Experience as a predictor of presidential performance ...

"I'd rather have someone with 2 years of experience and great ideas, than someone with 20 years of experience and the same boneheaded ideas he had 20 years ago."

I agree with you there.

Unfortunately, the statistical analysis presented in the linked article is not sound. Linear regression and correlation cannot be applied to rank data, which is what the presidential performance data presented is. This is because, for example, we cannot assume from the data that Abraham Lincoln was four times as good a president as Thomas Jefferson just because their ranks are first and fourth. Nonparametric methods need to be applied here. In fact, I did a Spearman's rank correlation test with the data as presented and got a rank correlation (NOT linear correlation) of about -.08, which is indeed not significant on n = 42.

Thus, the conclusions of the author actually are supported by the data-- experience is not a significant predictor of performance-- but the data analysis used to come to the conclusion is very wrong.

Prior experience should correlate to ability if the experience is relevant to the job. But is it? I can't see how being an ambassador, senator, or representative equips you to be President except by letting you meet and know many of the people you'll have to deal with, one way or another, once you're in the White House. General and Governor would be closer, since those are positions of command in some sense, but even there somebody stands in the line of authority between yourself and Hypothetical God.

If you look at how well the candidate did in their position, and how they acted, you might get a handle on raw ability. Raw ability is obviously something a President needs. But in the end, I suspect the only relevant experience to being Chief Executive of a country is the time you spend actually doing it. No President, elected for the first time, has any of that.