Fog Creek Software
Discussion Board

Joel on statistics

Uh, Joel, if you go to Vegas and play roulette with your eyes closed, does that give you a 50% chance to win?

No information does not imply a 50% chance of success; your server has already decided if it is going to work or not.

Of course "At Fog Creek we do calculations like this all the time".

The point of the article is good though, but don't underestimate the psychological impact of a bug-ridden website.

Ulf Ekström
Thursday, November 14, 2002

"And in the long run we scientists will win", says Joel, smugly.

Breandán Dalton
Thursday, November 14, 2002

I think this horse is already dead.

Thursday, November 14, 2002

In Vegas, your best shot at winning at roulette is no better than 48%, because of the impact of the 0 and 00 (which you always lose on).

Anonymous Gambler
Friday, November 15, 2002

stay, long enough and the house always wins.

-don't know who said that!

Prakash S
Saturday, November 16, 2002

With *no information* on the likelihood of an outcome, you could assume that there is a 50% chance that you are in the middle two quarters of the thing you are trying to estimate.

Princeton physicist J. Richard Gott wrote a paper in Nature where he used this technique to estimate the lifespan of the Berlin Wall and the longevity of Broadway plays.

There's a summary of the paper here that includes a nifty interactive duration calculator:

But did Joel really have no information about the number of users trying to use the free version of FogBUGZ? I think he has to have an estimate of market size, otherwise he wouldn't have produced the software in the first place.

Michael Bean
Tuesday, November 19, 2002

*  Recent Topics

*  Fog Creek Home