Humans are Naturally Tone Deaf to Probabilities | Hacker News Reader

Humans are Naturally Tone Deaf to Probabilities

michaelgr.com

11 pointsMikeCapone11 comments

I have consistently observed this and been frustrated at decision-makers for making what looks like fear-driven rather than rational decisions.

I like this article because it gives a hypothesis that explains this. But how do you educate people?

When I make cost-benefit analyses involving probabilities, an expected value calculation is really important, and I don't think everyone knows how to do it. So teaching that seems to be valuable. But when people don't have hard probabilities to look at, as the author notes, they seem to overestimate risk. That's something that can't be helped by teaching a formula - you have to teach an intuition.

Strategic games have probably helped me develop this intuition. Starcraft and the board game Puerto Rico both require you to continually estimate risks and rewards of taking certain actions.

I don't think education would completely fix the problem, but it would certainly be a step in the right direction. Even for people who never become very good at it, at least being aware of the problem would be better than nothing.

It's like learning about cognitive biases. You'll still have most of them, but just being aware that they are there is a good start.

I like your comment about strategic games. I played them too, and wonder if they had an impact.

Maybe this could actually be part of education - why should it have to be boring?

Yes, it's very frustrating to watch. I have a sneaky suspicion that politicians, or at least their advisors, are well aware of the bare statistical truth. But they can't take the counterintuitive but right stance because they would alienate their voters who don't understand it.

This is one of the downsides of democracy.

Exatly. Political leaders are rarely leaders at all, they are followers. If you want to change how politicians act, you almost have to change the rest of the voting public up to some significant portion.

This isn't a problem of democracy. It is far more general than that: it's a problem with humans. It persists even on the small scale -- in an oligarchy, and even in a monarchy or a dictatorship. Indeed, it might be an even worse in a dictatorship, because when the dictator decides to prefer an intuitive falsehood to a counterintuitive truth, there's no effective way to argue.

The answer to this problem, at every scale, is marketing. That's not what a roomful of rationalists wants to hear, of course, but it is the correct answer: Although it is wise to support proposals that are backed up by statistics (because those are the ones that are likely to work!) you also need a sales pitch. You must get people to associate the proposals with mom and apple pie, or with sexy people, or with Progress. Or you must convince them to follow the herd.

Well, you are definitely right about one thing: policy analysis is, unfortunately, a task too exhaustive for the average American. Therefore, I can definitely imagine a counterintuitive (but right) stance turning off voters.

However, one thing that I have learned throughout this whole financial crisis is that people do care. And regardless of whether or not the underlying problem(s) are generally understood, the public wants solutions...and good ones.

Political stances are always under a lot of scrutiny from the university professors and media networks. So if the general public cannot analyze and understand the full situation they will instead rely on and digest the consensus of the intellectual community. I cannot see politicians (at least smart ones) worrying about having their their voters understanding their policies. They have bigger fish to fry. They have to be right, they have a record to defend, and they want to carry on a political career that can only advance with a record of retrospectively, good judgment. Matthew Arnold put it best with, "The freethinking of one age is the common sense of the next."

Back in April, a number of papers were running a story on the Monty Hall problem in primate research. The NYT ran a story with a detailed explanation of how the problem works, and even provided a flash-animated demo so you could work through it yourself.

Nonetheless, it was amusing to read the wild-eyed diatribes in the user comments. You couldn't explain the problem in a clearer way, and yet plenty of people simply refuse to accept a result that seems counterintuitive.

After reading your comment, I immediately headed over to Wikipedia to refresh my memory on the Monty Hall problem. The results of my search are a good illustration of why I love the internet:

http://en.wikipedia.org/wiki/Monty_hall_problem

The article is nearly 6000 words and includes diagrams, detailed mathematical explanations, a very detailed synopsis of the history of the problem, and links to like 50 sources.

All for a probability puzzle. Unbelievable.

I find it interesting that the best model we have for how the brain operates is a Bayesian inference engine, but most people can't even apply proper Bayesian reasoning.

Log scale probabilities are a lot easier to grasp, but we still do not have a unit name for them :(

Reply on news.ycombinator.com