Top banner


An interview with Harry Collins

Greg Ross

As science and technology inform our society, we find ourselves increasingly reliant on experts. But what is an expert? How can we—professionals, policymakers, voters—assess the advice of others whose competence we don't share? And what does this mean for the enterprise of science and for our society in general?Harry CollinsClick to Enlarge Image

In Rethinking Expertise (University of Chicago Press, 2007), Cardiff University sociologists Harry Collins and Robert Evans consider these questions and offer a framework for exploring their import in science and in society. "Only this way," they write, "can the social sciences and philosophy contribute something positive to the resolution of the dilemmas that face us here and now."

American Scientist Online managing editor Greg Ross interviewed Collins by e-mail in March 2008.

What led you to this topic?

The idea of analyzing expertise grew out of my long study of the sociology of gravitational wave detection. I've slowly become a quasi-member of the gravitational wave community. This means I chat with my new colleagues in restaurants, cafeterias and coffee bars. I began to find I was talking physics—just the normal to-and-fro of science chat. Sometimes I would recommend that they try something different in the experiments and my remarks weren't just shrugged off; for instance, I might be putting a case that had been considered and rejected for physics reasons that I could follow, or, rarely, I might even get something right.

This began to strike me as interesting: Here was someone, all of whose university degrees were in sociology, talking physics with physicists. I could not do the math, design the circuits or solder wires, and I would never contribute to a physics paper. Yet I could still talk gravitational wave physics.

Then it struck me that the managers of the big gravitational wave experiments, such as the Laser Interferometer Gravitational-Wave Observatory (LIGO), were also not doing much in the way of maths, or designing and building experiments, or co-authoring research papers in the field. Most of what they did was mediated by the same kind of talk that I was doing. And I also realized that talk of this kind was what I heard when I sat in on review committees—it was talk that happened in these places, not calculating or experimenting. I could follow most of this talk, and, every now and again, I felt that I could even have offered something. This made me think about the nature of expertise: how my expertise differed from that of the scientists and the managers.

Expertise is important not only within science but also for understanding the public's relationship with science. Nowadays any parent of a young child, or anyone who can access the Internet, thinks their opinions on technical matters are sound. Many of my colleagues in the social sciences seem to think the same thing. The trouble is that the speed of politics is faster than the speed of scientific consensus formation, so politicians are often faced with making decisions without firm scientific answers to lean on, and this makes science look like anyone else's opinion. I found I wanted to work out how to value expertise without going back to the bad old days where anyone in a white coat was treated as an authority on anything scientific or technological. We have to solve the very hard problem of reconstructing the value of science when we know it can't deliver the certainty that people want. Studying expertise may do the trick.

How does one undertake a study of expertise?

As a philosophically inclined sociologist I begin to study expertise by looking around. Sometimes I will also talk to people about the way they decide things. For example, I interviewed Gary Sanders about a year after he moved from LIGO to become the project director of the Thirty Meter Telescope (TMT), and I interviewed Jay Marx when he was a new director of LIGO. I wanted to know what they brought to their decisions, given that they were steering projects in a science that was new to them.

My kind of expertise in physics—"interactional expertise"—turns out to be central. This goes along with "referred expertise," which is experience brought from another field and indirectly applicable. A paper on science management, co-authored with Gary, along with most of what is discussed here, is available here.

A so-called "Periodic Table of Expertises" is assembled in Rethinking Expertise. The table turns on the idea of "tacit knowledge"—things you can do but can't describe how. Riding a bike is the best-known example, but both everyday life and contributions to technical domains depend on tacit knowledge. The Periodic Table is divided up according to how much of what kind of tacit knowledge is involved. There are about 20 categories altogether, the bottom rows being about "meta-expertise," which is the expertise we use to judge others. "Referred expertise" is a meta-expertise.

Technical decision-making is often a matter of debating in committees and the like, so the way expertise works itself out in conversation was always going to be a central concern. We decided to use the forerunner of the "Turing test"—the "imitation game"—to see whether one kind of expert could be distinguished from another in conversational tests. In the imitation game, a judge asks open-ended questions of, say, a full-blown expert and someone with interactional expertise only, without knowing who is who. The judge tries to tell the difference. In the best-known of the experiments we did in Cardiff, color-blind people were found to be indistinguishable from color perceivers, and we argued this was because the former had been immersed in the language of the latter all their lives. I myself also managed to put up a good show answering questions about gravitational wave physics. A description of these experiments can be found in Rethinking Expertise and was also written up in Nature in 2006 (Jim Giles, "Sociologist Fools Physics Judges," 442:8).

