Comments on: IDEA #77 – Social Search Engine (take on Google)

By: Zmunz09

Zmunz09 — Wed, 05 Oct 2011 18:28:00 +0000

Google +1

By: alpha

alpha — Tue, 09 Oct 2007 20:45:27 +0000

I’ve been thinking about how I read RSS feeds recently, and I think a social ranking of feeds would make it so much easier to find good content… I mean if feeds and posts are socially tagged and ranked and added…

Because I don’t often search for content in blogs, I read them… it would be good if everyone else doing that helped out a little.

By: Jordan Mitchell

Jordan Mitchell — Tue, 25 Sep 2007 01:13:29 +0000

Hi Ivan, well I believe moving more into that direction is a good thing, but then again I doubt you’d list technology/ranking as your biggest challenge right now. I think adoption (network effect) is the biggest challenge we all have. For us it is anyway!

By: Ivan

Ivan — Fri, 21 Sep 2007 11:05:34 +0000

Jordan Mitchell: this is what yoorl.com is doing (in a way). The relevance of a site is determined implicitly, depending on how well that page spreads. Maybe moving more into that direction would be the right thing to do.
What do you think?

By: Jordan Mitchell

Jordan Mitchell — Thu, 20 Sep 2007 15:30:49 +0000

I’m convinced that the next generation of search WILL be social search, but fully believe that the vote will be passive/implicit, not active/explicit.

There are already well-funded companies offering social search based on explicit votes — either via a thumbs-up/down mechanism, bookmarking, manual ranking, etc. That’s not working.

People don’t want to build search results. But I think social search will work if they can vote passively. In essence, the users are the Web spiders and their behaviors form the indices.

By: alpha

alpha — Thu, 20 Sep 2007 04:13:25 +0000

You’d need a strong algorithm to give the users the power to actually find good results. Or you wait for the users to supply results and grade them, but this is a social directory – not search.

Also, when I search I browse all the links I think are useful (tabbing them in firefox) when I’ve finished scanning the search results I start looking through my tabbed results. So this is essentially providing Google with all the results I think are good.

There would be too many for me to find them if there wasn’t already a strong algorithm.

But you could use a basic algorithm to toss the crap. And then a sort of monte carlo algorithm to provide ‘random’ results to the user (random in order – because order is very important in search). Then weight the results as users provide click through… you could also give the click through a frame or similar to provide input for the result from the user, otherwise you would almost only be relying on click through rates… or users rating sites before they have seen them – based on snippet?

You’d need a well seeded algorithm to allow new sites to perform as well as old well visited sites – but this could provide a vulnerability in the search algorithm that many sites would enjoy exposing.

Ok, I think it can be done.

By: Ed Kohler

Ed Kohler — Thu, 20 Sep 2007 02:47:05 +0000

Social search engines sound like a good concept, but I’m convinced that anything requiring work on the part of searchers is doomed from the start. Winning search is based on calculating the actions people are passively making unrelated to search such as link popularity, on-site factors, search behavior, etc.

But I’d loved to be proven wrong. Anything that makes search better is a winner in my book.