There's a big battle a'brewing in the online search sector (check this video), as new human-powered "social search" networks are rising up against the machines (meaning Google/Yahoo!/Ask/Live/etc.). I recently wrote about the neo-natal state of Jimmy Wales' new search venture, Wikia Search, and gave my assessment that it shows great potential but there is still a lot of work to do before it can be a real competitor to the bot-powered behemoths like Google and Yahoo! (not to mention Microsoft's Live search). Since then I've also been testing out Jason Calacanis' people-powered search site, Mahalo.

The main difference between Wikia and Mahalo is that Wikia's results are culled by volunteers, whereas Mahalo's "editors" are paid to assemble search results by hand based on user requests. In essence, the Mahalo model is very enterprising. Mahalo has a program called the Mahalo Greenhouse which acts as a sort of brokerage for getting top-requested search-results pages built by its members. Topics that need results pages built are listed by vertical teams, such as Geography, and each team has a list of pages that need building with a price that will be paid for creation of that page, ranked by priority (and presumably based on popularity). So to use the Geography list as an example, there are roughly 50 countries on the "Most Wanted List", for which "Serbia" will net you $12 but "Angola" will only get you $10. Other search terms may pay as little as $1, but hey, it beats working for free (which is Wikia's model).

So how do these social search engines fare when compared to the algorithm-based engines? Well, honestly they have a lot of work to do, but they definitely show promise. The biggest advantage is that every link has supposedly been verified and vetted by an actual human. This should weed out the spam-link sites or otherwise unrelated results that can pollute a machine-compiled results page. For example, a search for the term "paris hilton" reveals some interesting differences. I chose this term specifically because it could mean more than one thing. I could be searching for a Hilton hotel in Paris, or I could be searching for info on the celebrity. I don't expect the search engine to know which one I meant, so the results could go either way.

So let's see how the humans fared vs the algorithms...

The folks behind the amazing user-generated encyclopedia Wikipedia, have just launched a new search engine based on a similar premise: user-generated search results should be better than results compiled by a bunch of robots. Sounds interesting, promising even, so I thought I would put it to the test. : )

The challenge: search for "pattycake.ca" using the term "pattycake" on Wikia, and compare results with Google, Yahoo!, Live Search, Ask, and a smaller player called IceRocket. The reason I chose this term is because I built and maintain Pattycake.ca, my wife's healthy baking and lifestyle blog, and we often check our search ranking against the "Sexy Pattycake" adult website, a children's singer/entertainer known as "Miss Pattycake", the perpetually-under-construction "Pattycake.com", and a bakery in the US called "Pattycake Vegan Bakery".

So, how does Wikia fare against the competition...? Here are the results, in order of best to worst, with respective ranking (in page and overall order):

1. Yahoo! (p1/3rd)
2. Google (p1/4th)
3. Ask (p1/7th)
4. IceRocket (p2/18th)
5. Live Search (p6/55th)

So where's Wikia in the list, you may ask? Well, unfortunately it doesn't even make the list because it didn't find Pattycake.ca at all! Surprisingly, it only returned 95 results in total (compared to 691,000 for Yahoo!, 432,000 for Google, and 56,700 for Ask). And of those 95 results, I could only get 73 to actually display. Hrmmm...

So it would seem that the little gnomes at Wikia have some work to do before they'll be unseating any of the encumbants. But then again, that's the model. The search ~should~ get better over time, as more humans filter and finesse the results. Till then, unfortunately, it isn't worth using. And that's the problem: if nobody uses it, it won't get better. If it doesn't get better, nobody will use it. Doh!