Getting an open source search engine would definitely be a great thing. Best thing would be if a cooperation were programmed in, so that we can have different search engines based on the same platform which cooperate at some places (especially spidering), but each serve somewhat different niches (for example, search engines for specific subjects, separate ones for entertainment, shopping and research, for fringe languages etcetera). And I definitely agree the Wikimedia Foundation should stand behind such an initiative.
However, I doubt whether the Foundation is actually the right group to be the organizer or main supporter of such a development. We have our mission statement, but we also have our background. And to me our background and strongpoint is definitely with 'massive volunteer collaboration'. If that's not the best mode of operation for this project, we may not be the best ones to do it.
My feeling is that if you put something like that into the workings of a search engine, what you end up with will be more like dmoz than with Google. If we want a real search engine, there will be a massive task of creating and maintaining the software, but hand-improving the database is not something that works very well with the 'thousands of volunteers' approach. In short, I think that an open source search engine is definitely something that should be worked on, better start yesterday than today, but I think it's a task that would be more fitting for the likes of the Free Software Foundation than for the Wikimedia Foundation.