People search engine Spock is still in invitiation only beta, but I had a chance to review the product. One of the first things that becomes evident is that all of its search results are about people. This is a vertical search engine through and through. This approach gives Spock substantial leverage. For example, if you enter in the query "NFL", you will get a list of people that have been tagged "NFL".
Spock appears to use a combination of algorithmic tagging, where it develops tags for people based on web crawling, and people powered tagging, where users can add their own tags. So the crawler will be a driving force in populating the search engine data set pre-launch, but over time input from people will have an increasing impact on the results.
As this article by Alex Iskold on ReadWriteWeb observed, the result will be that famous people will receive far more tags than others. This leaves the question as to how much information you will receive on the less famous.
This question aside, Spock provides a remarkably clean and efficient people search engine. The results are easy to interpret, and you can navigate among people who are "related" (i.e. that share the same tag, as in my NFL example above). I look forward to its public release.