Anyone have any news of how Wikia Search is doing?

Looking in the debates about it in the past, I see the article on 'wikipedia' was used as a test.


Well today it includes the gem in the mini article-

"In 2006, Wikipedia was already the world’s largest and most accurate encyclopedia,"

And the rest of the sites listed mainly seem to be the other language versions- although Wikipedia Watch is there.

With Alexa, you must take a grain of salt, but they say Wikia Search only accounts for 3% of Wikia's total traffic:

Where people go on Wikia.com:
yugioh.wikia.com - 16%
tibia.wikia.com - 15%
inciclopedia.wikia.com - 8%
starwars.wikia.com - 5%
dofus.wikia.com - 5%
nonsensopedia.wikia.com - 3%
search.wikia.com - 3%
evchk.wikia.com - 3%
pokemon.wikia.com - 3%
halo.wikia.com - 2%

Page Rank for Wikia overall went from about 600 in November 2007 to about 300 in early January, but it's scaled back down now to about 350.

Digital Journal recently mentioned Wikia Search:



Once again, we see the media confuses the "completely separate" entity of Wikia with the Wikimedia Foundation. I wonder why they keep doing that?

The blogosphere isn't impressed, either. The XODP Blog recently opined:



I wonder how Amazon is feeling about the $10,000,000 they invested in Wales and Penchina?

Greg

He's got quite a trail of failed projects in his wake. I'm not sure why he thinks this one will succeed. I guess that's one of the hazards of surrounding yourself with people who constantly tell you how great you are. You start to believe it. Anyone else is just a troll; who cares what they think?

Seth Finkelstein's latest blog post is on this subject

http://sethf.com/infothought/blog/archives/001325.html

The Wikia search "community" (three men and a dog) seem to be growing restless with the lack of progress.

Does anyone have a feel for how seriously the sort of people who use Wikipedia routinely would take Wikia search? They're probably "use the brand leader" types who will happily stick with Google or MSN.

Well, given that Microsoft, Ask Jeeves, AOL and all those other search people have not been able to beat Google, WikiaSearch needs two things:

* Search technology on a par with Google
* Something extra

Note that excellent search technology is mandatory: people will not tolerate anything much less than Google in performance, accuracy and the ability to dig out useful stuff from the crusty corners of the Internet. So I am thinking speed and reach.

Something extra: well I guess that is meant to be the "The search engine everyone can rank" ranking which I would have said is a bias on the search results, but there should still be an underlying Google-like ranking mechanism to cope with both topics without interest and also avoiding excessive bias by interest groups. If WikiaSearch took off, you can be fairly sure that the lobbyists would manage their searches and Jo Public would be entirely unaware of how they had been manipulated. In fact, I am surprised that Jimbo has not managed to get significant funding from the likes of the tobacco industry for the product.

Ranking does not work well from the models I've seen at Amazon. You find that negative reviews tend to attract negative ratings, presumably from fan warriors. Positive reviews tend to get good ratings, unless they are blatantly awful.

I don't see how Jimbo thinks that he is going to get this to fly. He needs big money search experts (it is not that it needs lots of programmers, but they need a star who can live and breath search to code and refine the search) and he needs to inspire people to be involved - but there is simply no payback - hey, guys, try searching on "nipple rings" - I did that I did. I note that even though I've mentioned it before, searching on Curry still suggests that Curry is an English national dish. Even if they sorted out the technology, they'll have to blow a bundle on marketing to inspire people to get the user feedback off the ground.

"Come and use a useless search engine and make it a bit better please, please. Oh, suit yourself."

Incidentally, those numbers probably won't include traffic from Memory-Alpha, WoWWiki, or Uncyclopedia.

Yes, those are ranked separately, since they have a different root domain.

Uncyclopedia.org and Memory-alpha.org, for example.



Well, didn't you already know that silver-thailand.com is the best Internet resource on nipple rings? Wikia Search is there to simply confirm that.

Greg

Remember Greg that a lot of the hits for Wikia Search over the last couple of months will have been me looking at the thing for laughs.

Shortly after its disastrous launch, the most significant and entertaining contributor was Wikipedia Review. And Jimbo's page still carries a weird Wikicommons picture of that crazy top-hatted bondage fiend.

http://re.search.wikia.com/search#jimbo%20wales

Presumably now identified as Max Mosley?

Jimmy has a predictable response:


Mind you, the person that wrote this appears to be the only guy who still cares. But hey, that's Jimmy's respect for altruism for ya! I expect that the reply will be in Jimmy's usual style anyway, which seems to be a toned down and well adjusted version of Charles Manson...By that I mean that they have both mastered the concept of using non sequitors and fall back phrases which works well on his minions but alienates everyone else--Jimmy is a little bit more subtle, but if you actually listen to what he says or writes, it makes very little sense most of the time. The other difference is that Jimmy does not have a group of people that are willing to kill for him...Or does he? Watch out, Greg! WikiBiz skelter, wikibiz skelter!


Seth blogged a lot about Wikia back in January and I believe he commented on this type of thing (who would participate and why). There may have an article he wrote about it, I'm not sure. He touches on something like you are asking here.

This just in:



Google ensnares about 70% of online search.

And if you don't have the tech on par with Google, your something extra had better be something the human mind does far better than a computer. Because Google has a quarter million servers and Jimbo has 40.

Now, the only slave-labor search stuff that humans still do better than computers, is visual recognition of patterns (humans still beat computers at the game of GO). If you could convert the net into some kind of visual representation and have people pick out something on a high-def screen, like looking for camo'd-armor in the forest, you might get this to work, somehow. A glass bead game... Or the human assisted computer search in classifying galaxies by type.

But the software to convert the net (which renews itself with new pages and re-pages at a truly mindboggling rate, which something has to automatically ignore), into the appropriate visual pattern to pick something out of, might TAKE a quarter million servers.

All in all, Jimbo is being a fool about this. But nothing new to see, there.

I hesitate to suggest this, lest it give somebody ideas, but a much better model for marriage of volunteers and machines, is to use people to clean up the end product of machine-translation. A lot of this can be done even with only understanding of the target language. Of course (duh) users usually do THAT for themselves. But perhaps there's a niche for people who only understand the origin language, not the target one, and just want something presentable in a business letter, and don't want to pay for complete professional translation, but are willing to pay enough just to avoid sounding ungrammatical, and non-idiomatic.

Also worth mentioning is the fact that every time I use Wikia Search to explore the contributions of various editors, etc., the site will invariably cause my web browser to seize up. That's an interesting way to make your site "sticky".

Cuil sucks (god I wish it didn't, it's so pretty and would be awesome if it actually WORKED) but it is orders of magnitude better than Wikia Search.

So three versions of Uncyclopedia eclipse Wikia search. And yugio.wikia.com is about 6 times more popular. Wow.

I'm still impressed with the accuracy of the mini-articles that add so much to the search.

April 2007:




August 2008:

Subscriber Feedback that was published in the May 2007 issue of Fast Company:

As a random aside, not so much, anymore, or at least, that gap is narrowing immensely: 2008 US Go Congress.