Wolfram expecting big bang from new computational tool
CHAMPAIGN – Wolfram Research is getting ready to launch its most ambitious project since Mathematica.
Called WolframAlpha, the new "computational knowledge engine" is scheduled to be released in May. With it, folks will be able to get answers they can't get from search engines.
Beyond that, officials of the Champaign-based company aren't saying much. The first public word about WolframAlpha came in early March when company founder Stephen Wolfram mentioned it on his blog.
But even after Wolfram's discussion of it, WolframAlpha remains, in the words of Winston Churchill, "a riddle wrapped in a mystery inside an enigma."
Company co-founder Theo Gray said at this pre-release point, he can't say what WolframAlpha is, but he can definitely say what it is not. It is not a search engine.
"Everyone thinks it's a new search engine. ... What we're actually doing is something that no one has ever done before. It sounds like a search engine. That's not what it is," Gray said in an interview last week at the company's headquarters.
Search engines are good only for finding answers that have already been written down by somebody and that appear on a Web page somewhere. But if you're asking a question that requires both reasoning and computation, WolframAlpha may be able to answer it.
An example: A student needs to find the mass of a certain number of atoms of sodium. Several things need to be known to get the answer. One is Avogadro's number (the number of molecules in a mole – a unit of measure – of the substance). Another is the atomic weight of sodium. The computer needs to come up with two pieces of factual data, then apply the correct formula to produce the answer.
WolframAlpha can do that, Gray said.
For a more routine example, suppose a cook needs to find the number of calories in a particular recipe. WolframAlpha could figure that by coming up with the number of calories in each ingredient – for example, in 5 grams of apples, in 3 tablespoons of sugar – and adding them up.
During the past year, experts in various disciplines have been hired by Wolfram Research to come up with "curated data" important to those disciplines.
Essentially, they've put the data in forms a computer can "understand."
All that information helps equip WolframAlpha for becoming a "natural-language question-answering system," something researchers have been trying to do since the advent of computers.
"A lot have tried and it's incredibly difficult to actually make it work," Gray said. But with tremendously complicated algorithmic code and vast amounts of raw data, WolframAlpha can do it, he said.
The tool will have applications in finance, economics and virtually any field involving quantitative data, ranging from disease statistics to sports scores.
"Everyone uses knowledge and wants actionable information," Gray said.
Gray calls WolframAlpha "the most important thing we've worked on in the history of the company, outside of Mathematica."
"We've shown it to a limited number of people," he said. "Everyone we've shown it to is someone not easily impressed, and they've been impressed by it."
But they've also signed confidentiality agreements not to talk about it, he added.
Exactly how Wolfram Research plans to generate revenue from WolframAlpha has not been disclosed. The tool will be available free to the public on the www.wolframalpha.com Web site in May. Even now, people can go to the site and apply for limited exposure to WolframAlpha.
Beyond the free public Web site, it's possible Wolfram Research might offer premium subscriptions for a wider variety of services. And it's likely WolframAlpha will be accessible through Mathematica software the company sells.
There's also the possibility Wolfram Research will form partnerships with other companies. Partners could include search engine companies, but they could also include any companies in the business of delivering information – mobile-device firms, for example.
The bottom line for how successful WolframAlpha will be is how it resonates with users, Gray said.
"The more people find it useful, the better," Gray said. "We think it will initially appeal to more technical sorts of people."
"There will be additional opportunities to be a success or failure if we get partnership deals, but if people don't like the stuff, it's going to fail," Gray said. "Every indication is we're going to succeed."
However, some people may be disappointed with the results if they ask WolframAlpha the wrong kind of question – for instance, who is Britney Spears' current boyfriend.
"We haven't curated the celebrity boyfriends data set yet," Gray said dryly.
Following Wolfram's initial blog about WolframAlpha, a number of high-tech bloggers have portrayed WolframAlpha as either a competitor to Google or as a potential Google acquisition.
