Daniel Kahneman

And the interesting thing here is that everybody has a number.

As soon as I told you that thing, a number came to mind.

Now, we know where that number came from.

We really, that's one of the few things that I'm reasonably sure I understand perfectly.

And this is that when you hear she read fluently at age four, you get an impression of how smart she is, of how precocious she was at age four.

And you could put that in percentiles.

You know, where does that put her on a percentile for sort of aptitude, ability?

And it's high.

It's not, you know, if she had read fluently at age two and a half, it would be more extreme.

But age four is pretty high.

So say at the 90th percentile.

And then the GPA that comes to your mind is around the 90th percentile in the distribution of GPA.

So you pick something, your prediction is as extreme as your impression.

And it's idiotic statistically, completely stupid, because clearly the age at which a child learned to read is not all that diagnostic with respect to GPA.

So it's better than nothing.

If you didn't know anything, you would predict the mean GPA, whatever it is, 3.1, 3.2.

Now, she's bright, so probably a little higher, but not 3.7.

You don't want to.

So that's cool.

That's a bias.