blogdex is a project of the
MIT Media Lab, and is currently hosted at
blogdex.media.mit.edu. It looks at the code of thousands of
weblogs, several times a day, and makes a note of any
URL that any of them
link to. It then
tabulates the resulting list, and presents on its front page the top 25 pages that have been most commonly linked to (recently, or throughout the life of blogdex) on the weblogs of the world. Hitting this front page is a great way to read today's hottest hits, news stories and
goofy eBay auction pages; hitting the all-time list is a great way to see a
fairly predictable list of what all
the kids link to all the time.
While blogdex has become popular as a tool for spotting trends in the nascent medium of personal weblogs, it isn't really doing anything complicated. It doesn't count the frequency of words, ideas, or forms (how often do teenaged bloggers post poetry?); it doesn't map the complex webs of weblog friendships; it just spots the easily-spotted <a> tag that is supposedly the currency and lifeblood of blogs.
Further, blogdex only indexes a blog if it is told to. If you have a blog and you want blogdex to count it, you must submit your URL to blogdex via a form on its web site. Your URL will then be checked, to see if it is really a weblog, by a human being - presumably, by whichever grad student has the unhappy duty today. What makes a blog a blog is very subjective, so there is likely a great deal of sponginess to blogdex's definition. It could be argued that the chief achievement of blogdex is one of collective librarianship. If the complete list of sites that blogdex indexes were made public, we could do something with the results, and perhaps learn more about what a weblog really is, and what kind of medium we're dealing with.
The blogdex code will soon be released as open source, according to its project page at SourceForge.