Two new badges today:
> > ## [Reversal](http://stackoverflow.com/badges/95/reversal) > > > > Provided an answer upvoted 20 or more times to a question that was downvoted 5 or more times. (gold)
> > ## [Pundit](http://stackoverflow.com/badges/94/pundit) > > > > Left 10 comments with a score of 10 or more. (silver)
I think I've discussed the Stack Overflow philosophy of badges ad nauseam by now, but the Reader's Digest condensed version is this: badges exist to encourage positive behavior -- both in the sense of contributing to the site and between the users participating.
In addition, when choosing new badges, I try to explore new dimensions, rewarding people for behavior that isn't necessarily accounted for within the existing reputation or badge system. Pundit is the first badge based on comment upvotes, for example. And Reversal is unlike any other badge to date, as it rewards, as TheTXI calls it, the "diamond in the rough" -- taking a bad question and miraculously turning it into something positive by providing a great answer, a phenomenon that continually amazes me:
Usually, it’s garbage-in, garbage-out. Bad questions beget bad answers. If you sort the Stack Overflow question list by votes and sink to the bottom of the barrel, you’ll find some truly horrible questions, as you might expect. But you’ll also find something you probably didn’t expect — some amazingly good answers! Now, these are questions judged by community votes to be of so little merit that I’d usually delete them without a second thought. But I can’t, because a well-intentioned Stack Overflow user has poured his or her heart into an incredibly insightful and helpful answer. Deleting the bad question would bury the good answer, too. It’s the web forum equivalent of turning lead into gold, and it happens far more often than I ever would have predicted. (This is also the reason why voting on questions should be, and is, independent of answer votes.)
I chose these two new badges with input from the meta feedback site, and my own observations of the underlying data. Enjoy!