Could Google Go Obsolete?

The programmers and executives at San Francisco–based startup Powerset are hoping that the answer to this question will be “yes” once they debut their search engine that operates on natural-language technology.

 

Powerset finally offered its first public preview on September 17, reports Associated Press writer Michael Liedtke, at a conference hosted by TechCrunch, “a blog widely read by venture capitalists and other high-tech luminaries” (“Search Startup Ready to Challenge Google,” Sep. 17, 2007).

 

Powerset’s natural-language search engine would give Internet users a platform to conduct their searches in plain English—in other words, in the kinds of sentences people speak in real life—as opposed to by typing in an assortment of keywords.

 

Right now, all the major search engines—Google, Yahoo and Microsoft—use the traditional keyword-based model, with Google cornering more than half of all search engine traffic.

 

The difference between the two types of searching technologies has to do with how the search engine processes the search terms you type in when you want to find something on the Internet.

 

A keyword-based engine, like Google, basically disregards “minor” words like articles (a, an, the) and prepositions (in, by, about, etc.), and hones in on what it considers to be the most important terms. Natural-language engines take into account a string of words as a whole rather than looking at them as individual pieces.

 

For example, if you wanted to find information on student loans, in a natural-language search, you would type your request in the same way you would ask a person: “How much money can I get as an undergrad?” In a Google search, on the other hand, you would search by keywords like “undergrad money college” and hope to get results with information on maximum loan amounts. Change the order of your keywords—“money college undergrad”—and get different search results.

 

This distinction, writes Liedtke, means “Web surfers will theoretically be able to get more meaningful results by typing more precise search requests in the form of straightforward questions.”

 

 

Searches Going Au Naturel

 

Perhaps more buzzworthy than the natural-language engine itself is the fact that Powerset is licensing its technology from veteran powerhouse Palo Alto Research Center, the subsidiary of Xerox famed for introducing one revolutionary technology after another—laser printing, the graphical user interface (GUI), Ethernet networking, and the first commercial use of the computer mouse, just to name a few.

 

Ronald Kaplan, PARC’s top natural-language specialist, is now Powerset’s chief technology and science officer, Liedtke reports.

 

“We have the best natural-language search technology that has ever been developed,” says Barney Pell, Powerset’s co-founder and chief executive.

 

And he plans to take that technology, along with Powerset’s $12.5 million in venture capital, and give Mother Google a run for her money.  

 

Although Powerset is not the first to push natural-language searching, with Ask Jeeves failing miserably in the 1990s and Hakia currently struggling to make inroads in the Internet search market, Pell believes his company’s technology will be able to fill in the holes that keyword searches leave behind.

 

He compares the problem of searching with keywords to trying to talk to a toddler with limited language skills.

 

“In one sense, you are happy you can talk to it at all, but you still really want it to grow up so you can hold a real conversation,” he says. 

 

Not everyone expresses the same optimism. Industry analyst Charlene Li of Forrester Research says the problem with natural-language searches lies not with faults in the technology’s algorithms, but with a computer’s inability to differentiate between variable meanings of the same word and discern the subtle nuances of language.

 

In a search, Li says, asking “What caused the collapse of Enron?” and “What caused the downfall of Enron?” will generally yield completely different search results even though the questions are essentially the same. 

 

For other skeptics, whether Powerset can achieve its goal of revolutionizing the search engine sector will depend on the company’s ability to transform the mindset and ritual practices of Internet searchers. These critics argue that the majority of current searches are short, only a few words long, and that Internet users, increasingly demanding that all their online interactions happen in high-speed, may not easily make, or even want to make, the shift to more typing and longer search strings.

 

Danny Sullivan, a blogger for SearchEngineWatch.com, says Pell fails to provide a convincing argument for why today’s keyword search users would miraculously transition to the other side.

 

Internet users “aren’t using keywordese now because they somehow have been trained to do it,” Sullivan writes in a blog post. “No one from Google sat the searchers down and said ‘only two words, and don’t use conjunctions.’ People search however they want—and right now, they use only a few words” (“Hello Natural Language Search, My Old Over-Hyped Search Friend,” Oct. 5, 2006).

 

 

Testing to Determine Powerset’s Potential

 

The natural-language debate aside, how successful the technology and the company will be remains to be seen. Powerset is rolling out its search engine gradually and will begin by analyzing feedback from the 16,000 people who have registered as users of its test site, Powerlabs. The temporary site is currently operating in a limited capacity, explains Liedtke, only indexing content from Wikipedia, the Web-based, user-generated encyclopedia.

 

Powerset has set up Powerlabs to display, alongside its own results, the results returned by competitor sites like Google and Yahoo when given the same search questions. Powerlabs then requires users to choose which search engine’s results worked better for them before they can proceed with their next search.

 

With all the press over the past two years surrounding Powerset’s natural-language model, Google execs may be starting to feel the pressure of a competitor threatening to break Google’s steel grip on Internet searchdom. According to Liedtke, Google has been working on its own version of natural-language search technology since the end of 2005, pumping nearly $2.2 billion into research and development and beefing up its staff by thousands.

 

But executives at Powerset aren’t deterred by the competition or the criticism.  

 

“Google is the king. … Their system does an amazing job, given what they have to go on” says Pell. “But we think they have plateaued.”