OK, folks time to get real here. The press has gone absolutely nuts over Wolfram Alpha.
This is a HUGE mistake.
I'm excited by anybody who tries to tackle these problems. But they are very, very difficult problems.
I wish Wolfram all the success in the world, because we'll all benefit. However, the hype reminds me very much of something else important to me -- the Segway. The launch (or leak) of the news that Dean Kamen had a new product produced incredible over-the-top predictions. In my opinion, those wild speculations ended up hurting the product and the company.
- The Segway is repeatedly termed a "failure", by those measuring its success by the inflated expecatations of the initial launch. After all these years of steadily increasing sales, with more and more people using them -- from celebrities to my mother (not to mention myself), "failure" seems a bit harsh. But if you were a late invester buying in at an inflated price, you might feel otherwise.
- There has been huge backlash. There is an entire Facebook group dedicated to hate speach directed to Segways. A large number of people feel compelled to issue threats of physical harm to people, merely for their posession of a Segway.
- Any time a newspaper reports that a police department has purchased a Segway to patrol, depsite their proven cost-effectiveness, a segment of people jump on the outraged comments about wasting taxpayer money. Apparently, their main concern is to maximize how much exercise the cops get. Spending money on more police, in expensive cars, doesn't faze them.
OMG! Sell your Google Stock! Sell Short!
There's more, but that's enough.
Fortunately, journalism isn't completely dead, not everywhere. The New York Times has one of the best responses, by Saul Hansell:
Let's be clear -- it's exciting to me, as a technologist, that Wolfram is taking on this challenge. But there are major technical challenges ahead for him. And he's got to build a huge infrastructure without running out of money. He could succeed technically and fail there. He could fail to get people to use the product, for good reason, or for silly ones. He could get all the usage in the world -- and fail to make money.
And, as he is trying to do all this, Google is quietly, but steadily, working to make their engine handle more and more challenges. I emailed Peter Norvig at Google about it. Not surprisingly, he doesn't have a lot to say about Wolfram's effort:
This approach is an interesting one that many groups have worked on, but so far we don't have anything to go on to tell how well Wolfram's group has done. You can see some of the work Google has done in this area (and others) at Papers Written by Googlers.
Even if you're not interested in reading research papers, I suggest you click that link, just to see the size and scope of Google Research. Google is NOT going to be sitting still. I have no fears for the health of Google. The odds are much better that Google will kill Wolfram Alpha, than the other way around. Though if you look at how many weaker search players struggle on, despite the mass of the Google behemoth, I think any talk of "killing" is highly premature.
Unfortunately, no such Darwinian effect appears to be operating in the world of Journalism. Journalists who write sensationalist articles and headlines should lose the trust of their readership, and thus be unable to obtain work in the field. (I am not referring to the Nova Spivack article I cited yesterday, despite Mr. Hansell's criticisms of it. The article was imperfect, but not straight hype).
Still, even if failure is the outcome, having someone come into the sandbox and play can only add to the richness of our toolset and our thinking about how to approach these problems.
My purpose yesterday was to highlight the future potential -- our ever expanding ability to ask questions of our computers and expect things we could not do ourselves.
My purpose today is to highlight how unpredictable and chaotic that path will be. Does anyone remember AltaVista? Despite being the first big search player, an outgrowth of work at Digital Equipment Corporation, it was not, in the end, the big success. (And Digital has been twice eaten since). Meanwhile, Google is a mere 11 years old.
My former boss at Symbolics, Ilene Lang, later became the first CEO of AltaVista. I hope to bring you some thoughts from Ilene after she returns from a trip later this month.
I considered waiting for her response to my email, but I didn't want to miss this great segue: Ilene now is CEO and President of Catalyst.org, an organization dedicated since 1962 to expanding the opportunities for women in business.
In the meantime, in celebration of Women in History Month, in conjunction with my younger daughter, Erika, we will bring you our selection of women in the math, sciences, and computer fields.
I'd never heard of Wolfram's "alpha" before, but the sensationalistic headlines, in my humble opinion, show a total misunderstading of Google's business model.
Google is in the advertising business, not the search business. Search is one of many distribution channels they have for that advertising. It lets them offer targeted ads, because what people search for can be used to target ads to people who might want to buy a product or service.
They've also figured out that if they give away products that involve high information content (mail, word processors, spreadsheets, etc.), targeted ads can be delivered unobtrusively in the margins, deduced from the information a person is working with.
It makes sense for Google to develop better programs in the information-processing space than Microsoft and give them away for free, since that drives eyeballs to Google's ads. You'll notice Google isn't building the G-Box 360; there's no information content there to be analyzed and monetized.
The Google phone gets you more deeply involved with your Google platform on mobile devices. My guess is that it's a just-in-case move, anticipating the possibility that mobile devices will develop into a big chunk of the information processing market (and thus advertising eyeballs).
Alpha may be able to answer factual questions directly, but it's not necessarily even in the same space as Google. Factual questions aren't likely to be very good at generating enough context to do good ad targeting. If I ask, "what is the tensile strength of steel," you don't have much information to use to target ads. You don't know why I want that information.
When I Google, however, I am typing in words associated with the actual information I need. I type in broader phrases, loaded with context. If I'm searching for "steel for skyscraper construction," it's easier for Google to find a host of relevant ads based on the query words and on the content of the top pages matching the query.
It's the very fuzziness of Google's search that makes it a good business for monetizing with ads.
Posted by: Stever Robbins | March 13, 2009 at 05:52 AM
That's a great point.
Wolfram Alpha, based on Doug Lenat's comments, is definitely not in the same space as Google-the-search-engine -- that is, it's not search. But if we assume they anticipate being advertising-driven based on query content, they're going to have to approach it very differently than Google.
I do see one advantage they have which may be to their advantage. People would come to their site for more focused questions, generally of higher value. If someone is looking for tensile strength of steel at a particular cost point in large quantity, and come to the Wolfram Alpha site as a result, they're more likely to be involved in high-value transactions.
So they may be able to charge higher advertising rates to a more select market, even if their keyword targeting is less effective.
In the end, however good their technology, and the adoption of their technology, it will come down to monetization.
After all, why don't we see newspapers scrambling to be the best, most reliable news sources, and raking in the money? Instead, we see old established papers in markets with little competition on the verge of going out of business, because neither their paper nor electronic versions can sufficiently monetize their content.
Posted by: Bob Kerns | March 13, 2009 at 07:18 AM