|
|
|
|
Competing with Google 20 June 05 “The search industry is ready for an architecture war.”
Charles Ferguson in Technology Review Jan 05
Google has had an excellent year. Its stock gained 220% in the past 12 months, the best in Business Week’s Info Tech 100. Google’s revenue grew by 103 percent in the same period.
Yet, as Charles Ferguson noted in the January 2005 issue of Technology Review, Google appears vulnerable to competition. After all, there is nothing preventing a user from moving to a new search site if she gets better results with it. Ferguson sees Microsoft as Google’s inevitable competitor, and forecasts a repeat of the Netscape / Internet Explorer browser wars of the late 1990s.
Michael Yang and Yeogirl Yun of Become.com
Michael Yang and Yeogirl Yun have a different perspective. These Korean-American entrepreneurs anticipate a world of increasingly specialized search sites which will take revenue from the generalized offerings of companies like Google.
They focus on internet shopping search sites, which account for a very large proportion of click-through advertising revenues. In February, 2005, Yang and Yun launched Become.com, a shopping comparison website that uses a new proprietary algorithm, called Affinity Index Ranking, or AIR. According to Yang and Yun, AIR delivers better search results than Google’s breakthrough PageRank approach.
This is the pair’s third foray into search and shopping. They developed MySimon.com, which they sold to CNET in 2000 for $700 million. Subsequent to that, Yun developed another search site, Wisenut, which he sold to Looksmart in 2002. Both of these search companies provided technical innovations which improved search results.
As with Yun and Yang’s prior startups, Become.com brings new technical innovation to this competition. In addition to the improved ranking algorithm, Become.com also has a human element – a small team of researchers locates authoritative websites, such as Consumer Reports, and gives them a higher ranking than they would receive if pages were ranked automatically.
Does this new search approach generate appreciably better results than Google? In the few experiments I tried, the answer was “not really.” In fact, the search results were very similar.
For example, I searched for information on Ugg Boots, an Australian sheepskin boot. This was one of Become.com’s suggested searches. Both Google and Become returned the same sites for their first two results. And with Google, I could move over to Froogle with one click to see pictures and prices of Ugg boots. Not so with Become.com.
In my second search, for tea, Become.com beat out Google only because Google listed the Texas Education Agency (known as TEA) first. After that, the listings were pretty much the same.
Competing on page ranking algorithms seems to be what Chan Kim would call a “red ocean” strategy. There is still plenty of blue ocean in the internet search business. As James Fallows pointed out in a June 2005 article in The New York Times, there are many areas of internet search where the keyword search algorithms yield unsatisfying results. Fallows gave the example of finding information that compared California’s per-capita spending on public education over time with that of other states. This kind of information is readily available from an expert, but very difficult to find with a keyword search.
Given the current business model in search, it’s unlikely that this fundamental problem will be solved by entrepreneurs, because there’s very little revenue attached to these kinds of searches. Trying to answer Fallows’ question about California via Google yielded more than 200,000 results, but there are none of the “sponsored links” that generate revenue for Google -- the right-hand side of the screen is noticeably blank.
So, for now, entrepreneurs like Yang and Yun will go where the money is. They will continue to refine and extend sophisticated technical algorithms in areas that are already pretty well-served by companies like Google.
In doing so, they leave untouched the “question answering” form of search. Rather than matching keywords, this question form scans, parses and "understands" vast quantities of information in order to respond to queries.
The government is currently funding some product development research in this “question” area. As Fallows reports, government intelligence organizations such as the CIA and the National Security Agency are sponsoring a project called Aquaint, which stands for “advanced question answering for intelligence.” These agencies want programs that will allow intelligence analysts to get answers to questions like: “did any potential terrorists just buy an airplane ticket?”
Perhaps, like other government-sponsored research, these innovations will one day find their way into more commercial search programs. For now, however, internet search development seems a bit lopsided – over-developed on keywords, under-developed on questions.
More information:
|
|
All content on this site licensed under a Creative Commons License.
|
