One factor that gets often overlooked. In America, at least, Android barely mattered until ~3 months ago, because prior to early October there was only one way to get an Android phone: switch to T-Mobile.
I know for a fact that Sprint has traditionally been popular among casual software developers... mainly, because they've usually had better phones (especially with regard to the first Palm phones circa 2001-3), data service, rates, and TOS than the others (Verizon=expensive control freak Nazis, AT&T's phones just plain sucked pre-GSM, and T-Mobile's data network was a cruel GPRS joke until very, VERY recently). 2 years ago at JavaOne, I'd conservatively estimate that *at least* half the guys there had Sprint logos on their phones.
More purely anecdotal evidence: check out the explosive growth of Android-related topics at xda-developers.com, and other sites like it, that have become de-facto "Android sites" over the past 2-3 months as practically their entire user base (admins and all) have jumped ship to Android as it became available from their carriers. Even moreso, when you look at the boards for the highest-end WM phones (like the Touch HD) and see that the #1 topics everyone cares about are the ones like "does (the guerrilla) Android (port) support ${bluetooth|gps|camera|etc} yet?"
Put another way, by next fall, the Android software situation is going to be a LOT more mature than it is now. And Microsoft will be filling its drawers when they throw the release party for WM7, and nobody really notices or cares.