Much of Joachims’ work revolves around improving search engines. Google, the mold from which all modern search engines are patterned, uses the interlinking of the Web to determine relevancy of pages. This algorithm, known as Page Rank, essentially grades pages based on their popularity with other pages already known to be relevant to a query. The circular nature of the algorithm has a kind of insane brilliance, and works wonderfully. But all good things come to pass.

“One problem that’s at least rumored, is that the link structure of the Web is actually degrading. In a nice, organic link structure, people write pages and the hyperlinks they put in can be interpreted as a recommendation. This may be becoming less frequent -- people write pages differently, and spam links to cheat Page Rank.”

As time goes on, we’re going to need new ways to rank the pages on the Web, a task that is really the heart of every search engine. After first blush, there are some other problems with Page Rank as well. Only a very small percentage of the users of the Internet ever make Web pages, and thus the relevancy of pages is determined, in effect, by a small minority. “For example, my Mom. Never written a Web page, probably never will. But she’s on Google every day.”

Joachims’ is testing a way to rank pages that is simultaneously more democratic than Page Rank, and independent of the link structure of the pages.