Brian Epstein – TRANSCRIPT
Let me take you back to the autumn of 2008, when the global financial system almost collapsed. Back in March, Bear Stearns had gone bankrupt, but people weren’t really sure then whether that was a one-off event, or whether it was foretelling something worse. Then in July, IndyMac, a huge mortgage bank, went under, and markets started to get nervous.
By the middle of September, they were in full panic. On September 15th, Lehman Brothers collapsed. The very next day, AIG had to be bailed out for 85 billion dollars. Then on September 29th, the Dow fell 778 points. And so began a multiyear global depression that we’re only now getting out of. It sure is a good thing we have 15,000 economists in the United States. You might have thought they could have let us know this was coming. You might have thought they could have given us some policies for when it did arrive. But they didn’t.
Before the crisis, the greatest economists in the world didn’t see it. When the crisis came, they bickered about what to do. Today, years later, they’re still debating the fundamental causes. No wonder we don’t have good policies in place for avoiding the next crisis. This represents a great failure of social science. The social sciences are not working. Sociology, economics, political science, history; we need good answers from these disciplines.
Imagine what we could accomplish if we were actually good at social science. Imagine the crises we could avoid. Imagine the lives we could save. Today I want to talk to you about some new ideas for fixing the foundations of social science, for rebuilding them from the ground up. At the heart of this talk is one simple point: There’s a certain kind of question that we’re not really asking in the social sciences, something that we’re largely overlooking. It’s what I’ll call a “What is it?” question. These are questions we rarely ask, and when we do bother to ask them, we tend to get the answers wrong. The reason is the social sciences are largely in the Dark Ages, they’re relying on old assumptions and dogmas that we need to overturn.
So let me start by distinguishing two kinds of questions in the sciences: “What is it?” questions and “How does it work?” questions. These questions are fairly familiar in the physical sciences or natural sciences, like physics and chemistry and biology. “What is it?” questions are questions about structure or composition. Like, suppose you have a molecule of DNA. What is the structure of it? What are it’s building blocks? What is it? “How does it work?” questions are questions about mechanisms or processes. Like, imagine that you have a strand of DNA and some enzymes out of which a protein is being built. What’s the process by which that’s happening? What’s the sequence of causes and effects?
Scientists spend some of their time answering the first kind of question and some of their time answering the second kind of question. And importantly, those two questions interact with one another. If you want to do a good job answering “How does it work?” questions, it’s a really good idea to do a good job answering “What is it?” questions as well. The thing about the social sciences is we spend a lot of time on “How does it work?” questions. But “What is it?” questions? We don’t really do that. What is money? What is a company? What is an institution? What’s a social group? What are credit cards, financial instruments, contracts? These are questions we really don’t ask in social sciences.
Here’s an interesting comparison: I had a research associate look at two of the top journals. One in the natural sciences and one in the social sciences, Nature and the American Economic Review. We looked at the 400 or so articles that came out over the last year and coded them in terms of what kind of questions they were addressing. Here are the results: The blue bars represent Nature, and as you can see, about 16 per cent of the articles in Nature are addressing purely “What is it?” questions. Then another 12 per cent are addressing a mix of questions. The American Economic Review is very different. Out of the 242 articles we looked at, only two of them are addressing purely “What is it?” questions. And only 13 were even addressing a mix. These are things we’re not really paying attention to. Why do the natural sciences spend so much time addressing these questions?
Let me tell you a story about what happens if you get your answers to “What is it?” questions wrong. This is a story from biology, from the 1860s. But as you’ll see, the things that were going wrong then are remarkably similar to the things that are going wrong in the social sciences today. Here’s a guy named Rudolf Virchow. Back in the 19th century, he was a very prominent biologist arguing for the cell theory of organisms. Now it may be hard to believe, but at the time, there were a lot of debates about what organisms like us are made of. Some people thought that we’re made of some stuff called protoplasm, whatever that is, some people thought that we’re made of fluids with a vital life force flowing up and down us.
