Howard Burton - The Social World, Reexamined: A Conversation with Brian Epstein

Brian Epstein likes to talk about anthropocentric ontologies and frame principles, enthusiastically holding forth on topics like grounding, anchoring and supervenience. From all of this you may be tempted to discount him as just another abstruse academic peddling jargon that has nothing to do with the real world.

But youd be dead wrong.

Because while Brian certainly has the necessary academic credentialsAssociate Professor of Philosophy at Tufts University, after degrees from Princeton, Oxford and Stanfordhe is nothing less than deeply absorbed by the everyday social world surrounding us. So much so, in fact, that he has devoted much of his professional career trying to puzzle out why so many of our current models of economics, politics, and other areas of social science so often go terribly wrong.

Not coincidentally, his curiosity about such issues was first piqued far from the ivory tower, during his years as a management consultant.

One of the projects that I worked on was incredibly illuminating to me. It was a macro-trend study for a mutual fund company in Japan. An American mutual fund company had some mutual funds they were investing in Japan, and they wanted to know what was going to be happening in the Japanese market. This was in the early 90s and, although the bubble had started to burst in Japan, it was still really the centre of the global economy. So we went in and tried to predict what was going to happen in the Japanese marketplace.

We came up with, maybe, ten predictions; and every one of them was completely wrong. I was amazed. We knew, of course, that the Japanese population was agingthat was one thing that we didnt get wrongbut our predictions based on that were completely wrong. Everything that we did was wrong.

A less committed fellow might have just shrugged his shoulders and continued climbing the lucrative corporate ladder of blithely making uniformly wrong predictions. But Brian took a strikingly different route, enrolling in a PhD program at Stanford in philosophy of language.

One of the core parts of philosophy of language is the theory of reference: how it is that we attach words to things in the world, how we name things like electrons or water or gold or things that are independent of us. Theres been a tremendous amount of work over the last 20 or 30 years on that kind of problem.

What theres been less work on is that its also very easy for us to attach words to non-natural things in the world, like glass or plastic or curtains or buildings or economies things that we dont necessarily understand any better than the ancients understood the nature of water or gold. We dont really understand what these things are, yet we talk about them very easily.

What was needed, then, was a theory of reference to the social world, which was something that nobody had been thinking about. That was the problem that I started to really pay attention to in my dissertation work: how language worked in the social world. Over time, it just became clearer and clearer that it had a lot less to do with language, and what I was really interested in was the metaphysics of the social world.

Well, thats nice, you might think, and very much the sort of thing that people with PhDs in philosophy would tend to talk about. But how many of the rest of us can afford to spend our time puzzling over the metaphysics of the social world, whatever that might be. There seems to be a very large relevance gap between social metaphysics and our day-to-day lives.

But there most definitely isnt: which is precisely Brians point.

A key point that hes keen to emphasize is that, up until now, weve assumed that everything in the social world around us is built from peoples thoughts or beliefs or actions or interactions with other peoplewhat he calls an anthropocentric perspective. But careful reflection shows that cant be right.

Lets say that you have an insurance policy on your house. This policy is a product of human agreementsyou agree with your insurance company on these terms and conditionsand then there are these facts like, your insurance company owes you $10,000 because something happened. Well, what are those terms and conditions? What do those depend on? What are the building blocks of that fact?

These sorts of factsfacts about liabilities, debt, and moneyare extremely central to our social world. They are elemental parts of how weve built up our complex, social world, so we have to think seriously about them.

If you take a fact like the bank owes you $10,000, and you ask yourself about the ontology of that fact and what it depends on, you might realize, after looking around, that it depends on nothing human at all. It might depend on a forest in Virginia, or a house out in the Turks and Caicos.

In other words, the ontology of the facts about a contract are really whatever you stipulate them to be. The point is that the building blocks of these sorts of social objectsa bond, a liability, a contract, a university, a corporationdont have to be human-centric at all. They could be dog-centric, they could be airplane-centric. Or they might not have any kind of simple building blocks at all.

Well, OK, you might grudgingly agree, he does seem to have a point. And an insurance policy is, indeed, something sufficiently tangible and non-academic. But stillwhat real difference does all this make?

Well, Brian tells us, the problem is that if we dont appreciate what actually makes up the social worldif we naively assume, say, that its based on something that its notthen its hardly surprising that all of our models will turn out to be wrong too.

Which means that its all very well and good to be excited about the revolutionary potential of the dawning of the big data age of unprecedently large computer power, but if our underlying models are flawed, then were still not going to get anywhere.

Theres a standard way of building simulations of the social world, which is called agent-based modelling. Basically, modellers take a bunch of people as agents, put them in an environment and then let them go off and interact with one another.