Try searching people with a simple keyword like “female” in a large network like San Francisco. It claims there are 406 results yet the tab claims 500+ results. There are, of course, around a quarter million profiles in the SF network, half of which are female. The first page of results contains around 25% redundant profiles. The second page contains around 25% redundant (within the page) and around 25% from the first page (around 50% total). The third page contains 25% new redundant results, etc. By the fifth page or so, you’ve reached the point of diminishing returns. The search has broken down to the point where it’s useless. But if you search all 40 pages you’ll conclude that there are perhaps a few dozen true results for “female”.

At least they recently fixed a long-standing bug where each page returned a completely random number of results (1-10), not the 10 which it was supposed to.

Yes, the guys who run Facebook certainly aren’t users of their own website or they’d make the search engine work so they could use it themselves. History has shown that if a company’s employees aren’t users of their own product there’s a very high likelihood that it’s doomed. And thus next-to-worthless from an investment standpoint.

Right, any site with a lot of users probably has high upstream numbers like MySpace. The challenge then is about keeping users on your site to do the search rather than navigating to Google.com or using a Google toolbar. As an example, Yahoo does a good job funneling their users to Yahoo web search once they’re done using Yahoo pages.