Bryan Caplan  

Jane Galt's Self-Criticism

Tranquility for A Dollar a Day... James Hamilton on the Great Ra...

One of the main reasons to study psychological biases is to help us stop making them, but even many specialists don't bother to try to reform their thought processes. But don't give up hope. Jane Galt is a role model of intellectual integrity:

As the days go by, I find myself less and less worried about terrorism, and more and more worried about things like Lindsay Graham's proposed bill to strip habeas corpus rights from non-citizens...

The reason for this mental shift is not because I have carefully studied the issues, and concluded that terrorism is not as great a threat as I once thought. The reason is simply that time is putting ever more distance between me and 9/11.

But this is not a good reason to believe that terrorism is less of a risk than it once was. Terrorism was a HUGE threat on September 10th, 2001, even though it had been 8 years since Al Qaeda attempted anything on American soil.

I can feel myself committing the basic decision error of believing that something that hasn't happened in a while is therefore less likely to happen tomorrow, even though for many events, such as earthquakes, the fact that it has been a while since the last one makes it more likely, not less, that a big event will occur in the near future...

I don't mean to suggest that my current assessment is necessarily wrong; indeed, I'm sure that my analysis of the relative risks of civil liberties curtailment and terrorism was wildly skewed in the days after the World Trade Centre was destroyed. But changing my priorities should be the result of rational analysis, not the simple passage of time.

Comments and Sharing

COMMENTS (9 to date)
Elkharth writes:

....the 'study psychological biases' should focus more upon the distorted view of actual "risks" in our daily lives.

Jane Galt still finds the 'risk' of terrorism to be "huge" -- although the actual harm to Americans (..including 9/11) is trivial relative to other risk.

About 40,000 Americans die horrible deaths 'every' year from automobile accidents, with many times that number seriously injured --- yet no one gives a second thought to hopping in a car each day.

Emotionalism always seems to trump reason, especially in politics.

Jane Galt writes:

But I can almost entirely remove my risk of dying in a car crash by not getting into a car--and indeed have done so by deciding to live in a city and getting in a car only a few times a year. Even if I drive, I can lower my risk my driving slowly in bad conditions, obeying the speed limit, and keeping a lookout for drunk drivers. Reducing my risk of terror attack, on the other hand, requires considerably more cooperation with my fellow citizens, and thus is the proper focus of a policy blog.

Terorism has a number of other features that make it appropriate to worry about it in a different way from car crashes. It results from deliberate human action, rather than negligence or bad luck, which violates our norms about justice; it is a large magnitude event; its victims have no reasonable chance of avoiding the attack, unlike many people in car accidents; and allowing some of it may encourage more of it, by providing fodder for fundraising and recruitment, and in teh case of Madrid, convincing terrorists that their methods are effective. In contrast, no one I know decides to crash their car because it worked out so well for the neighbours.

gamigin writes:

Yes it's very emotionally honest. It's also overly introspective and irrelevant to any real issues.

Bryan Caplan, why don't you speak with the same honesty and admit that you and the other well educated bloggers at GMU wouldn't pay any kind of attention to this nonsense if Jane Galt were either:
A) A man
B) fat
C) old

Will writes:

the fact that it has been a while since the last one makes it more likely, not less, that a big event will occur in the near future...

Come on Jane Galt, this is a classic statement of the Gambler's Fallacy. For any stationary probalistic process in which events are independent from period to period (an assumption to be sure), the record of events prior to the present is completely irrelevant. 200 heads in a row doesn't make it more likely that the next flip will be a tails (assuming you know a priori that the coin is fair).

Jane Galt writes:

Indeed, Will, you are correct. But there are many events in this world which are not independant from period to period, such as earthquakes, which was the example I cited; the longer an earthquake zone goes without a major seismic event, the more stress builds up, and the more probability at any given time that a major event will occur.

Terrorist events are probably somewhat independant, but only somewhat. Any group that plans an event will use up resources to execute it; assuming that there is a relatively short supply of things like bombmakers and competent people willing to blow themselves up on trains or fly planes into buildings, you are safest from a repeat attack right after the last one has occurred. The immediate aftermath of an attack could well also discourage other organisations from launching their attacks, because of heightened security and alertness, and because they do not want their "brand" to be lost in the shadow of a bigger attack; had we ever figured out who the anthrax mailer was, his name would be much less well known than that of Al Qaeda. (Though to be fair, attacks could also inspire copycats.)

If you think that the probability of serial terrorist attacks can be calculated the way that the probability of serial coin tosses can be, I'm not the one committing a statistical error.

Jane Galt writes:

And Gamigin, I'm flattered that you find me so fetching that you think my beauty overwhelms my otherwise lacklustre ideas.

Bernard Yomtov writes:


I don't think it's the gambler's fallacy. After all, we don't know the probability of a terrorist attack in any given time period. We are guessing. Why shouldn't we change our guess as a result of experience. And if we reduce our subjective probability estimate doesn't it make sense to reduce our fear of an attack?

For example, most parents are probably fairly nervous when their teenagers begin to drive. But if time goes by and the teenager has no accidents or tickets, that nervousness sensibly subsides a bit, for the same reason - the parents' estimate of the chance of a problem drops based on experience.

My objection to Jane's thinking is that she seems to assume that measures like Graham's help protect us from terrorism. I think that's dubious.

Paul N writes:

I appreciate the logic of Jane's statement - she may side with Taleb that we underestimate the probability of unlikely events that we don't think about - but I think this is a bad example. I for one have thought terrorism was overemphasized before 9/11, on 9/11, and every day since 9/11.

At some point you have to look at numbers and cost-benefit, and with terrorism it's just not there. From a Bayesian perspective, the more years pass between major terrorist events, the less likely one is to happen again.

gamigin writes:

Wow, my post actually got a reply. LOL! Take care!

Comments for this entry have been closed
Return to top