It can simultaneously be true that to be successful in your field, you need to be unconventional relative to that field, and also that the means and methods for being usefully unconventional transfer across fields.

"This article is self-contradictory. The article says good programmers write code in a way that doesn't cause surprises for the reader. But the only way for a program to be totally unsurprising is if it's an exact copy of another program which already exists!"

...

"The formulae are all driven by survivorship bias anyway."

In PG's case, he's had a chance to observe a lot of startups, both successes and failures.

But even aside from this, survivorship bias is mainly a problem in the presence of downside risk. Let's say we look at Olympic champions and find that they all trained really hard. You cry: "Survivorship bias! It could be that you will train really hard and not become an Olympic champion!" Yes. It could be. But since the risks of training hard are relatively low, you may as well go ahead and train hard if being an Olympic champion is a serious goal for you.

Perhaps I'm misusing the term or using the wrong term. Could by hindsight bias. I'm talking about when a person a group accomplishes something, looks back at what they did, and they tries to decide post hoc which of the things they did contributed to their success.

He claims you have to be unconventional to be successful in pursuits which reward independent thinking. He named at least one profession where being conventional is desirable: an administrator.