Tim Almond

Fascinating subject, search. How to make search engines that give people the information they need, whilst removing what they don’t need, and avoiding people "gaming" them to get their answers to the top.

The original approach to search was with keywords. Search engines would examine a site for keywords and those with the most keywords would rise to the top. This was quickly "gamed" by companies writing a keyword in hidden text hundreds of times.

Google came up with the general principle of PageRank. That is, those pages with the most incoming links by words get pushed to the top. It’s slightly modified to prevent certain forms of "gaming" (like "link farms") but works well.

It looks like Jimmy Wales of Wikipedia is setting up a new search engine which will be based on human input. This sounds a bit like the way that Digg works - users "digg" stories and that pushes them up the list. So, if you digg story #2, it’s more likely to push it up.

I also think there’s a lot that could be done on "personal search". When I search for "python", there’s a big difference between my needs and that of a snake breeder or a comedy fan. Perhaps if we had something with more bayesian logic, we could do better searching.

Tags: Internet //