The Google Stalker

Getting Results

ZH recently found out how effective ePrécis could be by plugging it into a homegrown “Scrape Site,” www.eprecis.com. The beta version of the site made its debut last November.

“It gives you a more robust output than you get with Google,” Johnson says. “It differentiates between sites with a lot of good content and those with very little content that happen to match the search terms. If you use ePrécis to look for medical content, you’re going to get fabulous results, because it treats all the technical terms it finds as top priority. Legal, medical, petrochemical—same thing.”

Scraping a search engine like Google is the ideal way for ePrécis to strut its stuff. Only one problem: Google doesn’t like to be scraped. Shortly after the beta debut, Google’s trackers discovered that its site was getting hit especially hard at a certain Internet protocol address, and it soon shut down ePrécis’s access.

“Our intention in [scraping Google] wasn’t to make money, it was to show the technology,” Neils says. “We knew people were lined up 500 deep to get in Google’s door and push their product, so we did it this way instead. And a week later, we got a call from Google.”

That call led to meetings at the Mountain View, California, “Googleplex.” That, in turn, resulted in an informal agreement by which ePrécis will continue to piggyback on Google’s searches (at a cost to ZH) while the two companies try to work out a more in-depth licensing agreement. ZH also is in discussions with Microsoft, IBM, Yahoo, the Central Intelligence Agency, even West Publishing. ZH hopes that it will either be able to tailor access to the ePrécis technology for each client, or else find a buyer that would have exclusive use of it.

“We’re still working out a price structure,” Johnson says. “If someone in the indexing business wants to use it, maybe it’ll be two cents per page, or five cents. If a search engine wants to use it, maybe it’ll be millions for unlimited, exclusive access. It just depends on what we work out.”

Even if the ongoing dance between ZH and Google doesn’t lead to long-term romance, ZH is confident that another search engine, looking to get a leg up on the king of the search sites, would be happy to do business. One client already on board is San Francisco–based Zimbio, which runs a network of public Internet portals (interactive Web sites that serve as entry points to other sites). Zimbio CEO Tony Mamone says his company plans to use ePrécis to create short abstracts of Web site content and news articles.

The one stipulation that ZH is insisting on with all suitors is that any agreement means access only to ePrécis’s technology—not to its nuts and bolts. “We’re never going to take the ‘black box’ off our server,” says Neils. “They don’t need to know what’s in the secret sauce. When I went to Google, the first thing they asked was, ‘How does it work?’ I said, ‘It just works.’ Even if I wanted to, I couldn’t explain it to them.”

Schultz himself insists that the all but the most basic aspects of ePrécis are too deep and gnarly to be explained at all by anyone but him. Nobody can duplicate ePrécis, he says, “because there was an art to how it was designed and assembled. Just because you explain it to someone doesn’t mean they can do it. If we ever do have a transfer of technology, I’m probably going to have to go somewhere with a team of linguists and spend six months explaining it.”