Gray dismisses the notion of competing with Google and shoots down the idea of selling the company.
"No one in their right mind goes against Google. You don't compete against Google in search. They do it extremely well," he said – adding that Google co-founder Sergey Brin was once "one of the more successful" summer interns at Wolfram Research.
Instead, WolframAlpha will be a complement to search engines and operate "quite differently from the way a search engine works," he said.
"We're not interested in being bought," Gray added. "The company is not for sale."
Wolfram Research employs about 300 in Champaign.
"If we're successful, more people will be added here – a lot more here, as well as other places," he said.
Some bloggers have compared WolframAlpha to the computer aboard the Starship Enterprise of "Star Trek" fame. But it's not that advanced, Gray said, nor is it like the computer in "2001: A Space Odyssey."
"It's not the HAL 9000," Gray said. "Right now, you still have to think."
WolframAlpha is another step in that direction, he said, but there will have to be more "big steps before we get to HAL 9000."
What Stephen Wolfram says about WolframAlpha:
(excerpted from his March 5 blog)
"Fifty years ago, when computers were young, people assumed ... that one would be able to ask a computer any factual question and have it compute the answer. But it didn't work out that way."
"We can only answer questions that have been literally asked before. We can look things up, but we can't figure anything new out."
"Some people have thought the way forward must be to somehow automatically understand the natural language that exists on the Web ... I realized there's another way: explicitly implement methods and models, as algorithms, and explicitly curate all data so that it is immediately computable."
"The way humans normally communicate is through natural language. ... Getting computers to deal with natural language has turned out to be naturally difficult. ... But if one has already made knowledge computable ... all one needs to be able to do is to take questions people ask in natural language and represent them in a precise form that fits into the computations one can do."
"Pulling all of this together to create a true computational knowledge engine is a very difficult task. ... But I'm happy to say we've almost reached the point where we feel we can expose the first part of it."
"It's going to be a Web site – www.wolframalpha.com – with one simple input field that gives access to a huge system, with trillions of pieces of curated data and millions of lines of algorithms."
"That almost gets us to what people thought computers would be able to do 50 years ago!"
What others are saying about WolframAlpha:
Nova Spivack, CEO of Radar Networks, on Twine.com: "Stephen Wolfram is building something new – and it is really impressive and significant. In fact it may be as important for the Web (and the world) as Google, but for a different purpose. It's not a 'Google killer' – it does something different. It's an 'answer engine' rather than a search engine."
Doug Lenat, founder of Cyc, on Semantic Universe: "Stephen Wolfram generously gave me a two-hour demo of WolframAlpha last evening, and I was quite positively impressed. As he said, it's not AI, and not aiming to be, so it shouldn't be measured by contrasting it with HAL or Cyc, but with Google or Yahoo. ... I would invest in this, literally and figuratively. If it is not gobbled up by one of the existing industry superpowers, his company may well grow to become one of them in a small number of years, with most of us setting our default browser to be Wolfram Alpha."
Saul Hansell in his New York Times blog: "Wolfram's search engine, called WolframAlpha, is meant to be able to answer specific factual questions in a far more precise way than any search engine before it. For example, it will parse questions like "What is the location of Timbuktu?" or "How many protons are in a hydrogen atom?" to answer the questions rather than simply pull up sites that have the answer on them.
"If it achieves its very ambitious goal, it could be quite useful and influential. ... But Mr. Spivack's post has a critical logical flaw, one that too many people make: Google is a company, while Wolf-ram Alpha is a technology. They are very different. And it is Google's success with users and advertisers that made it 'important.'"
Jon "Hannibal" Stokes, cited in ComputerWorld blogs: "Wolfram can fairly claim to have revolutionized the math software niche with the 1988 launch of Mathematica. ... I expect that the May launch of Wolfram's service will remind everyone who overestimates the power of computers that this process is not something that can be automated."