A Science of Social Prediction?
September 24, 2003 3:24 PM   Subscribe

"You'd think that predicting human behavior would be easy...everyone should be a rational economizer, busy calculating their individual costs and benefits, and acting accordingly. Right?" So begins the review of Socionomics: The Science of History and Social Prediction on slashdot. I've always thought the Elliot Wave Theory sounded like psuedoscience, but found the rational choice theory problematic as well, even ridiculous at times. What's voodoo, and what's promising in advancing predictive social sciences?
posted by weston (15 comments total)
These sociologists and political scientists have tried to build theories around the idea that all action is fundamentally 'rational' in character and that people calculate the likely costs and benefits of any action before deciding what to do. This approach to theory is known as rational choice theory, and its application to social interaction takes the form of exchange theory.

Human beings act rationally? Since when?

Economics is totally out of my field, but I find this intersection of events in the market/human behaviour to be beyond reductive. Is this the generally accepted wisdom in economics? If so, why does it work so poorly?

I'm all ears, honestly.
posted by jokeefe at 4:20 PM on September 24, 2003

It is the generally accepted wisdom -- at least, it's one of the underpinnings of neoclassical economic theory. But as a lot of folks have observed, economic theory isn't remarkably good as a predictive science, and sometimes is inadequate to explain what's happened. These folks, non-linear dynamics heavies from academia, were recruited by Wall Street types a while back in order to see if better prediction could be done in markets. And the book referenced in my post is something of a challenge to rational choice theory, from what I can tell, a competing theory.

Personally, I still think that economics is at the same point that physics was when they were still talking about things like impetus.
posted by weston at 4:29 PM on September 24, 2003

Human beings act rationally? Since when?

You're coming up against terms of art.

Rational doesn't mean "sane" in rat-choice parlance, it only means "instrumental" or "goal-oriented." That is, if I'm performing some action that I think will get me something that I think I want, I'm being rational. In a rat-choice model, Jeff Dahmer is rational -- he knows what he wants (success in life), and he does things he thinks will get him that, like killing and eating people. He just had odd beliefs about the consequences of his actions. Call it "instrumental choice" if it makes you feel better.

Rational choice models work pretty well when you can meet two conditions:

(1) You can (mostly) suss out people's preferences fairly well and fairly simply -- firms want to make money, political candidates want to win, etc, with a few wrinkles added if you're feeling adventurous.

(2) You can appeal to something like an evolutionary pressure to drive out actors who choose irrationally. By and large, firms that don't profit-maximize get driven out of business by ones that do, and politicians who aren't good at winning elections get replaced by ones who are better at it. This is what's going to make a lot of stuff a bad choice for a rat-choice model -- if there are no real, substantial consequences either way, then there's no pressure to "behave." For example, rational-choice models of voter turnout, or of mass voting, generally don't do as well as other models*, in large part because voting is more expressive than consequential behavior.

*they're still mostly right in their comparative statics, though, which is cool
posted by ROU_Xenophobe at 5:03 PM on September 24, 2003

Human behavior is actually quite simple to predict - I for example predict that there will be an anti-Bush/pro-Bush flamewar in less than 3 hours.
Someone will use the word 'patriotic' in a sarcastic sense. Otherwise sane people will insist that lying to start a war is a commendable thing.

My predictions never fail.
posted by spazzm at 5:04 PM on September 24, 2003

Okay, thanks. I'll read the links.

Does this theory factor in things like ideology, or fantasy?

Guess I'll find out.
posted by jokeefe at 5:04 PM on September 24, 2003

Also thanks ROU_Xenophobe.
posted by jokeefe at 5:05 PM on September 24, 2003

Predicting human behaviour is easy - if you always predict that people will act like arseholes, you will be right at least 50% of the time.
posted by dg at 5:09 PM on September 24, 2003

Does this theory factor in things like ideology, or fantasy?

