If you look at a sample of people who started companies and find a few repeat winners, you probably will have found repeat winners produced by chance. But if you look at a repeat winner who keeps producing wins after you've started watching them, the same argument doesn't apply -- odds are overwhelmingly against the hypothesis that they continue winning randomly.

If X happens to 1 in 100 people, then there's a very high chance (~1) that X happens to 1 of the 100 people, but a much smaller (~1/100) chance that it will happen to person #56, for example.

The parent post is saying that "starting to watch someone" constitutes naming the #56.