Reading Yudkowsky, part 4

by Luke Muehlhauser on December 4, 2010 in Eliezer Yudkowsky,Resources,Reviews

AI researcher Eliezer Yudkowsky is something of an expert at human rationality, and at teaching it to others. His hundreds of posts at Overcoming Bias (now moved to Less Wrong) are a treasure trove for those who want to improve their own rationality. As such, I’m reading all of them, chronologically.

I suspect some of my readers want to improve their rationality, too. So I’m keeping a diary of my Yudkowsky reading. Feel free to follow along.

Yudkowsky’s 24th post is Statistical Bias, a somewhat technical explanation of a term you probably thought you understood. In Useful Statistical Biases, Yudkowsky notes that some statistical biases are helpful because you might, for example, be able to eliminate a huge amount of the variance by introducing a small amount of bias.

In The Error of Crowds, he argues against a particular kind of “wisdom of crowds” argument. The Majority is Always Wrong offers an unconvincing argument that the most popular option around is probably the worst.

Next: Knowing About Biases Can Hurt People. Here, Yudkowsky cites at least one study that seems to confirm the following:

  1. Prior attitude effect. Subjects who feel strongly about an issue – even when encouraged to be objective – will evaluate supportive arguments more favorably than contrary arguments.
  2. Disconfirmation bias. Subjects will spend more time and cognitive resources denigrating contrary arguments than supportive arguments.
  3. Confirmation bias. Subjects free to choose their information sources will seek out supportive rather than contrary sources.
  4. Attitude polarization. Exposing subjects to an apparently balanced set of pro and con arguments will exaggerate their initial polarization.
  5. Attitude strength effect. Subjects voicing stronger attitudes will be more prone to the above biases.
  6. Sophistication effect. Politically knowledgeable subjects, because they possess greater ammunition with which to counter-argue incongruent facts and arguments, will be more prone to the above biases.

That last one is especially annoying. It says that the more you know, the more prone you are too bias, because you have more ammunition with which to attack opposing views.

In Yudkowsky’s experience, this also happens with knowledge of biases. Those who know how biases work can very easily explain how all their opponents are falling prey to biases, and thus dismiss their arguments.

In Debiasing as Non-Self-Destruction, Yudkowsky says that all this debiasing training is “not so much about how to be extraordinarily clever, as, rather, how to not be stupid. Each profession has its own way to be clever, but their ways of not being stupid have much more in common.” Thus:

The great victories of debiasing are exactly the lottery tickets we didn’t buy – the hopes and dreams we kept in the real world, instead of diverting them into infinitesimal probabilities. The triumphs of debiasing are cults not joined; optimistic assumptions rejected during planning; time not wasted on blind alleys. It is the art of non-self-destruction.

Previous post:

Next post:

{ 8 comments… read them below or add one }

Joseph December 4, 2010 at 9:42 am

“Yudkowsky says that all this debiasing training is not so much about how to be extraordinarily clever, as, rather, how to not be stupid.”

That in itself is a bias.

No matter how much knowledge you have or don’t have, you can’t escape but forming a position, and that will constitute a “bias”. What we must remember is that one shouldn’t fear of making mistakes, of being able to recognize them and learn from them, and be bold to think outside the box.

  (Quote)

Luke Muehlhauser December 4, 2010 at 10:04 am

Joseph,

No, it’s not a bias. As I understand it, Yudkowsky is using “bias” in the formal sense: that is, a systematic departure from actual probabilities (calculated with Bayes). Yudkowsky is trying to rein human reasoning in toward ideal probabilistic calculations. Bayes doesn’t care what the end result is, only that you’re following the inescapable laws of probabilistic reasoning.

  (Quote)

Joseph December 4, 2010 at 11:19 am

I would like to see that calculation. Can you provide a link?

  (Quote)

Luke Muehlhauser December 4, 2010 at 11:21 am

Joseph,

Which calculation? This one?

  (Quote)

Joseph December 4, 2010 at 12:16 pm

I know how Bayes theorem works. I fail to see its relevance to the six types of biases enumerated in the article and to Yudkowsky’s conclusion about debiasing.

  (Quote)

Steven J December 4, 2010 at 8:14 pm

Yuddowsky should do some case studies of these types of bias. For instance, PZ Myers would provide an excellent example of all six types of bias working maximally and in conjunction. Richard Dawkins could possibly provide an example of the six biases working at a lesser level of effectiveness, perhaps due to the decaying effects of age and/or slightly higher levels, in general, of awareness (due to prolonged and extensive public polemical activity?) than Myers.

  (Quote)

Joseph December 5, 2010 at 2:38 am

Yuddowsky should do some case studies of these types of bias. For instance, PZ Myers would provide an excellent example of all six types of bias working maximally and in conjunction. Richard Dawkins could possibly provide an example of the six biases working at a lesser level of effectiveness, perhaps due to the decaying effects of age and/or slightly higher levels, in general, of awareness (due to prolonged and extensive public polemical activity?) than Myers.  (Quote)

While at it, we can ask him to carry on his studies on William Lane Craig and how his bias of limited knowledge of math, physics, cosmology and biology permeates the stupidity of his debates, or Gary Habermas and how his bias of christian interpretation in his textual analysis of alleged false copy of ancient manuscripts leads him to his propensity of wild speculation.

  (Quote)

DaVead December 5, 2010 at 7:50 pm

“William Lane Craig and how his bias of limited knowledge of math, physics, cosmology and biology permeates the stupidity of his debates”

?

It is not because of Craig’s limited knowledge that his debating tactics are criticizable. Craig is an extremely intelligent, sharp-minded, well-educated and very, very well-read philosopher. The problem with Craig is that he uses debates as opportunities to win souls for Jesus, not to engage in arguments or Bayesian analyses at a scholarly level. Not to mention, his opponents would almost never be able to follow him if he did this. He takes his stance in debates based on unspoken philosophical positions that he holds quite justifiably (and what I mean is that we really cannot fault him for ignorance or stupidity in this) based on his life’s work and research. To all the readers who keep insisting that Craig is an idiot, have you read any of his scholarly work? Have you listened to or seen him present philosophical papers or debate people in a professional forum and not in public in front of students? There he becomes a different kind of Craig completely.

What I think he is guilty for is oversimplifying things, which makes his lay-arguments open to a lot of criticism. But in the face of this criticism he just redefines his terms and reformulates his arguments to dodge these obviously holes in his arguments. You see him do this all the time in his debates. In responding to objectiosn he sometimes slips into philosopher mode and says things that go way over the audience’s head, or he’ll save time and say that such and such issues were dealt with by so and so in some book. He does the same thing in his popular works. But as much as this might be frustrating to non-believers, all philosophers do this in the public sphere.

Try reading Peter Singer’s popular book Practical Ethics and give it a full-blown philosophical analysis. Without being able to read what’s going on behind the scenes, books like this often fail philosophically. The same holds for an author like C. S. Lewis. His works are full of fallacious arguments, but look at how much he is respected by professional scholars. And then when people like Reppert or Plantinga come along and explicate what’s going on behind the scenes, like in the argument from reason from Miracles, you see how dead-on and how ahead of their time that Lewis’s intuitions and arguments actually were.

  (Quote)

Leave a Comment