Via Network World, a report that appears to validate something I have long suspected: what you find, when you're searching the web, depends very heavily on which search engine you use. That is, rather than Google, Yahoo, Microsoft Live, Alta Vista, Ask, etc providing overlapping views of the central data repository that is the World Wide Web, each returns a particular sampling of that data, which can differ to a startling degree from the other samples.

For instance, the study compared the first-page searches from major engines and found that on average:

• 69.6% of Google’s [first page results] were unique to Google.


• 79.4% of Yahoo’s were unique to Yahoo.


• 80.1% of Live’s were unique to Live.


• 75.0% Ask’s were unique to Ask.

All in all, according to the survey, only 1% of results appeared on the front page of all four search engines.

Now, the study was commissioned by Dogpile, a "metasearch" engine that combines results from many different engines -- and therefore has an obvious self-interested stake in the idea that no single engine produces a balanced view. But it was conducted by real academics, at Penn State and Queensland University of Technology in Australia, and it appears to be on the up-and-up. It involves comparisons of results on more than 19,000 separate search queries. (Oddly, the Network World article has no link to the study itself; via Dogpile, I found a summary of it here and the full text, in PDF format, here.)

Let's allow the possibility that, because of Dogpile's commercial interest in exactly this research finding, some other study will come to a different conclusion. Still the general topic of how computerized "intelligence" -- in this case, the reliance on search engines to produce data people used to have to remember or look up in books -- affects "real" intelligence will only become more interesting. The implications of this report, about the separate realms of knowledge we discover via internet search, will be worth coming back to.

Bonus

I was going to add a link to a New York Times column I did on this subject three years ago, but it's now behind their firewall. For those with NYT access, it is here. The relevant passage (and I assume I have the right to quote myself) was this:

When the computer age began, some people warned that the rise of word-processing systems would mean the decline of skillful writing. The idea was that computers would make writing so automatic and easy - yeah, sure -that fine points of thought and language would be buffed away, leaving depersonalized, machinelike prose.

In retrospect, there was nothing to worry about. Books, articles and lectures are now as good, and bad, as they have ever been. One area that technology has obviously changed is personal communication, through e-mail and instant messaging. But there, its main effect has been positive, in reviving what had been the moribund idea that people, even teenagers, could stay in touch through written as well as spoken words.

The broader ways in which computers will change our modes of thought and interaction are hard to predict, so any early indicator is interesting. Electronic calculators, for instance, have already eliminated one ingredient from the traditional concept of being "smart." From the invention of arithmetic until about 1970, speed and accuracy in handling numbers were a mark of intellectual distinction. Now computational skill is a parlor trick because the most gifted human prodigy cannot keep up with the cheapest hand-held device.

A "sticky" mind, one that retains names and ideas and retrieves them on demand, has traditionally been a proxy for one kind of intelligence. But how long will that matter, as search engines grow faster and more precise? We'll know the change has come when a schoolchild with Google can knock off any "Jeopardy" champ.

Long before that happens, another change driven by Google could have a cultural, and perhaps even political, impact...

[And on to a discussion of the way Google's "AdSense" program could make small blogs profitable, and could make Google rich. This was a few months before Google's IPO, which I was "sensible" enough to take no part in. That looks stupid now, but not as stupid as pooh-poohing posts like this.]