Silicon Valley is full of imperial visionaries whose mission is to take down Google, which now controls about 64 percent of the search market worldwide. They range from Microsoft to newcomers like Cuil, but all of them have ended up merely battling one another on the outskirts of Google's dominion, fighting for the shrinking number of searches that Google hasn't yet figured out how to conquer.
Then there is physicist Stephen Wolfram. When he wrote a blog post in early March announcing the imminent release of a new, highly sophisticated search engine, technology watchers from the Bay Area to Bangalore wondered if this was going to be The One. Wolfram claimed a breakthrough, an engine that does not merely crawl over Web sites seeking to find one that has already posted an answer to the question at hand. Instead, Wolfram|Alpha, as the technology is awkwardly named, has at its disposal 10 trillion (and counting) points of data from fields like chemistry, meteorology, history and astronomy. It also houses a vast number of equations and algorithms to connect the numbers, giving it the ability to compute completely original responses. "How old was Britney Spears on Sept. 11, 2001?" might be a question that has never been asked before, but Alpha knows the answer (she was 19 years, 9 months and 9 days old). Curious how unhealthy your grandmother's original chocolate-chip cookie recipe is? Input the ingredients, and Alpha calculates the calories.
Wolfram|Alpha is so new its impact is hard to predict, but some people believe it could transform search. Doug Lenat, founder of Cycorp, an Austin, Texas–based company working on artificial intelligence, says that by promising to answer the kinds of questions you would ask only of "a colleague or an assistant or another intelligent human being," Wolfram|Alpha represents the next step on the way to "something very much like the HAL 9000 computer from 2001: A Space Odyssey, but minus the homicidal-maniac tendencies." Compared to that, says Lenat, searching on Google is "like asking your dog to fetch a newspaper."
Wolfram says his creation is not so much a search engine as a "computational knowledge engine." It has a single input field, like a search engine, but users can pose complex questions. What is the date of the next total solar eclipse visible from Paris? (Answer: Sept. 23, 2090.) What is the current orbital location of the International Space Station? "Computing where the ISS is right now is not a trivial computation," says Wolfram. "You have to actually solve some differential equations for the motion of the aircraft in the Earth's gravitational field." And yet the result is returned as quickly as a Google search.
"You can think of it as a giant Excel spreadsheet, with all the laws of physics, economics equations, and formulas from various fields of science, mathematics and more," says Nova Spivack, the CEO and founder of Web startup Twine.com, who got a sneak peek of Alpha in March.
At this early stage, Wolfram|Alpha's knowledge reflects Wolfram's scientific background. He earned a Ph.D. in theoretical physics at Caltech by age 20 and became the youngest winner of the MacArthur "genius" grant in 1981, just after his 22nd birthday. But for questions not rooted in science or numbers, Wolfram|Alpha throws up its hands. In two weeks of testing, the most common answer I received was, "Wolfram|Alpha isn't sure what to do with your input." It paints the world with a broad brush, and has little data at the local level—for instance, it can't tell you when the Maine Lobster Festival occurs or provide a synopsis for a book or a film.
Wolfram insists the system is learning rapidly, but he concedes that it will always be limited to "systematically computable knowledge, and only for a certain set of things has our civilization gotten to the point where the knowledge is systematic." For that reason, when Alpha opens to the public this week, the service is sure to disappoint many who come to it expecting a sort of supercharged Google, an expectation Wolfram has fed by calling his creation "a new paradigm for using computers and the Web."
Wolfram's involvement in and of itself has generated plenty of hype, because of his reputation for disrupting the status quo. The London-born physicist's first hit came in 1988 with Mathematica, a program that did for complex calculations what Microsoft Word did for writing: provided a powerful but simple-to-use tool that quickly became an industry standard. With a few lines of Mathematica code, Boeing engineers could model the flow of air over an experimental airplane wing. Mathematica now has 2 million users and has made Wolfram fantastically wealthy. (Wolfram Research, the parent company behind both Wolfram|Alpha and Mathematica, is still private and brings in an estimated $22 million a year.)
In some ways, Wolfram|Alpha is the next logical step. With Alpha, Wolfram is bringing the computing power and data of Mathematica to a more general audience. "We're bringing expert-level knowledge to everybody," he says of Alpha.
The money he made with Mathematica gave Wolfram the freedom to tackle his next major project: reinventing science. It was about as humble an undertaking as it sounds. He obsessed over the idea that extremely complex and seemingly random events, including weather patterns and the way a zebra's stripes form, actually stem from a few, simple rules. The result of his 10-year odyssey was A New Kind of Science, a 1,200-page brick of a book that Wolfram self-published. The reception was generally lukewarm. The Nobel Prize–-winning physicist Steven Weinberg wrote that "the strongest reaction I have seen by scientists to this new book has been outrage at Wolfram's exaggeration of the importance of his own contributions."
It follows, then, that Wolfram is not humble about his creations, and Wolfram|Alpha is no exception. Given the technical nature of most of Alpha's data, it would be reasonable to expect that most of its users would be professional scientists, engineers and the like. But Wolfram dismissed that idea. When asked who he envisioned using it, he seemed almost frustrated by the question. "People," he said. "Out there. All kinds of people." Then, purposefully or not, he compared it to Google: "Just like people have gotten used to the idea that there's the Web out there, and we can go and search for things on it, people will get used to the idea that there's Wolfram|Alpha out there and we can go compute things on it."
Google won't say what it thinks of Alpha, but there are signs that it's watching closely. In the past, Google has made a few halfhearted attempts to give users hard-and-fast answers, as Alpha does. Most recently, Google integrated public government data into its results, so now when someone searches for "unemployment rate" they receive a custom-made graph and the latest figures. But Google chose an interesting time to announce the feature: the same hour that Wolfram gave the first public demonstration of Alpha, at Harvard Law School in late April. "My guess is that Google or Microsoft may do deals with Alpha," says Spivack. "With a service like Alpha behind it, [Google] could do much more sophisticated calculations." Google, of course, has 20,000 employees and $18 billion in cash—it could simply decide to build its own Alpha. As Spivack puts it, "Alpha may not be a Google-killer, but Google could be an Alpha-killer."
That might not give quite enough credit to Wolfram, who is not only a hyper-intelli-gent scientist but an extraordinarily successful businessman. When asked how he plans to make money off Alpha, he rattles off a half dozen plausible options, from corporate sponsorship to consulting. "We've been building a core technology," says Wolfram, "I think there'll be lots of interesting partnering."
That would make Wolfram even more wealthy than he is now, but he says that's not what drives him. "My main reason for doing it is, that, well, I was curious whether it was possible," he said. "But also, I think it's useful, and I like doing things that are useful." Doing something that is merely "useful" may not sound visionary, but then again, branding has never been one of Wolfram's strong points.