Engineering a software product involves many tough decisions, both technical and product. You rarely have enough information to be sure of which technology is best, what features customers will like, etc. And we spend a lot of time debating these tough calls amongst ourselves--- engineers have different intuition, and product people have different guesses about what customers want.

Fortunately, one of those tough decisions IS NOT which web browsers your product needs to support. Compared to a typical product decision, it's stupidly simple--- you look at your server logs to see who is using what browser, and then you simply calculate what % of users you are screwing by failing to support Browser XXX. (And when you work for a certain popular search company, you have pretty good data.)

So when I need to have a technical debate about this (and it's sad that there ever is a debate, but whatever), it's one of the few arguments that I feel I can be confident about. Discussions typically go something like this:

Engineer: The product works great!
Me: It doesn't seem to even load in Internet Explorer 6.
Engineer: Oh sure. But it works great in FireFox!
Me: Uh huh. We have 4 times more IE6 users than FireFox users.
Engineer: But I hate IE6! Who uses IE6?
Me: Outside of a 50 mile radius of Palo Alto? Everyone.