Virchow argued for the cell theory of organisms, and here are a couple of the principles he put forward. One is: All organisms are made exclusively out of cells. A second is: Cells are the functional and structural units of organisms. Cell theory really is very good. It’s much better than life-force fluid theory. But the problem is it’s not quite true. Think about a body. There are lots of cells in the body, but lots of parts of the body are non-cellular as well. Think, for an instance, about the eye. There are lots of cells in the retina or cells in the optic nerve, but most of the eye is non-cellular; the cornea has almost no cells, the lens. Most of the eye, is the vitreous humor and that also has almost no cells. The eye is an amazing structural and functional device, and yet it’s not made out of cells. The same is true for lots of other parts of the body as well. The bones, 15 per cent of the body. The hair, the teeth, all the fluids in the body. Virchow got the answer to the “What is it?” question about the body wrong. It was too simple. It was too uniform.
One more point about this: Suppose you want to use Virchow’s answer to “What is it?” questions to then answer “How does it work?” questions. For instance, suppose you want to construct a simulation, and in the simulation of the body as a whole, you only simulate the cells. And you ignore, overlook, all the non-cellular stuff. This will not be a good simulation of the body. You might simulate a few things well. You might simulate the muscles well, or you might simulate the organs well. But as a simulation of the body as a whole, it will be terrible. You can probably predict what’s going to happen.
With a bad answer to “What is it?” questions, we don’t have a prayer of giving a good answer to “How does it work?” questions. There’s an assumption that’s deeply built into the social sciences today, which is that the social world is built exclusively out of people. What’s an economy? It’s a bunch of people interacting with one another. What’s a company? It’s a bunch of people interacting with one another. Let me give you three examples of cutting-edge social science to illustrate this: Here’s a picture I took from a recent book on simulating corporations.
As you can see, what they’ve done in terms of organizing a simulation is to take people and then to cluster them into hierarchies or groups. If you followed their instructions for simulating a corporation, you are going to simulate the people. Nothing more. Here’s an example of a field of social science that’s getting an enormous amount of attention: social network theory. In this diagram, you can see that a society as a whole is represented as a graph, with people at the nodes and relations between people as the lines between the nodes.
The society as a whole is represented by people and their relationships, nothing more. Or general equilibrium theory in economics. This is the workhorse of economic theory, it’s what macro-economists spend much of their time doing. These models are slightly more complicated in terms of their building blocks than the ones I just showed you, but not much more, they’re still very simple. They start with a group of people, or what they call “A cohort of agents,” then they add just a few more building blocks. They add some resources, they add some firms which are basically understood as “black boxes that take inputs and transform them into outputs,” and then they sometimes will add a government or some bonds. A little more complicated but still very simple.
I was talking to an economist the other day, and he was like: “Sure, society is built out of people, and these are the building blocks of our model, but what else is there? Are you suggesting that there’s some sort of dark matter in the social world that we’re not really representing? Some invisible stuff we can’t see?” Now I’ve gotten this reaction quite a bit, but whenever I get it, I’m always kind of puzzled because there is no dark matter. All we need to do is look around us. Look around you right now. Sure you see some people, but there’s a lot more than people. Or look in your company. Or look in your university. Or in your living room, or your car, or look at your wallet, at the bills and cards in your wallet. Each one of these things is built on a complex hierarchy of parts.
To see this vividly, think about the building blocks of an economy. Or maybe that’s too complicated. Think about the building blocks of a single company. Or less complicated, think about the building blocks of the company’s balance sheet. Or even one account. Or even simpler, think about my bank account. As you can see from this ATM receipt, I have a $78 balance in my bank account. This is what you get for going to academia instead of Silicon Valley.
