I think the real question here is something like: what does the sum of the conversations I have with my friends look like to the outside world? I think that in the case of Facebook conversations, you may 1) want this conversation to be private 2) not want it to be indexed by the world; or 3) want the conversation to be relevant, but only to a small group of people.

I think that the microformat “indexing” is already happening by proxy. Sites such as Facebook, Twitter, and Friendfeed make it possible to follow a large number of conversations among close and “loosely tied” friends without a lot of effort to tag content, figure out what format it should be, or restrictive templates that force people to tag “pictures” or “conversations” or whathaveyou.

I agree with you that this content needs to be searchable to harness the power of information on the web. I also think that it needs to be so easy to use that people don’t even realize that they are tagging and categorizing and sorting that information. Automatic semantic “sorting” is happening already — and it’s so much better than 10 years ago it’s just sick — and I think we will look back on the search engines of 2008 by 2018 in much the way we look back at the Excite and Altavista and Webcrawler today. The advent of Google Onebox like searches that combine deep linking (like you’re proposing) with traditional searching will mean new ways of mashing up information that we can’t even imagine today.

I for one can’t wait for the first usable Voice/Mobile search mixup personal agent to do my bidding