Selection bias and post hoc rationalization. You can always frame success in any way you want- it'll sound good, because if the company's successful then your a

Selection bias and post hoc rationalization. You can always frame success in any way you want- it'll sound good, because if the company's successful then your argument proves itself. How many successful companies don't fit the mold that was just described? I can think of many examples that had fairly straightforward road to success: Facebook, Dropbox, Stripe, usw. The real measure of a theory is how successful it is at predicting whatever it predicts. So out of all of the companies that you predicted could become the next "Microsoft of X", how many did? If your method doesn't perform better than randomized selection, then it's worthless. I suspect that's the case here. Of course, the law of large numbers combined with the high returns offered by tech companies means that even if investors aren't performing better than a randomized process, they feel like their theories about what makes a successful companies/entrepreneur are valid. Here's an experiment that will never happen, but that I'd love to see YC perform: for a given batch, select who makes it to the batch randomly from the pool of preselected applicants (you want preselected applicants and not all applicants because it's fairly likely that the 90% of people who are applying with their "Facebook for cats" idea that they came up with after beers 2 hours before the submission deadline won't make it. Or will they?). Then, look how that batch performs. It wouldn't be surprising if the batch performed no worse or no better.