I didn't want to phrase that statement too strongly because there are more degrees of success than failure. It may be that avoiding denial is enough to avoid failure, but that to be a big success you need other things as well, like being smart or being in a hot market.

The vast majority of violent criminals in the USA are male, but the majority of males are not violent criminals.

And yet being male still almost perfectly correlates with being a violent criminal. asdfologist probably picks on a loose use of word "correlation", correlation is by definition symmetric. pg is really talking about conditional probabilities, but honestly I think everyone understands what he means.

You're right-I was also thinking of conditional probabilities. I appreciate the correction-I'm trying to learn to be more precise.

I did interpret it to be symmetric, but now that I think about it you seem to be right that colloquially it does (generally) refer to conditional probability. I disagree though that everyone interprets it that way, based on the upvotes my comment got.

Self examination in this case would be a necessary but not sufficient condition.

Not if more people have this trait than are successful.