Some years ago , I wrote a short "proposal" (more like food for thought, than proposal) on how we could create a distributed search engine.
The idea is simple. Each content/data source would expose a search API, and would also suggest other relevant/trusted sources that expose the same search API -something like a social network of content/data sources. What I wanted most, was a search engine that would query my friends blogs and their friends blogs, my "slice of the web".
It never got traction, it didn't even start a conversation.
A lot has changed since, but I feel that the #indieweb movement might be interested in the idea. It should be relatively easy to implement it. I would start by extending opensearch -but right now, I have no time to start such a project, not even as a side-project.
So this is the Distributed Query System proposal. 
 I only did some limited editing to the 2004 text, mainly typos. I left the rest of the document untouched -if you've been around long enough, you will probably have some fun reading mentions to altavista, dmoz, etc. Don't be too judgemental on the details, I was younger back then :-)