If you have a hundred would-be Warren Buffetts who each make one investment per year, and a hundred would-be George Soroses who each make dozens of investments per year, the variance on the Buffetts' investments will be a lot bigger than the variance on the Soroses'. Consequently the richest investors out of the whole group will almost certainly be Buffetts, even if the two groups do equally well on average. In fact, the average of the Buffetts has to be quite a bit worse than that of the Soroses in order for a Soros to come out on top.

That's the argument Taleb is making: that Buffett's success is weaker evidence for the expected return of his investment strategy than Soros's success is for his.

Taleb is correct. If you think he's wrong, you don't understand his argument.

Now, you (and Buffett and Munger) argue that there's an additional reason to believe that Buffett's strategy is a good one: because fewer decisions means that each decision will be smarter. Well, you could be right. Historically, though, human beings are pretty bad at distinguishing good investment strategies from bad ones by logically analyzing their premises. So the statistical evidence Taleb is discussing counts for more, in my book.