[eek2121]

Curious as to how these engines accept new sites into their ranks. A big problem most of us have had is spammy results outranking everything else by building large, fake networks of sites that boost each other's rankings via interlinking. Many of the higher end networks are undetectable, as they have legitimate content and never link to more than 1 other internal site (among a mix of external sites, some affiliated with still other networks).

[marginalia_nu]

I use Personalized PageRank giving disproportionate voting power to a bunch of people whose entire identity is their passionate dislike for the commercial web. To get a really high rank in my search engine, you need to convince a those people to link to your site.

[joe_the_user]

One approach is to have human testers. When a low quality site gets high rank, you investigate in detail how that happened and downrank the linking sites.

It shouldn't be that hard to find the bad network if you're systematically investigating all the time. Google has people testing search results often.

The problem is that this is fairly expensive. But quite possibly not the largest cost a search engine would have.