PORTALS By LEE GOMES
What Are Web Surfers Seeking? Well, It's Just What You'd Think
One thing about us Internet users: We like our music, we like our pictures, we like our sex -- and we like them all free.
Last week, AOL released a trove of what it thought were anonymous Web-search data from 650,000 of its customers. While intending to help researchers, AOL instead set off a privacy controversy because some of the users could, in fact, be tracked down. But taking up AOL on its original intentions, I got hold of the data set -- 2.27 gigabytes' worth, loaded it into my shiny new SQL Server database software, and started my own research project into how people really use the Web.
One learns, for instance, that excepting prepositions and conjunctions, the most commonly used word in the 17.15 million separate searches was "free." If something isn't free, it better at least be "new," as that was the next-most common word.
Excluding proper nouns, the next most popular words were "lyrics," "county," "school," "city," "home," "state," "pictures," "music," "sale," "beach," "high," "map," "center" and "sex."
Ah, sex. The Web turns out to be every bit the domain of the unbounded id we always thought it was. According to a research paper about the data prepared by an AOL-led team, p*rn was the third most common activity of Web searchers, behind entertainment and shopping. My study showed that 14% of all users made some form of explicit sexual search. And sex was No. 44 on the list of Greatest Hits words; usually, it's around 2,500 for standard usage, such as in English-language novels.
Among the sex searchers, there were 50,549 inquires for nude pictures. Perhaps for the first time in Internet history, the person most requested wasn't Pamela Anderson. In fact, it wasn't even a woman, but Peter Wentz, boyish singer of pop group Fall Out Boy. Ms. Anderson was second, followed by Paris Hilton.
How good is the Web with queries not involving naked celebrities? Users seem to think it needs improving, because in 47% of all searches, they didn't click on any of the results presented to them. (Although that could also mean they got the answer they needed just from the information in the list of links they were given.)
The AOL researchers noted that 28% of all searches were refinements of earlier searches, as users reshaped their queries to make the results more in sync with what they were looking for.
For those searchers who did click on something, in 42% of the time, they clicked on the first link presented to them. That factoid explains why Web sites spend so much money boosting their search-engine rankings.
In looking through all of the queries, I found 413,638 that were questions, that is, beginning with one of the five W's. Some 35% of these questioners never clicked on an answer, though I am not sure what even the best search engine could have done with a question like, "How to find your eye color."
How good is the Web at actually answering questions? I took a random sample of 50 "question" searches, and then visited the Web page that questioners had clicked on.
By my reckoning, 60% of these sites provided reasonable answers to what the user wanted to know, which included, "How to propose to a man," and "How to install more memory in a computer."
Unanswered queries included, "How to potty train a new puppy" and "How strong are gymnasts." The first went to a company that sold dog food, the second to a Russian gymnast training school looking to raise money.
Returning to the list of most frequently used words -- I need to thank Greg Sadestky of Poly9, fine purveyor of "mash up" map software, for his help -- "Google" was the 17th most common. While some might scoff at the notion of "searching for Google," it's actually efficient Web surfing. You're relying on the friendly, type-fixing search box, rather than the unforgiving URL bar at the top of your browser.
In the same vein, don't do what 15% of people in the sample did, and include the full URL, including "http" and "www" and ".com" in the search box. It's a waste of time and keystrokes.
The data were collected from March through May. Here are some random searches, and the numbers for each: Britney Spears -- 3,938; God --
3,279; Madonna -- 1,881; Mother Teresa -- 165; Stephen Hawking -- 41; Kofi Annan -- 12. While clearly celebrities did well, writers seemed to fare badly. Indeed, there were no searches at all for such modern masters as Malcolm Lowry, Martin Amis and Lee Gomes.Data aside, reading these queries is like listening in on random phone calls; even if you don't know who is talking, the experience can be wrenching.
Consider the person who, over the course of a few minutes, searched for "What to do when your Christian husband turns away from God," "How to deal with mental abuse in a Christian marriage" and "Do I stay or go when a Christian husband is on drugs and alcohol."
One of the recommended sites gave thoughtful answers to important life questions from an evangelical Christian perspective. The other hawked Bible books.
Write to Lee Gomes at snipped-for-privacy@wsj.com
Copyright 2006 Wall Street Journal
NOTE: For more telecom/internet/networking/computer news from the daily media, check out our feature 'Telecom Digest Extra' each day at