Wolfram Alpha searching for its niche

One week after a shaky debut, Wolfram Alpha is a lot more stable but is still having trouble defining exactly why information seekers should give it a try.

Any new search service that attempts to launch in the Age of Google is in for inevitable comparisons to the search giant. (And any search hopeful should also strive to learn from the unhappy launch of Google challenger Cuil last summer.)

But Wolfram Alpha says it represents something very different, and that people should not treat its "computational engine" the same way they do Google's search box. CNET readers learned that the hard way.

We asked readers to give us their impressions of Wolfram Alpha following its debut last Friday, and the results were not good. Asked to judge how happy they were with the outcome of their searches, readers gave Wolfram Alpha an average score of 3.55, with 1 being "most satisfied" and 5 being least.

For the most part, readers were dissatisfied with Wolfram Alpha's ability to produce results for anything outside of a relatively narrow set of queries related to math, science, or statistics. Forty percent said they would not recommend Wolfram Alpha to friends, while 28 percent thought it was only appropriate for "serious data nerds." (Percentages based on 1,459 responses.)

"This might not be what Wolfram Alpha is intended for but I think people used to Google and Wikipedia will expect to get some kind of answer," wrote one reader. "I think you should always give them something, even if the query is a little out of scope."

Another reader wrote, "For the moment their data seems incomplete. They can't answer many questions on anatomy or prescription medication."

Wolfram Alpha does crawl the Web for information. But it is distinct from Google in that a human being on Wolfram Research's staff must vet any sources used to generate data, said Jean Buck, director of computable data initiatives. Of the 200 people or so at Wolfram Research working on Wolfram Alpha, just 25 or so are working full time on data curation, she said.

Given Wolfram Research's history as the developers of the Mathematica software, the company is laden with experts in sophisticated math and science topics, Buck said, and therefore it shouldn't be all that surprising that results for those types of queries produce far more useful answers.

The harder part is about to come, as Wolfram adds data related to socioeconomic trends and attempts to diplomatically handle issues such as how to categorize Taiwan. Just this week, Wolfram staffers had a meeting over what to do about the entry on Macedonia, which is a country recognized by the United Nations as emerging from the remnants of the former republic of Yugoslavia but also refers to a wider region that incorporates parts of Greece and other countries on the Balkan Peninsula.

All those decisions are made by humans, not computers, which means the number of people dedicated to reviewing data for Wolfram Alpha is going to have to dramatically increase as the company beefs up sections on sports, cars, and food, three priorities listed for the coming months.

This also means that Wolfram Alpha visitors should never expect to find results for queries such as "causes of the Civil War," Buck said. Right now, a query for "Civil War" on Wolfram Alpha returns a reference to a book called "The Civil War," which is essentially the text version of the interminable movie made by Ken Burns.

"Our emphasis is always going to be on the computational aspect of things," Buck said. Wolfram Alpha, at some point, will be able to verify the dates of the Civil War, generals who were involved, and the location of key battles, but won't provide information on the causes of the war or the thinking behind strategic decisions.

But even that will take time, meaning that Wolfram Alpha users interested in those types of queries are probably better off searching the Web the old-fashioned way. "People think it's replacing search, and it's not. It's a different thing, it's complementary to search," said John Ekizian, a Wolfram spokesman.

Some CNET readers grasped that after a few spins around the block with Wolfram Alpha. "I found nothing to like except the audacity of the attempt and the hope that it will improve," wrote one reader.

Another reader perhaps summed it up best: "It looks like it'll be good for something, but I'm not yet sure how it will serve me, an ordinary person. It's not readily apparent how to use it most effectively. It may just be a matter of a learning curve for a new way of thinking about inquiries."