What miracle you will need

Yesterday I heard Elad Gil talk about how startups that require multiple miracles to make are usually doomed for failure due to the low probability of success of multiple events with low probability of success. Although I generally agree with his premise, I think that there are a couple of points worth considering.

First, miracles are not well defined events. To me any startups needs multiple miracles, planet alignments, and strikes of lightning in order to be successful. Jokes aside I do think that miracles are not a clearly defined concept so maybe startups needs one big miracle and 10 small miracles to become successful. I think that what constitutes multiple miracles is about perspective.

Second, miracles are not independent events. Although the probability of any given miracle is low, the probability of every miracle should be evaluated in terms of likelihood of a team making it happen at that specific time and under those specific circumstances, while taking into account everything that they’ve done together until that point. This bayesian approach doesn’t disprove the point that Elad is trying to make. Instead it tries to account for the fact that startups that are able to make it past their first “miracle” seem to become increasingly good at making more miracles(: