September 16, 2006

The rise of Wikipedia

Tim Bray observes that Wikipedia is showing more and more on top of Google searches, and while it's a phenomenon I have observed as well, his interpretation is quite surprising:
Cast your eyes back across those web addresses. What are your chances of guessing them? Of remembering them? Of writing them down accurately? If you bookmark them, how confident are you that they’ll be there after the next site re-org?
According to Tim, Wikipedia is getting high rankings because its competitors have complicated URI's...

I think a more reasonable answer is that these various other Web sites that, Tim deplores, "don't get enough attention", have very poorly structured information. This is not Wikipedia's nor Google's fault. Nobody links to them because, just like Tim experienced, it's very hard to find the correct information there.

What I like about the effect that Google, and more recently, Digg, have brought to the Internet is that they turned the Web into a popular meritocracy. If a page gets a high score, it's because a lot (a lot!) of users have expressed that they found this particular page or site useful by casting a vote for it.

As for standardizing URI's... I do remember an effort from the W3C about ten years ago to do just that, but the truth is: except for hackers, nobody cared about URI's back then and nobody cares about them now.

Posted by cedric at September 16, 2006 08:18 AM

Comments

I think you're mistaken: a *lot* of people care about the URIs since it infuences your page rank. eBay, for example, spent millions and millions of dollars cleaning up their URIs and pages, to make them search engine friendly. What was http://listings.ebay.com/83889.html became http://crafts.listings.ebay.com/Bead-Art_Beaded-Flowers_W0QQfromZR4QQsacatZ83889QQsocmdZListingItemList. As a result, eBay started showing up a lot more in google results. The SEO industry is all about optimizing your URIs and page content to get you higher ranking. Of course, some of its members are pushing the limits a bit, and constructing link farms, but that's a different story...

Posted by: JM at September 16, 2006 09:41 AM

I have to agree with JM. I worked on a shop with a very good site (mine!) but a terrible ranking and almost no pages viewed. We changed the url, the links, and the tags. This one week work gave us +/- 10% increasing sells. And the more you have a good page ranking the more you increase it.
Look at amazon, with all its links: it a google trap!

Posted by: G; at September 16, 2006 03:07 PM

Since you can't track me back, notice I replied your entry at http://nicolas-delsaux.is-a-geek.net/wordpress/index.php/archives/2006/cedric-et-la-wikipedia/

Posted by: Nicolas Delsaux at September 17, 2006 12:16 PM

Cedric has a point. True, if I'm looking for a word, say, Guacamole, Google will increase the score of a page that contains the word in its URL.

But a high page rank weights much more. While placing the word in a URL might win you a few positions, a high page rank will win you a few dozen positions. The page rank goes for the root domain, and since a lot of people link to Wikipedia, its page rank should be high.

One last point is the lexical analysis of links. A log of pages talking about mexican food might have a link to Wikipedia's Guacamole entry. The link has the word in the URL, but probably also in the title and in the link text. Finally, the entry will also talk about mexican food and have a lot of common terms with the original page. This combination means jackpot.

Posted by: Tiago Silveira at September 17, 2006 01:28 PM

Complicated URLs might indicate a disorganized site but I don't think that's the only or even primary reason Wikipedia gets so much Google weightage. Lot of people link to Wikipedia because it's good pretty good info on almost all topics. I'd be careful about relying on Wikipedia to figure out who killed President Kennedy or to learn the Pope's real views on Islam but alteast for technical info or other non-controversial info, it's pretty darn good. Tim has a point that if someone mucks with obscure info like the population of Saskatchewan then you are out of luck but for most topics it's undeniably decent. Also read Nick Carr's point of view at http://www.roughtype.com/archives/2006/08/the_centralizat_1.php

Posted by: Raghu at September 18, 2006 10:18 AM

...because it's GOT pretty good info on almost...

Posted by: Raghu at September 18, 2006 10:19 AM
Post a comment






Remember personal info?