Teoma, the newcomer search engine that unabashedly targets industry leader Google, has officially launched its presence on the Web. Last August, Paul Gardi, Teoma's then president and CEO, claimed that the engine performed superior searches to other engines 99 percent of the time, "and we're working on the 1 percent." (See the August 20, 2001 NewsBreak at http://newsbreaks.infotoday.com/nbreader.asp?ArticleID=17526.) Gardi stated then that Teoma would leave beta status when the size of its database had grown sufficiently to rival its competitors.
In September 2001, Ask Jeeves Web Properties acquired Teoma for $1.67 million in cash and stock options. Gardi remains with the company as vice president of search. Teoma founder Apostolos Gerasoulis, a computer science professor at Rutgers University, remains as vice president of research and development. The company's chief scientist is Tao Yang, an expert in parallel computing at the University of California-Santa Barbara.
Ask Jeeves, which built its reputation with technology that allows users to search against a large, manually constructed database of frequently asked questions, uses a more conventional engine to handle those searches for which its database lacks close matches. In January, Ask Jeeves announced that it had integrated Teoma as its fallback search engine, claiming a 25-percent improvement in search results by doing so.
Teoma officially left beta status on April 2, with the site actually going live on April 1. The launch garnered considerable attention in the mainstream press. Echoing the brashness noted in last August's NewsBreak, the Los Angeles Times ran a half-page article on April 1 titled "On a Search Mission to Outdo Google." It quoted Gerasoulis as saying, "We are the next generation in search." The New York Times also took note of the upstart. In addition, CNN picked up an Associated Press piece that said, "Gerasoulis has a message for everyone who relies on Google as an online guide: It's time to move on."
Such bravado in taking on Google, which handles some 150 million end-user queries per day, might lead one to conclude that "Teoma" means "brash" or perhaps "hubris." Gardi told me last August that Teoma means "cunning." More recent Teoma press releases claim the word means "expert." A quick check of a Gaelic dictionary reveals meanings of "skillful," "expert," or "dexterous."
The essence of Teoma's claim to superiority is that a simple linear hit list cannot reveal to the user relevant pages whose importance vary by experts in different communities of interest. Even Google, with its "page rank" analysis of links (essentially a measurement of popularity) cannot, according to Teoma, tease out specialized—but highly relevant and highly trusted—content. Essentially, Teoma's claim is that a pure popularity contest can't deliver subject-specific expertise.
Indeed, Teoma has coined a new service mark, "Subject-Specific Popularity," to distinguish how its engine works. The way Teoma exposes its unique features to users, along with its nomenclature, has changed a bit since its beta days. The following results groupings now have simpler labels:
Some sample searches reveal how Teoma differs from engines with simple linear hit lists. Performing Gardi's suggested example search of "alcohol abuse" yields hits in the following groupings:
- "Web pages" is now labeled "Results," with a subheading of "Relevant web pages."
- "Web pages grouped by topic" is now "Refine," with a subheading of "Suggestions to narrow your search."
- "Experts' Links" is now "Resources," with a subheading of "Link collections from experts and enthusiasts."
How well can this methodology serve users? Lou Rosenfeld, a best-selling author and expert in information architecture, says that layered or segmented result sets can offer real benefits to the user. "There is some scientific data that clustered search results can be better than simple relevance-ranked lists. And we're beginning to see more sites employ clustered results, so perhaps there is increasing anecdotal evidence that it benefits searchers. But those sites (e.g., BBC) are providing access to much smaller subject domains and content collections. They can employ more human expertise to manually derive some of their clusters."
- "Results"—Offers 460,100 hits. The National Institute on Alcohol Abuse and Alcoholism provides a page on "Alcohol: Problems and Solutions" and "CSAT Treatment Improvement Exchange TIE: A substance abuse treatment resource."
- "Refine"—Suggests topic areas of "Alcohol Drug Abuse," "Substance Abuse," and "Alcohol Research"
- "Resources"—Offers a collection of sites, most of them in turn offer their own notion of the most important links on alcohol abuse
Rosenfeld wonders if machine-derived clustering can deliver the same benefit to general users, noting, for instance, that Northern Light's Custom Search Folders did not yield significant advantages as a general Web search engine. (The company recently withdrew from the general Web search marketplace just before its purchase by divine, Inc.)
Significantly, all of Teoma's hit list hyperlinks are not direct links to the corresponding page or site, but instead are links to a redirector at Teoma. For instance, a link labeled "National Institute on Alcohol Abuse and Alcoholism" under the URL http://www.niaaa.nih.gov is in fact a link to:
A user who clicks on a hit list result is redirected immediately to the labeled URL. In the meantime, no doubt, Teoma's logs take note of which hit list items are clicked on.
Teoma's advertising partner appears to be Overture.com (formerly known as Goto.com). The search for "alcohol abuse" yields sponsored links (clearly identified as such) for a treatment center in Malibu and for the Center for Law and Addiction.
Any reader who conducts a search on his or her own name or Web presence at Teoma and finds no expert links can take comfort in the fact that a search at Teoma for "Teoma" yields this notation: "We found no link collections for your search Teoma."
How likely is Teoma to topple its named target, Google? Rosenfeld is dubious. "I don't know that Teoma will bury Google. It's really, really hard to unseat so established a service. Google was able to unseat its competitors because it represented something radically different than the bloated portals that the AltaVistas and Excites had become. Teoma's approach, if it indeed is beneficial, is evolutionary, not revolutionary, so it certainly won't take the world by storm, and may never touch Google."
Greg R. Notess, editor of SearchEngineShowdown.com, is also skeptical of the prospects for a quick win by Teoma. He told me: "While it is great to see new search engines appearing with some interesting approaches, Teoma still has a lot of growing to do. The idea of using Web link patterns to identify communities on the Web may help for certain searches. But Teoma is still missing many features that make Google so important for the professional searcher. Teoma has no advanced searching capabilities, no cached copies of pages, and no indexing of PDF and other file types. The relatively small size of Teoma's database means that hundreds of millions of Web pages just cannot be found at Teoma."
What about Google itself? I asked Google spokesperson David Krane to comment on the company's brash new rival. He reacted with generosity and equanimity. "Google welcomes competition in the search marketplace because we think it's a good thing for consumers. Google has been widely recognized as the leading Web search engine because Google consistently delivers high-quality results that are relevant to what our users are looking for. We are hopeful that Teoma helps raise awareness of the importance of search engines in people's everyday lives."
On April 1, the day reporters first peeked at the new Teoma, Google issued a press release demonstrating that no matter how seriously Teoma takes itself, Google's self-image is much less self-important. The release revealed that Google's Page-Rank technology is in fact powered by a large team of trained pigeons. (See http://www.google.com/technology/pigeonrank.html)