If you search Google for site:www.jefftk.com [1], the sixth result is currently:

The text it's showing for the description is from a comment. This is strange, because the comments from my posts are pulled in from from Facebook and Google Plus via javascript. So in the indexing process Google is running this script on another server, computing the change to the page that the "document.write" makes, and saving the text as the post-javascript page.

I can see how in the modern javascript-everywhere web this helps people get better search results, but this feels erie. It's a violation of how I expect bots to work. Next we'll see Google having indexed pages that require form submissions to get to.

Update 2012-01-11: The example above of javascript is from a Google+ comment, and while I think it's unlikely that Google is using G+ comments on link shares to pull up results, to test this I checked whether it also could pull things out of Facebook comments. It can:

[1] Why was I doing this? I recently added a <meta name="description" content="[first 400 characters of post]" /> to my blog entries because Google was putting the header ("Jeff Kaufman  ::  Blog Posts  ::  RSS Feed ...") into the search results and I wanted to tell it where the real text started:

The change seems to have fixed some results but not all of them. They previously all looked like the image above, but now some are the beginning of the text, as 'description' should hint: