Google Weblog

May 20, 2003: Myths

Here's what Google representative Urs Hölzle listed as some Google non-issues at Search Engine Day:

Metadata. He thought meta data was fairly useless. Users don't enter it. He thought the semantic web project would fail. He said users around the globe will not work to enter descriptive data of any kind when there is no user benefit.
Deep Web. This is the idea that "oh no! only 2% of the web is indexed!". He said when he asks for "San Francisco Airline" he doesn't really need a list of ALL possible flights to San Francisco. He just needs a page about flights to San Francisco. Everything you really want is 0-1 steps from an indexed page.
P2P web search. This solves the wrong problem. Crawling is easy. Ranking is hard. And the system can be hacked and spammed.

If anyone else has a report from Search Engine Day, feel free to pass it along.

Another common myth these days seems to be "oh no! the blogs are ruining all the searches". A few years ago this was "oh no! the mailing lists are ruining all the searches". Two problems with this:

First, it's hard to make a general comment about all blogs or all mailing lists. Yes, there are many pages with less-than-useful information, but some have quite valuable information and it would be a shame to lose it.

Second, often when you get some bad results, it's because there are no good results. Removing the bad results will only leave you with worse ones, or no results. For example, a recent Times article claimed that because Google had lots of results with 9/11 conspiracy theories and none refuting them, "search engines aren't working as they should". The possibility that no one has yet spent their time debunking all the conspiracies didn't seem to cross his mind.

I'm sure we all agree that Google isn't perfect, but taking out all the blogs seems more like medieval bloodletting than a solution.

Posted by Aaron Swartz on May 20, 2003 09:12 AM