Mahalo – the latest and somewhat controversial entrant to the search engine space – is reported to employ about 40 editors who manually edit search results in order to provide spam-free and therefore better results for its users. I can’t help but think that these editors would be better utilised by allowing users to rate search results and having the editors moderate those ratings such that people cannot game the system.
By crowdsourcing ratings in this way, perhaps Mahalo can ease some of the concerns raised regarding the scalability of its current approach of using editors to edit results and moderate pages crafted by members of its Greenhouse scheme, which by the way does employ crowdsourcing - but in a different way to how I am suggesting.
Crowdsourcing ratings and in effect results quality also fits somewhat with an idea that I had a while back, for a social search engine where users could own search phrases by being one of the first to rate results generated for them and then share in any advertising revenues resulting from those phrases.
My thinking was that the first person to rate the results would get the lion’s share, followed by the next and the next after that etc – up to an appropriate number. This way the first few people to rate are rewarded relatively well. To encourage other users to vote, the remaining revenue is given to a random user that rated results for the phrase.
The benefit here is that as long as ratings are monitored and moderated, it should result in better results. After all, 75-80% of searches are repeat queries - going by Udi Manber’s revelation, as reported on Read/WriteWeb, that 20 to 25% of the queries Google see today have never seen before.