This article is based largely on the assumption that priority number one will be to port everything to Windows and I think this is an absurd assumption to make.
Are we to believe that Microsoft is willing to risk throwing away $100M+ in order to migrate this thing to Windows?
Don't forget that Microsoft ran Hotmail on Unix for years. I imagine by the time they got around to porting it to Windows they were about ready for an upgrade anyway. (Plus, Windows was finally getting to the point where a project like that could even be considered.)
My guess is the meat of this deal was about bringing in some new IP and some bright engineers- not about running some search engine that almost no one has heard of. In that case, the platform it runs on is largely irrelevant.
http://www.techcrunch.com/2008/07/02/interview-with-barney-p...
Maybe I don't know what I'm talking about but having half of Live search running off Windows search and the remaining part being handled by Powerset's unix boxes sounds silly. Or at least I will think I am seeing pigs fly if it every happens.
The article you linked to mentions that only 5% of searches contain natural language elements. This integration they talk about could be as simple as identifying these searches and passing them off to the Powerset stuff in the background.
I don't doubt that Microsoft will eventually want to port everything to Windows, but I think it's a false assumption that this is going to be a top priority. They've made repeated statements about their desire to see this put to use quickly and speculation about a lengthy and complicated port seems to directly contradict that.
