OT here. Didn't know that this had been posted - here's my discussion post: http://news.ycombinator.com/item?id=3649114 http://news.ycombinator.com/item?id=3649

OT here. Didn't know that this had been posted - here's my discussion post: http://news.ycombinator.com/item?id=3649114 http://news.ycombinator.com/item?id=3649114 Neglecting the 100- and 400-year problems with leap years, that seems to be a good frequentist estimate. But if you look at it from a Bayesian perspective the question is how much you're interested in incorporating into the model. And there's quite a lot of (Also, the problem as originally stated in 140 chars isn't really well-defined - though if you want to get cheeky with an answer of 100%, let's think about the probability for an unspecified NYC startup, rather than foursquare!)