New idea for an algorithm?

[John_Caius]

A short article in last week's New Scientist talks about a recent patent of a search algorithm incorporating user choice into the ranking system. As I understand it, a page receives a boost if it is clicked on from a particular set of search results, raising its ranking the next time that keyword is searched for. The idea is to reduce the impact of spam results, which might occupy top SERPs but receive fewer clicks if they contain obviously irrelevant results.

I've searched the archive and I think this is the patent they were talking about, although I'm not completely sure as I've lost the New Scientist article!

Patent Here [patft.uspto.gov]

So would this kind of algorithm improve Google's results?

[le_gber]

IMHO I'm not sure 'cause the spammers will be/are amongst the first returned by SERP, hence everybody will click on the link and increase their rank next time.

Leo

[jeremy goodrich]

Hm, that patent filing says, 1998 on it.

Reminds me of direct hit. afaik, that was their entire algorithm...seems like it's a bit dated material :)

[John_Caius]

I thought it said July 16 2002 at the top... :o

<added>

Oh, I see where you mean. Anyone got a copy of last week's New Scientist then? I can't find the article on the web version.

</added>

[martinibuster]

Sounds like part of what AdWords is already using.

[j_anstice]

Which version of New Scientist was it? And do you remember the issue date (or the cover picture?)

thanks

[John_Caius]

The UK version, something like April 10th I think (last week anyway).

[mack]

The google toolbar coulld be activly used to enable such an adition to their algo. It could count how many pages on a site you view after you click through from google serp. One page view = poor result whereas lots of views could mean a small boost in the serps.

Cant see something like this happeneing though. Or at least not for a long time.

[benihana]

One page view = poor result

maybe...
one page view = immediately finding what they are searching for