It can, but probably not in the way you're thinking.

There are models of ideology and party identification that see them as, in part anyway, tools for economizing on information gathering and processing. If I adopt an ideology, or pick my favorite party, then I can save lots of time (and not have to think about political stuff) and I'll almost always cast the same vote I would have if I'd had all the info.

There are also psychological models of the same thing. *shrug* In the end, a lot of the differences are just different language being used to describe the same black box.
posted by ROU_Xenophobe at 5:16 PM on September 24, 2003

This reminds me of Robert Wright's book, Nonzero, which approaches human history from a game theory perspective. It's largely concerned with social evolution, with some biological evolution thrown in. Maybe one or two chapters deal with the predictive value of the approach.
posted by mbd1mbd1 at 5:28 PM on September 24, 2003

That's a lowball estimate, dg.
posted by trondant at 8:38 PM on September 24, 2003

Paging Hari Seldon
posted by Dagobert at 11:57 PM on September 24, 2003

Scott Adams writes in The Dilbert Future that there are three constants about human behavior:

We are stupid.
We are selfish.
We are horny.

Applying these three basic principles, I typically hit r-squareds in the 0.92 - 0.95 range.

'Nuf said.
posted by ZenMasterThis at 5:49 AM on September 25, 2003

I work for The Socionomics Institute, so take all of this with a grain of salt. (I will say, however, that the Institute exists entirely without a profit model - we don't have anything to sell anybody. If our theory is wrong, at least our heart is in the right place.)

Discussion of socionomics usually seems to careen pretty quickly to arguments about the respective merits of behavioral finance theory and the efficient market hypothesis. This is good, because EMH represents an economic paradigm that has been fairly successfully discredited by a number of academic research studies.

(Google these papers, or instead, sit on your butt and think about for a few minutes. When you see footage of the daily carnage that erupts on the trading floor on Wall Street, does it seem like these people could possibly be rationally calculating their choices?

Were investors acting rationally when they bid up .com IPOs in the late nineties to multiple hundreds of dollars a share, even without the existence of something like a feasible profit model driving the company?)

Although socionomics is probably rightly classified as a branch of behavioral finance, the theory is actually a lot more radical than almost any other well-known branch of economics. It's as ambitious of a social theory as it is an economic one.

Look, there's no question that aggregate social mood is an intelligible concept - the national zeitgeist is so palpable in America right now you can taste it. The question is: what drives it? The socionomic hypothesis is that the main causal motivator behind important social trends is located outside of the social system itself.

In my history with this company, I went from being a slacker copywriter on the dole, to actually coming to believe in this idea very passionately. My change of heart didn't result from the academic research. After the stock market top in April, 2000, I watched my country change in a way that I would have never, never imagined possible, and I've come to see similarities between social forces now at work and social forces at work in other bear markets in history.

I don't think that bear markets themselves send us to war. I think that fear does, and fear seems to be cyclical.
posted by Pinwheel at 6:50 AM on September 25, 2003

I don't think it's in one's individual best interest to be 100% predictable, which, if one were 100% rational one would be (of course if you were 100% rational you'd figure this out and be unpredictable once in a while). So what you have is people who are mostly rational given their knowledge of a given situation, but have an eccentricity factor built into their brains which makes predicting their behavior harder.
posted by wobh at 5:31 PM on September 25, 2003

I don't think it's in one's individual best interest to be 100% predictable, which, if one were 100% rational one would be (of course if you were 100% rational you'd figure this out and be unpredictable once in a while).

You've just described part of the eminently rational "mixed-strategies Nash equilibrium." People think long and hard about exactly which way to behave randomly is the best they can do -- think of football teams figuring out how often to pass, or F1 teams figuring out their pit strategies.
posted by ROU_Xenophobe at 6:10 PM on September 25, 2003

« Older 'Do not call' list on hold   |   And You Believed It? Newer »

This thread has been archived and is closed to new comments