Powerset Turns Out To Not Be All That Powerful
from the powering-down dept
We never understood the hype around Powerset. It was the latest in an extremely long line of startups that claimed to focus on “natural language search” — which is one of those holy grails for computer scientists who never stop to ask whether or not there’s actually any market demand for it. As Google has shown, people don’t need to use natural language to search. They’re just fine doing keyword search. Yet, for some unclear reason, Powerset was able to raise a ton of money at a ridiculous valuation, and did so using all sorts of buzzwords (and vague patent threats). But when it finally released a product (just to search Wikipedia) it proved to be rather ho hum. Searching Wikipedia via other means was still more effective.
Now comes the news (first leaked last week) that Microsoft has bought Powerset. While both sides will present this as a big win, the numbers being tossed about ($100 million) are not a big win at all for Powerset’s investors, and the exit certainly falls well short of the hype around Powerset. If Powerset was actually seeing any traction at all it never would have agreed to sell at that price. Basically, Powerset discovered what was widely known by industry watchers for years: natural language search is a neat challenge, but it’s not something the market is demanding. It will be interesting to see if Microsoft actually does anything with the technology, but my guess is that it will slowly fade away. If anything, Microsoft may do a little saber-rattling over the patents Powerset hyped up, but little else.
Filed Under: buyouts, hype, natural language search, search, startups
Companies: microsoft, powerset
Comments on “Powerset Turns Out To Not Be All That Powerful”
Let’s track the number of sales of startups for $100MM in the second half of 2008 – bet it is less than 5. Makes the sale of Powerset not so bad.
Re: Re:
Let’s track the number of sales of startups for $100MM in the second half of 2008 – bet it is less than 5. Makes the sale of Powerset not so bad.
Um, if you’re really a VC, you should know that the price itself is meaningless if not compared to the valuation and the amount invested. Based on Powerset’s initial valuation this is not a win.
Companies are still just salivating and annoyed at the Success of Google. Trying to ‘make the next Google’ now – but I’m fine and prefer keyword searches – they work well enough.
Sounds like they found some sucker investors though.
great analysis, Mike…
It will be back
Not Powerset, but natural language search. Eventually, computer users won’t even understand the concept of searching for content containing particular words. Why would you want to do that if you could search for content about a particular topic?* It’s a subtle but IMO critical distinction. I for one am NOT satisfied with keyword search. It works pretty well more than half the time, which means there’s lots of room for improvement.
* I’m exaggerating. Keyword search will still have a place, certainly at least in combination with language search – look for content about such-and-such that also contains word xyz, for example.
Re: It will be back
Natural language search is one of those things that sounds good on paper, but — as Mike points out — isn’t really needed. Even when it is beneficial for an individual, this utility drops over time.
For example, say that there was a search engine that would give you meaningful results from the following query “Who was the first president of the United States?” To someone who is unfamiliar with computers, this natural language approach would be inviting and easy-to-use. But it wouldn’t be too long before this same person would switch to using queries like “first us president”. Once you understand the principles of keyword searches, there’s far less return on the investment of typing out your query in a complete sentence.
Also, as the younger generation grows up with computers, there will be even less need for the initial natural language phase; they’ll “skip” right to the “first us president” phase. To them, keyword searching will be natural.
So, natural language search not only has limited utility now, but that utility will continue to decrease over time. Would you invest in technology that would only be used by newbies for a few years?
Good Analysis
It’s refreshing to see someone finally pointing out the clothing state of the Powerset emperor.