So, in answer to the question of how one studies expertise, I would say "in as many ways as possible." I have used immersion in the social life of my respondents (what I call "participant comprehension") along with distanced reflection on my experience, philosophical/theoretical analysis, interviews, experiments with and without statistical analysis of the results, anecdotes, reflective analysis of one's own understanding of a text, and surveys. I would like to do many more imitation game experiments and really work out their technicalities—there are many questions to ask—but so far all our applications to funding agencies have failed.

Can you give an example of interactional expertise?

To give a nice example I'll put together some of things Gary Sanders says in our joint paper. He talks about the problem of building the adaptive optics for the Thirty Meter Telescope's main mirror:

This whole art of sensing the wavefront distortion and doing the mathematics to unfold the correct corrections is an extremely arcane field. … But I've found that, remarkably, what you call interactional expertise was not hard to achieve. … I can sit down in a group of adaptive optics experts who will come to me and say, "Gary, you're wrong, multi-object adaptive optics will be ready at first light, and it will give the following advantages"—and others will say, "No, it's multi-conjugative adaptive optics," and I can give them four reasons why we should go with multi-conjugative adaptive optics based on the kind of science we want to do, the readiness of the technical components, when we need them and so on. And I will see as I am talking about it that the room is looking back at me, and they're saying, "He does have a line, he's thought it through, and yes." But if someone said to me, "Okay, Sanders, we agree with you, now go and design a multi-conjugative adaptive optics system," I couldn't do it. I couldn't sit down and write out the equations. But I can draw a diagram of what each part does, where the technological readiness of each one is, what the hard parts are—I know the language.

That's a nice clear example, but once you "cotton on" to the idea you realize that science, and indeed life in general, would be impossible without interactional expertise. Take a big project like LIGO or the TMT—such projects are made up of many small groups of experts with the hands-on ability to do narrowly specified things such as polish mirrors, design suspension systems, calculate waveforms or work out multi-adaptive optics. The point is that they cannot do each others' jobs. But to make something into a project rather than a collection of isolated specialists, the different sub-groups have to be able to talk knowledgeably to each others' jobs. Extend the idea and you can see that there could be no division of labor—no teamwork and no society—if there was no interactional expertise. Interactional expertise is everywhere. The strange thing is that no one has noticed until now.

Though maybe it is not so strange. There is a perpetual battle in life between the doers and the talkers. We are always disdaining those who can "talk the talk" but cannot "walk the walk," and mostly we are right to do so. Interactional expertise is not simply "talking the talk," however. It isn't a matter of bluff or passing yourself off as something that you are not. It is really hard to acquire interactional expertise, and if it can be acquired it takes its place alongside the other practical capacities. Someone with interactional expertise does more than talk the talk, they "walk the talk." That interactional expertise has not been given credit until now is probably because it sits "between the lines" in the old battle between the mouth and the hand.

If political decisions can't be informed by careful science, does this spell trouble for democracy as the world grows more complex?

I do not think that there has to be any trouble for democracy, but we live in dangerous times. The argument goes back at least as far as Plato's suggestion that the Republic should be controlled by "philosopher-kings." We now know that cannot work—experts are too fallible, and too much power corrupts. In the last resort, all decisions have to be made through the machinery of democratic politics if we want to preserve a society like ours.

On the other hand, it is vital to preserve the separate spheres of the technical and the political. In practice this can never be achieved, but if we don't try we will destroy the very idea of science and of expertise as a whole. I would say that the danger to democracy that my own discipline—social studies of science—is not doing enough to combat is the collapse of the idea of expertise. Current social studies of science has difficulty with the notion of expertise. The attitude that anyone's opinion on any topic is equally valuable could spread, and there are some indications, such as widespread vaccine scares, that suggest it is happening. A world in which there is said to be no difference between those who know what they are talking about and those who don't is not one that anyone who thinks about it wants. Such a society would be like one's worst nightmare, exhibiting many of the characteristics of the most vile epochs of human history.

Philosopher-king fascism won't work, but a reaction to it that creates technological populism is just as bad. It is very hard to work out how to find a rationale for a middle position which does not replace one extreme with the other. Our studies of expertise perhaps indicate how we might establish a middle view now that we know that science and technology cannot deliver the kind of certainties that politicians need at the speed that policy unfolds.

As you developed these ideas, did they affect your perspective on your own research?

One way the Periodic Table has affected me and my research group (The Centre for the Study of Knowledge, Expertise and Science—KES) is that, now, when we do a piece of research or write a paper, we often reflect on the expertises we bring to it. We are adding a little "inventory of expertises" to some of the papers we are submitting. Members of the Cardiff group even undertook to list their expertises as social scientists, and we then held a seminar to discuss them. It was quite a difficult and interesting exercise. One of the group members has plans to write it up.

comments powered by Disqus


Bottom Banner