So what are the building blocks of that? Well partly, it’s actions by me, a person – my deposits and my withdrawals, maybe interacting with tellers and machines – but there’s a lot more behind it than just that. For instance, in order for me to have that bank balance, Bank of America needs to be a depository institution, it has to be able to take deposits. And that depends on an enormous hierarchy of complicated stuff. It partly depends on their employees, but it also depends on things like licenses, which have lots of complicated terms and conditions, and those conditions have their own dependencies. Then there are things like bank capital, and relationships with other banks like the Federal Reserve, which then has it’s own dependencies.
And there’s more. For instance, the Bank of America needs to be solvent. For it to be solvent, depends on it’s capital and on its loan portfolio, and deposits and other stuff. Or if it’s not solvent, then there’s the FDIC, which has its own dependencies. The point is not the details of this particular example. The point is that what looks like a very simple fact is actually built out of a complicated and heterogeneous set of building blocks. And that is just one guy’s bank account.
Think about a company. Think about an economy. The idea that these things are built exclusively out of people is crazy. What do we do when we encounter something like the financial crisis? We want to understand what happened, we want to understand how it works. But to understand that, we need to understand the mechanisms behind the parts of that. And in order to understand that, we need to understand what those parts are. This is something we have to take seriously, at least as seriously as we take DNA in the natural sciences. It’s not that social scientists are blind. They can look around them and see that the world is a complicated place.
But what they’re doing is they think that society is made exclusively out of people, and so it interacts with the other parts of the world. So you come up with this clean separation of society and people here, and resources here and firms here, and so on. But if you just take one glance at this complicated hierarchy, it’s clear that couldn’t possibly be right. There is no way of pulling out and separating the people in our models.
Our approach to the social world overall is too centered on people; it’s too people-centric, too anthropocentric. 500 years ago, Copernicus argued against anthropocentrism, about the solar system. Before Copernicus, astronomers couldn’t imagine a universe where people weren’t right at the center. Those astronomers, of course, were proven wrong. We are part of the universe, but we are not at the center of the universe.
Amazingly, 500 years later, the social sciences are still relying on a similar kind of assumption. But this is not the way it has to be. We don’t have to labor under the dogmas of an earlier generation. All we need to do is take the “What is it?” questions even half as seriously in social science as we do in the natural sciences, and we have the potential to make some real progress.
Here’s some steps we can take. First: We have to recognize the problem. There’s no way that we are going to improve this stuff until we see that “What is it?” questions are important and that we are not addressing them well. Also, critical, is to go beyond just thinking about psychology. Lots of people are thinking about how to fix the social sciences, but by large the way they are doing it is by improving our models of the human psyche. For instance, the biggest trend, by far, in the social sciences today is taking account of the fact that people are not always rational. Sometimes we are irrational.
Now, this is something we’ve known for a very long time, but people are starting to systematically incorporate it into their models. This is a good thing. It’s a good thing for people to do. But we have to understand that even if we got the human psyche perfect, we’re still only addressing a small fraction of the social world. Second: We need to experiment. We need to try things out in a variety of places, this is not a task for big science to try to do a moonshot. It’s a task for a lot of us to experiment and build on things. We’re very much at the beginning of understanding the nature of the social world. A lot of different people and disciplines need to start contributing to it.
And thirdly: We need to build out our intellectual infrastructure, and we need to build out our modelling infrastructure. I come from philosophy, and in recent years, we’ve made some huge strides in cracking some of these problems about the social world. But still, we’re only just starting. And of course, philosophy’s just one place where this needs to happen. It needs to happen in the social sciences.
Modelling infrastructure needs to be built in computer science and in mathematics, and examples need to be drawn from business and from the humanities. Let me conclude by saying that this is not something that we are going to solve in a day or a week or a month. It’s an ongoing project in social science, just as it is in natural science. But even if we can make small strides, little by little, it can have an enormous impact. The stakes are huge in the social sciences. And if we can start making improvements, we just might find that after years and years and years of effort, that we can finally draw on the social sciences to solve some of the most pressing problems of our time. Thank you.