Interesting, while retweeting, back-linking and posting here and there I decided it was time to throw a link into my Google Buzz panel. I did so and at first didn't notice anything interesting.
What I pasted into my buzz box was simply the title of the post and a bit.y generated link from, bit.ly. It was from a page/site that validates to XHTML + RDFa.
What happened next didn't register until later today when it didn't happen. Later today I wrote an article, posted it and decided again that I was going to post it into my buzz box.
Just after posting todays link I looked at yesterdays link and noticed it was formatted differently, nicer looking, more descriptive - you know - better for the user.
What happened yesterday was, moments after I pasted the link into my buzz box the words "Loading..." displayed and thereafter a rich snippet containing the title of the post and the first paragraph as well all this in my local buzz report.
Today's link didn't, but simply said "no images". Both posts had no images. The page todays link is on doesn't validate as of yet.
To make matters worse, I checked if http://www.w3.org validated and it did. I then pasted that same URL into my buzz box and the same thing happened - buzz displayed a rich snippet with the title and a few sentences describing the page.
For sake of argument I pasted in http://bit.ly of which didn't validate to XHTML 1.0 Transitional but buzz did display the title of the website.
I just moments ago, also decided I wanted to post this post into my buzz box and although this page didn't validate to XHTML 1.0 Transitional it did display a rich snippet title and opening paragraph. The interesting point is that it took the first lines of this post as the opening paragraph and nothing prior this post that is on the page. I then checked the meta description of this page and it too, had a snippet of the first paragraph of this post so it seems Google buzz is able to take information from the meta description tag just as other websites like Facebook, Bit.ly and so forth are able to.
I'm not sure if it makes it any easier for the algo to determine what data it should snipper unless the based on whether a page validates or not.
What do you think? Has anyone looked into this? I haven't been able to find much of anything related to this yet.