15 August 2008

People Are Fundamentally Rational

There's been some focus recently (for example, this David Brooks article) about the ways in which people are reliably irrational. This focus is largely based on the Nobel Prize (Economics) winning work of two psychologists, Amos Tversky and Daniel Kahneman. This work has led to the development of behavioral finance, which investigates the various ways that individuals differ from the simple homo economicus of basic economics. (Does that mean that economics is junk? No, but why it doesn't is beyond the scope of this post. The short answer is that rationality is a simplifying assumption of economics, but not a fundamental assumption.)

Kahneman and Tversky's work is interesting and accessible. There are worse ways to spend a few hours than reading some of their articles. My favorite K&T experiment is simplicity itself: in front of a group of people, bring in a large spinning wheel with numbers on it. Spin the wheel and get a number. Then ask the audience to estimate the number of, for example, languages spoken in the world (about 7000, but shrinking). Unless you've got a linguist (or a wise-ass blogger) in the audience, the guesses will vary around the number on the wheel, even though the audience knows that that number was chosen randomly.

But saying that people are predictably irrational is not the same as saying that they are irrational. In particular, there is logic embedded in the stable patterns of interaction between people. This is, in fact, part of the answer of traditional economics to behavioral finance: people are, for example, loss averse and thus, in the right condition, risk preferring, but the market is demonstrably not irrational is that particular way. So, that individuals are irrational does not imply that groups or society in general are irrational. In fact, conservatism is, in large measure, the insight that there is a wisdom embedded in the stable interactions of large groups of people that is more reliable than perhaps irrational decisions made by individuals: the system knows more than any single component.

What got me thinking about all this? The miserable blog interface used by the Maui News, which came up, tangentially, at Thought-Mesh. Blogging is less than 10 years old, but already there are embedded expectations and ways of doing things. The standard blog package might seem path-dependent and inefficient, but you deviate from the accepted standard at your peril. People might be irrational, but that doesn't mean that you can simply ignore the way things are done. In fact, it might mean the opposite.


Ali said...

In sort of the same way, Thought Mesh's design is slightly upsetting. I'm so used to internet pages having black text against a white background. TM looks like I'm reading from a blackboard.

Harry Eagar said...

Tom Wolfe called it the Genteel Beast (in 'The Right Stuff').

We expect our heroes to be not only good but handsome, and irrationally also expect handsome people to be good.

Thurber satirized it in a short story whose title I've forgotten but was about a Lindbergh character who was such a crude social misfit that he had to be thrown out a window and made to look like a horrible mishap in order to preserve the amour propre of the public.

That said, I don't see the spinning wheel example as irrational, just inconsequential.

It would be irrational if someone spun the wheel and the game was to agree to pay some amount of money for, say, a car, and then the audiences made bids based on the number on the wheel.

Maybe they would. People do get caught up in irrational valuations during viva voce auctions.

However, you're never going to persuade me that the mass market decisions of the many are superior the individual market decision of me as long as bottled water sells the way it does.

joe shropshire said...

Well, yeah. Say hello to your installed code base, and to the fact that it lives in the heads of your users, just as much as does in the code.

aog said...

We called it the "user conceptual model" and spent as much time designing that as we did the actual interface.