3 ms·
This is one of the few cases where this is worth repeating: correlation is not causation. PG is definitely one of the foremost researchers in the realm of entr
by FD3SA 13y ago
This is one of the few cases where this is worth repeating: correlation is not causation.
PG is definitely one of the foremost researchers in the realm of entrepreneurial success factors, but it is important to step back for a moment when analyzing such things as verbal accents and "Zuckerberg likeness" correlating with failure and success, respectively.
Just as Noam Chomsky criticized Peter Norvig because of his focus on statistical methods versus fundamental models, I would suggest that inferring success based on statistical observation without an underlying model can become a confusing and unrewarding process.
Statistics is a tool to test fundamental models, not a model to explain phenomena all in itself. As such, I would guess that founder success is more likely based upon mundane traits such as intrinsic motivation, intellect, experience, access to capital and key personnel, and most importantly, luck. We see this time and time again in superstars such as Elon Musk, Jeff Bezos, John Carmack, Bill Gates, etc.
Extremely smart people are more prone to analyzing every tiny variable, which sometimes causes them to give additional weight to trivial factors in a complex equation.
- jonkelly 13y agoPG offered a clear underlying model in this essay. His model is that a) founders need to sell well to succeed and that b) it's hard to sell well if you can't be easily understood. You have a "guess" that mundane traits are more important for success. He has a theory and data that support his theory.
- vasilipupkin 13y agoThat's fine just as long as what he did wasn't the following: see some small number of founders with a fairly strong accent, say < 30 or < 50 or < 60, then observe that those startups happened to do poorly and then extrapolate a theory from those observations to explain what is happening. I am not necessarily disagreeing with his thesis, but in statistical data analysis you have to be very careful about drawing conclusions from small samples when the underlying distribution is non-normal, with fat tails
- tlb 13y agoThis is an epidemiological observation, and those rarely indicate a particular mechanism. Even with giant populations and teams of scientists we have only tentative and shifting theories for why, for example, Americans are fatter than Europeans. Controlled studies of startup founders are not practical.
- clavalle 13y agoBut pg has a pretty good idea of the 'why' of this observed phenomena so there is your underlying model: it is fundamentally important for CEOs to be able to communicate clearly and effectively with stakeholders. Your 'correlation is not causation' mantra seems misplaced unless you can argue that effective communication is not fundamentally important.