Fog Creek is currently demo-ing their new bug/task tracking system FogBugz.
Most of the systems which we currently see (for example Jira, Bug tracker) are based on the fact that they are good at reporting, maintaining and tracking (somewhat) bugs and tasks. That's it. The role of a bug/task tracker system usually ends there. (if you been making more use of a bug tracking/task management system, drop me an email).
There is where FogBugz is different from rest of the systems. Probably the most interesting part of FogBugz is Evidence-Based Scheduling, which uses a statistical technique called bootstrapping (a variation on Monte Carlo) to determine the probability that you'll ship on any given date. In short, this is how it works. You estimate features as usual. But then, instead of adding up all the developers estimates—instead of taking them on faith—FogBugz does a Monte Carlo simulation looking at what speeds developers worked at in the past, vis-à-vis their estimates.
In Joel's words:
Using that same distribution of probabilities that we had in the past and if we run a simulation of 60 futures each of which will occur with equal probability, what we get, instead of a date, is a probability distribution curve that shows the probability that the product will ship on such-and-such a date.
Now thats what i call an awesome sfotware tool, which really stands out. Well how reliable and accurate FogBugz is, is another question.
I think i got to try it to find out!!!