February 13, 2007, 11:45 AM — What bloggers are saying about the latest in information technology
Google, Yahoo and MSN, the top dogs of the search business, actually use technology that's fairly simple. Their search engines don't really understand language as people do, they merely recognize individual words and do statistical analyses on them to generate results that are likely to have something to do with what the searcher actually wants.
Google, by far the largest and deep-pocketed of the three, got to where it is today not because of excellent technology, but because they found a good way to monetize it with contextual ads. And it's money, more than pure science and technology, that drives the growth of the Internet. My theory about it is that an Internet company with "pretty good" technology and a strong focus on monetization is more likely to beat out an Internet company with superior technology and less focus on monetization. As consumers, we tend to accept "pretty good" so long as it is convenient. That's why there are so many dollar stores. Why do people shop there? Because things are cheap, there's probably one in your neighborhood, and the things you buy there may not be the best, but they are good enough to get you by.
A startup called Powerset announced $12.5 million in first-round venture funding last week, and an aggressive plan to change the very nature of search technology. Of course, by Google standards, $12.5 million is pocket change, and Powerset has some very well-moneyed rivals to take on. But, on the other hand, successful tech companies have been launched with less, and I'll give them credit for taking on the 800-pound gorilla. And of course, their technology is from Xerox PARC, which is by no means dollar-store technology.
The big difference is Powerset's natural language technology, which understands queries made in natural language. If you type "what is the name of Paris Hilton's dog" into the Google search bar, Google's engine sees eight individual words, and attempts to generate results that are probably relevant based on a statistical analysis of that clump of words. (The answer is "Tinkerbell".) Powerset, on the other hand, recognizes a sentence, and each word's relationship with the other.
There's still not much evidence of the superiority of the technology out there though outside of a controlled demo, and TechCrunch's Michael Arrington wonders if it's just a "house of cards". Arrington's confidence may have been shaken by the fact that at Powerset's launch party, someone from D7TV was there doing a "Party Crashers" segment. For the uninitiated, D7TV is a mostly useless "content network for mobile devices and the internet", and "Party Crashers" is an insipid series of shorts that shows the host crashing Silicon Valley parties. Fortunately, when host Sarah Meyers cornered Powerset CEO Barney Pell, he pulled off a decent impromptu interview -- but unfortunately, Meyers also cornered an unknown employee who didn't seem to know much of anything and stammered over his words. Advice to Pell: Keep this guy in the back office!
The SearchEngineWatch blog offers up a good old-fashioned rant about why natural language search isn't the "killer technology" people make it out to be, and the blogger is probably correct. But Powerset doesn't have to be a "killer," it just has to be better than the other search technologies (which it probably is), just as easy to use (don't know yet), and-most importantly-they have to have their revenue model down pat. Not much has been said about that to date. Are they planning to take on the Google monster and compete in the realm of contextual click-through advertising? During the "boom," I met plenty of dotcommers (amazingly, with venture capital to burn) who had great technology but no clue as to how to make money from it. I think Pell knows better than those early dotcommers, but he's just not talking. The most logical course of action of course, would be to simply acknowledge Google as the alpha dog of the search business and submit to an acquisition, should one be offered. Although, I admit it would be great fun to see him take Google head-on.
