that specialty. So I’m a little worried that in fact there isn’t enough pointless experimentation in networks. People have argued that the major concepts we’re using were developed in the 70s and early 80s. It’s just all stuff developed then. On the other hand, I’ve always felt that most of the work in the 70s and 80s was invented in the 30s. Moreno was doing his sociometry, and we just gave it a new spin. The cycle could be a cohort thing. We’ll be forgotten, and a later cohort will re-discover today’s models.  But perhaps not, if we can keep to the discipline of where network structure is linked to performance.

JK: In your talk, you presented a wide range of different data points and some might say, why do you have to replicate the same thing over and over? But it seems like that was a source of how you can tease out the mechanisms and that accumulation of data.

RB: It’s interesting how you frame that. You frame it as an N problem—how many. It was much more strategic. The five study populations I pulled were deliberately selected in the following sense. There’s a graph in the manuscript that shows the very densely integrated investment bankers, who are within a couple of steps of each other—the longest step is five steps, with most bankers connected in one, two, or three steps. At the other extreme is the Asia-Pacific software launch in which the longest chain is fourteen steps weaving its way out of China into Singapore into California and then over to the East Coast. The other populations are between those extremes. Friction is much more likely to erode information along long inter-group relationships. By showing that I get same results in relatively friction-free versus friction-full system, I rule out friction as an explanation. By showing the mechanism works across a variety of circumstances where it should be different if these explanations are working, you create a robust explanation. So often now, because of the pressure of publishing, people will publish on a single firm, but the data are highly clustered samples.  They cannot be representative of organizations. The strategic juxtaposition of populations is something that we in OMT are perfectly positioned to do to demonstrate the stability of our results.

JK: Why do you think that is?

RB: I think the pressure to publish. It’s a lot of work. Access can be an issue. I was fortunate because of my consulting activities. I get exposed to a lot of things, and one of the conditions for me doing a consulting project is that I’m going to gather data that are on some issue that I have a question about.  Then you start stapling these things together. It’s absolutely a powerful way to go.

JK: That’s a nice segue into a question I had about your career in general. In your book, you mention that your career is a good example of social capital and brokerage in that you occupied positions in both the business school, sociology and the “real” world. It’s very interesting that you practice what you preach…

RB: But Jim Baron’s observation which quickly follows is absolutely right: “it is like getting two surgeries for the price of one.” You don’t understand the stress of brokerage until you do it. It’s horrible. Dave Krackhardt once said to me, “You know Ron, the thing you don’t talk about is that this is really stressful.” Always confronting people who are pooh-poohing your ideas. I think anyone who does organization work at a business school must feel this. The two people that had tremendous influence on me personally were Nan Lin bringing me into Sociology, and Jeff Pfeffer bringing me into the business school environment. I was a full-time sociologist at Columbia, but Jeff made it possible and plausible for me to shift to the more applied setting of a business school. One of the things Jeff stressed was that you are accustomed to being among people like yourself, and being prominent among them, but when you enter a business school, you are going to be among people that are different than you and you will be seen as a lower caste. At Chicago it’s stark. Maybe as stark as it gets anywhere. On the other hand, you have to love the challenge.

JK: What advice would you give young scholars in the OMT division who wanted to develop similar brokerage positions, integrating new ideas so that they develop the capability to be cutting edge innovative ideas?

RB: I think two things are valuable. The first is not to do it quickly. That doesn’t mean not to do it from your first year, but not to do it without experience. The experience is reframing work. One of the lovely attractions of the University of Chicago are the workshops. It’s an oral culture. The way you add value as audience is reframing what the presenter is doing. Not in a destructive way, but just to see what their argument would look like when viewed through another lens. The exercise of looking at things through different lenses is incredibly valuable for working in a heterogeneous environment. If all you know is a particular way to do things, you will typically fail at brokerage.  We can see it in colleagues who insist that every thing must be viewed through their lens.  For a young person hoping to avoid that sort of synaptic wither, there’s nothing that substitutes for sitting in workshops and reframing work. Not only does that make you a stronger person in terms of doing whatever kind of work you are about, but it adds phenomenal value to the group where you are; so people are heard to say “Jeez, I haven’t thought about it that way.” Every time an idea gets reframed it’s a new paper. The work I presented in the OMT lecture is an example. I began with an innocuous idea: “Lets measure advantage as it spills over from adjacent networks.”  It turned out that the advantage of brokerage did not spillover at all, and we’re into a more social-psychological concept of how brokerage works.  Reframings are how we can be pushed into doing things in new ways.

JK:  You mentioned Second-Life in your presentation. This proliferation of technology has led to a growth in network related services, so we now have blogs, Linkedin, etc. How will these developments impact network research? Will it extend existing research in terms of testing, or will it provide new insights, and how should we go about leveraging those developments in an exciting way?

RB: People say that the internet will get rid of the local groundedness of information. Blogs communicate to everybody yadi yadi yada. I think the point missed is the function served for the user, which leads to the perennial question: the more we are split between different groups the less we know who we are, which means the more prone we are to need sympathetic souls that we think are like us: Such that blogs start to accumulate walled gardens. Such that the Internet is used to verify what I already know rather than expose me to variation. That kind of confirmatory bias has gone on since we lived in caves. I don’t expect the internet to change it. At the same time, the setting for the age-old process of brokerage and closure has absolutely shifted, and inside Second Life I expect to see it play out in a literal way. I expect to see high death rates on ties to people who don’t fit what I’m looking for, who aren’t “cool” like I am. I expect to see an accumulation around people who are cool hoping it rubs off on me. What is different about electronic networks such as one finds in Second Life, is clickstream data as relations are enacted as opposed to panels of sociometric data where we have to disrupt somebody and ask them who are you talking with etc. The continuous time data should allow us to understand processes a lot better than we’ve had in the past. There have been some great sessions here on using email. And you shouldn’t be surprised to learn that they’re finding exactly what we find with the traditional 1930s sociometric methods asking “Who do you ____?” The great thing is that they’re results have a substantive detail way beyond what is possible with traditional sociometic methods. We’re getting down to the laser surgery.

JK: My final question, and building on the Second Life comment…

RB: Are you on Second-Life? I’m trying to imagine your avatar… [Laughter] JK: No. But that was exactly my question. I’m sure OMT newsletter readers are very curious a) if you have an avatar, and b) what that avatar is: CEO, movie star, hopefully not a super smart academic?

RB: I picked the boy next door, and as near as I can tell, he’s still running around barefoot. [Laughter] He’s in whatever form comes out of the shop. I like the neutrality. The great thing about New York, when I was there, is that you can fade into the background and watch people. You see the social process of interaction. When you are on point, you have architectural concerns like making sure people are happy and managing things, and you lose track of what’s going on in the room. Chicago is a city of tremendous heterogeneity, but phenomenal homogeneity in each place. If you add all the groups up, it looks very heterogeneous, but in any one place, everybody looks the same. I was accustomed to New York and San Francisco, the two places where I had full time jobs, where you were always surrounded by people different from you. I liked it, because you could be whatever you wanted and just watch people, et cetera. In Second Life I wanted something very neutral. No feathers, no fur, no wings, no weapons.

JK: Well, with that image in mind, I’d like to thank you for your time, and the stimulating talk.

Share/Save/Bookmark