Arnold Kling says

Also, I am in the process of re-reading the Converse issue of Critical Review, which I first blogged about five years ago. In Jeffrey Friedman’s essay, he raises the issue of how Rush Limbaugh and Paul Krugman could each be sure that he is right.

I think one can model this metaphorically as the outcome of a hill-climbing algorithm where you can get stuck at a local maximum. I will explain this below the fold.

A hill-climbing algorithm is a way to solve for the maximum of a function. Imagine that you were plopped down in the middle of some topographically varied terrain and were trying to find the highest mountain peak. Using a hill-climbing algorithm, you would send out small probes in all directions and move in the direction where altitude is increasing. Then repeat, until you get to a point where altitude is declining in every direction.

If there is only one peak in the terrain, this method will find it. But if there are many hills, it is also possible to get stuck at the top of a small hill and never find the peak of the highest mountain.

Think of Limbaugh and Krugman as being stuck on their own hills. Based on where they started, and the paths that their experiences took them, each is at a point where he cannot see any way to improve his understanding of the world by changing his mind. Even though their views are incompatible

What occurs to me is that the conviction that one can persuade is key. A necessary condition of feeling comfortable I am right and some other folks are wrong is that I believe that I can persuade them that I am correct.

If not then I have to concede that no evidence exists which could convince a person of any origin of the truth. If that is the case then no evidence exists which could convince me of the truth irrespective of my origin. Thus my knowledge of the truth depends crucially on my origin, which is presumably orthogonal to the truth about most issues.