Update:

The much-rumoured Microsoft acquisition of “natural-language search” startup Powerset is now official, with a statement from MSFT and one from Powerset. Mike Arrington says that sources close to the deal tell him the rumoured $100-million asking price is in the ball park. Not bad for a company that has virtually no actual operating business.

Original post:

Matt Marshall over at VentureBeat says he has it on good authority that Microsoft is planning to make an offer for Powerset, the “semantic search” startup that has been in stealth mode for quite awhile now, popping up only long enough for a party or two, and recently poked its head out with a small-scale demo of its technology as a Wikipedia search engine. The rumoured dollar value for this deal? $100-million. If true, that would be a hell of a payday for something that hasn’t really shown much in the way of spectacular results so far, and is based at least in part on 30-year-old technology that the company licensed from Xerox’s PARC labs. TechCrunch says the deal could still be derailed by the Microsoft-Yahoo mess.

Of course, for a company like Microsoft, $100-million is chicken feed — Bill and Steve find that kind of money stuffed under the couch cushions when they vaccuum the Microsoft HQ. And the idea of an acquisition has been around before, with rumours floating here and there. It’s a painfully well-known fact that Microsoft’s search is a distant third place to Google and Yahoo, which is one of the main reasons the software behemoth continues its on-again, off-again (currently on-again) pursuit of Yahoo’s search business. If it could use Powerset to add natural-language search tools to its arsenal, that might help to close the gap with Google — although as Danny Sullivan has noted many times, we’ve been around this particular racetrack many times before.

Matt Marshall over at VentureBeat says he has it on good authority that Microsoft is planning to make an offer for Powerset, the “semantic search” startup that has been in stealth mode for quite awhile now, popping up only long enough for a party or two, and recently poked its head out with a small-scale demo of its technology as a Wikipedia search engine. The rumoured dollar value for this deal? $100-million. If true, that would be a hell of a payday for something that hasn’t really shown much in the way of spectacular results so far, and is based at least in part on 30-year-old technology that the company licensed from Xerox’s PARC labs. TechCrunch says the deal could still be derailed by the Microsoft-Yahoo mess.

Of course, for a company like Microsoft, $100-million is chicken feed — Bill and Steve find that kind of money stuffed under the couch cushions when they vacuum the Microsoft HQ. And the idea of an acquisition has been around before, with rumours floating here and there. It’s a painfully well-known fact that Microsoft’s search is a distant third place to Google and Yahoo, which is one of the main reasons the software behemoth continues its on-again, off-again (currently on-again) pursuit of Yahoo’s search business. If it could use Powerset to add natural-language search tools to its arsenal, that might help to close the gap with Google — although as Danny Sullivan has noted many times, we’ve been around this particular racetrack many times before.

Mike Arrington over at TechCrunch has taken another run at Powerset, which has been hitting the headlines in the blogosphere because the company continues to raise giant sums of money and get all sorts of attention with very little to show for it. Mike figures that the “natural language” search startup could be a “house of cards” and that CEO Barney Pell seems to be staffing the thing with Yahoo engineers, in the hope that the company will eventually be acquired by Yahoo in its eternal quest to beat Google.

That’s all fascinating and everything, but I have to say the most interesting part of his post is the video clip, which features platinum-haired Valley girl Sarah Myers of d7tv.com doing a feature called “PartyCrashers” (Matt Marshall has the video at Venture Beat too). She gets giggly with several staffers and has a brief interview with Barney Pell — in which he says he will probably raise more money this year — but the video really gets good near the end, when she interviews some poor schmuck who has either been hit on the head with a large object, is mesmerized by Sarah’s platinum bob and party dress, or has had way too much to drink (or possibly all three).

His explanation of what Powerset does is so incredibly obtuse that it is hard to believe. At one point, he says: “We have a demo where you can, like, search web pages and, like, get results.” It’s incredible. I don’t know who this guy is, but he is just about the worst advertisement for the company it is possible to imagine. One investor who says he put $100,000 into the company posted a comment on TechCrunch that said he was “very worried about this company.”

Admittedly, an offhand comment at a party by some faceless staffer who has consumed too many free beverages isn’t a fair judgment on an entire company, but still. It’s hilarious and painful at the same time. And a warning for startups: everyone who works for you is part of your sales team.

Human beings are funny. If you’re the underdog, the plucky startup going head-to-head with a big player, people will root for you regardless of whether you have a hope in hell of actually succeeding — but as soon as you become the dominant player, they will dump you in a heartbeat and start rooting for the next underdog. Google knows this better than anyone. Not that long ago they were the one getting all the cheers, and now we are all looking at companies like Powerset as the next potential “Google killer.”

For a company that doesn’t even have a real product yet, Powerset continues to get a surprising amount of publicity, including the story in the New York Times and a long feature at VentureBeat about the company’s licensing of natural-language search technology from Xerox’s legendary PARC research centre. Mike Arrington goes so far as to call Matt Marshall of VentureBeat a “cheerleader” for the company, which got $12.5-million in venture financing last year (contingent on the licensing of Xerox’s technology, according to Matt). He also points to a long treatise on natural language search that Danny Sullivan wrote here, which is worth reading.

Is Powerset the next big thing in search? Perhaps. But right now, it seems like a straw man dressed up in hopes and dreams, sent in to battle the now seemingly invincible Google. My friend Paul Kedrosky seems similarly skeptical of Powerset’s chances.

It’s interesting to see the spectrum of opinion on Powerset, the search startup that just raised a bunch of cash from some high-profile Silicon Valley types including Peter “PayPal” Thiel and Esther “Release 1.0″ Dyson (daughter of legendary astrophysicist Freeman Dyson).

Some critics, including Danny Sullivan of Searchenginewatch — who posted a comment on Matt Marshall’s piece at VentureBeat and has more thoughts here — seem to feel that Powerset is doomed, because so many other companies (including Ask) have tried to unseat Google by offering “natural language” search. Others have a “search is broken, so best of luck” attitude, and to his credit Barney Pell of Powerset links to representatives of both camps from his blog.

It’s obvious that in addition to getting some heavy hitters interested in his company, Mr. Pell is a pretty smart guy. He studied symbolic systems at Stanford and then got his PhD in computer science from Cambridge University, and wound up working on AI control systems for NASA’s Ames Research Center, according to this old bio page from there. Matt has some more history in his Venture Beat piece, but it’s clear Barney knows a thing or two.

In other words, he must know that plenty of people have tried the natural-language search model before, and yet he obviously feels that there is still something worth doing there, as he describes on his blog. And who wouldn’t agree that search is somewhat broken? Nine times out of ten, the first page of results from the average Google search is garbage.

If natural language search can help with that, I’m all for it. Of course, one of Mr. Pell’s critics, Steve Bryant, says having a good search engine isn’t enough to compete with Google.