Quartz’s Christopher Mims has written a piece about Blippex, a search engine which weights results based on how long people spend on a site and how many times its users have visited (as distinct from Google’s PageRank, which weights more on how many other pages on the web link to it).
Blippex’s algorithm, called DwellRank, decides relevance based on how long users spend on a site and how many times Blippex users have visited it. Researchers at the University of Massachusetts Amherst have, independently of the Blippex team, established that the amount of time someone spends on a web page or document is, not surprisingly, a pretty good measure of how important and relevant it is (pdf).
The consequence of this is that the only pages in Blippex’s search index are those its own users have visited. It has only two million pages, compared to the tens of billions of pages and trillions of links that have been indexed by Google. (There are many more links than pages on the web because most are spam, duplicates, or unhelpfully different from one another.)
The result is a search directory that’s currently only as good as the (mostly tech-focused) people who are early adopters of new web services. That means it’s great for things related to computer programming, pretty good for recent events, and nearly useless for more obscure search terms.