April 22, 2008
Wikia Search - When You Have A Wikipedia, Everything Looks Like An Edit
I continue to be unimpressed by Wikia Search, the Wikipedia-model
search engine. They've
released a bunch of new "social" features for
editing and annotating specific results, which are not in and of
themselves bad things. Yet I keep wanting to say, but, but, but,
there's no real search engine there. Nobody gushing over all the
pretty buzzword-compliant aspects seems to care that when putting
lipstick on a pig, underneath, it's still a pig. Maybe they think a
sufficient amount of lipstick on the pig emergently create a useful
As I understood the initial Wikia Search concept, volunteers were
supposed to build Jimmy Wales a search engine for free, that the
venture-capital backed start-up Wikia Inc. could then monetize with
ads (or presumably flip in a sale if the opportunity arose), because
this would then prove The People open-source amateurs can challenge
the closed proprietary Google elitists. Or something like that -
obviously it wasn't stated so bluntly.
A major problem here is that search engines require very specialized
expertise, for which companies are willing to pay big money, so almost
nobody wants to give it away for the good of Wales's
gold-plated washing machine, I mean, humanity.
But the latest iteration of Wikia Search seems to be trying to use
the very poor search technology as a seed page for human-edited
results. That is, an unpaid, voting-driven,
I suppose if the workers aren't paid, anything at all is profit.
But I can't help but think this is turning into a proof of my theory
that Wikipedia is basically a weird thing, which doesn't export the secret of
the fountain of free labor.
By Seth Finkelstein |
posted in wikia-search
on April 22, 2008 11:59 PM
[Seth Finkelstein: I have removed this comment because Jimmy Wales made a reasonable-sounding claim of libel, and otherwise the comment has only common criticism. This was a voluntary kindness on my part, not a reaction to a legal threat.]
Having built a voting based social news engine (the second largest behind digg), I can tell you voting is the WORST way to figure out if something is good online.
Voting is way easy to game.
all the best,
I agree with you. Wikipedia's search engine services are not good enough. Normally I type keywords in Google, and then add "wiki",I will get good results there.
LOL. WMF has 120 servers and Google has 450,000. Jimbo's cadre of teenagers cannot compete such raw computer power in looking through the web. Humans can and do compete with computers in certain social functions which require massively parallel processing with small data-input. But a web-search isn't that kind of task.
Let Larry talk, Seth... Let *us* decide if the part you did not regard as credible libel was "common criticism"... (now you just made me curious...)
Milton: I don't think the idea, in general, is a bad one (using people to improve search). I just don't think it could or should be done the way Jimmy is trying to do it (Seth is right on this).
About $14 million has been wasted - er... - invested in Wikia, Inc. so far. I will be stunned if another dollar is sunk into that organization after Jimbo's most recent 5 months in the news, on several fronts.
Jason: Good insight.
Milton: I think the idea is that since so much is free, he only has to attract a relatively small audience to be profitable (but it's unclear if that works here!).
Delia: Sort of defeats the purpose.
Gregory: Do you have any idea what goes on in high finance? Those people put all sorts of things on company expense accounts. That's Wales's problem in part, coming from and dealing with, that culture. If anything, the moneybags are going to laugh at him not for trying the expenses, but for failing.
re: "Delia: Sort of defeats the purpose."
Seth: I would have asked Larry to post again without the credibly libelous part. I mean, it's not like all comments bring original criticism here...
P.S.not a big deal, though (I just thought I'd let you know what I thought) D.
The search technology is not that bad. The problem at the moment is that the index quality is poor. Google and other search engines tend to spend a lot of time working on index quality and that's before they even have a viable index. I think that this area (index quality) is where wikiasearch is weak. Unless it hires some real search engine talent it is going to remain a more current version of Dmoz.
There are massive differences between hobbyist search engines (wikiasearch) and genuine commercial search engines. Genuine SEs tend to be very highly organised with clear chains of command. And the expertise needed for developing a good SE is rare. One of Wikipedia's strengths was that anyone could contribute. The problem with wikiasearch is that it is trying to humanise what is in reality a highly automated and self-contained system that depends on repeatable, mechanised processes.
Wikiasearch may have a future but it just cannot compete with generic SEs like Google. I think it would be better to explode wikiasearch into niche search engines that would take advantage of Wikia's existing clientbase.
re: "what is in reality a highly automated and self-contained system that depends on *repeatable, mechanized processes*" [my emphasis]
I think this is a serious limitation of search engines. Plenty of things are just not amenable to using "repeatable, mechanized processes." There is definitely a place for people as far as improving search.