I don't see how analyzing comments on ex-post known successful businesses can tell us anything about anything -- it (at least) needs to be compared to how comme

I don't see how analyzing comments on ex-post known successful businesses can tell us anything about anything -- it (at least) needs to be compared to how comments on average startup proposals do, to see if there is any predictive power. If, for example, successful startups had 20% positive comments vs. 10% positive on average, and the difference was significant, that would suggest that HN shows awesome crowd wisdom -- even if 80% of people here hate everything :) Accounting for confounding variables would be nice too (for example, if field X is generally more liked by HN than field Y, equally positive feedback on a startup doing X would tell us more than that on a startup doing Y). Still an interesting idea, I just think that people should not forget that "data science" is a subset of statistics, and any kind of data analysis is not really useful without at least some familiarity with it.