The Google Stalker

Schultz’s lifelong work “involves all the syntactical elements of language, all the linguistic elements of language, all the contextual values of language,” says Ward Johnson, a director with technology incubator ZH Computer, which owns ePrécis. “It really takes into account every meaning of a particular word.” Johnson concludes with a massive understatement: “It’s a non-trivial task.”

“I think Arnie’s on his fourth pass through the dictionary,” says Henry Neils, ZH’s cofounder and president. “It probably takes him six to eight months to go through the dictionary, a word at a time.”

Each time Schultz makes his pass through the English lexicon, he’s like a football team’s defensive unit being penalized half the distance to its own goal line: He gets a little bit closer to an endpoint that he’ll never reach, since new words are always being created and definitions of old ones shift. “Arnie views the world as a semantic swamp,” Neils says. “He needs to get to the bottom layer of the swamp, and every time he tries, he gets a little closer.”

And now ePrécis may be getting closer to the big time.

A Program’s Travels

Like any incubator, ZH Computer—which is owned by a handful of local investors along with Neils and Johnson—has several irons in the fire at any one time.  Among its current hot properties are DART, a consumer-level software program that helps users clean up recordings before putting them into a digital form; and Assessment.com, an online career skills–assessment program.

EPrécis may turn out to be ZH Computer’s biggest thing ever. But while the application appears on the brink of big things, that isn’t the first time that’s happened. The technology has spent much of its life passing through the hands of companies that could never quite figure out what to do with it.

Schultz’s long-ago request from the U of M Medical School to index abstracts led to a grant from the National Institutes of Health to develop a computer-text search-and-retrieval program. That program was notable for its pioneering use of so-called syntactic structures, which allowed it to pick out the words most likely to contribute to the meaning of sentences, and then form meaningful output from the results.

In the 1970s, Schultz sold the rights to that program, then called Masquerade, to Mara-thon Oil, which was diversifying in some unusual directions. Marathon relinquished the rights back to him after its contract with him lapsed. Schultz, meanwhile, went to work for West Law (now Eagan-based West Publishing), where he converted Masquerade into a program that could electronically create back-of-the-book indices for law texts. But Schultz says that West lost interest in the idea.

Leaving West in the 1970s, Schultz attracted some venture capital investors and, with several partners, formed Syntactic Analyzer, Inc. Working with Don Malcolm, a former computer designer at Control Data, Schultz refined the program’s dictionary, developing subroutines that strengthened the program’s ability to produce abstracts as well as indices. They named the new iteration Syntactica.

ZH Computer bought the rights to the technology in 1992, after converting the program from IBM’s primitive PL/1 programming language to the more useful C++, which is used extensively for large-scale applications, particularly those that work in tandem with the Internet. Soon afterward, it sold Syntactica for $1 million and future royalties to Maple Plain–based hard-drive circuit-connection maker Innovex, which formed a subsidiary called Iconovex specifically to develop products related to Syntactica. The company took home best-in-show prizes at Comdex, the big U.S. technology trade show, for its Syntactica-based products EchoSearch (in 1995) and Indexicon (in 1996), both of which performed tasks similar to what ePrécis does now. But like Marathon, Innovex couldn’t find a suitable market for the products, and the rights to the technology reverted to ZH. Neither product is currently on the market.

All this occurred just before the massive tech boom of the late ’90s. Up until then, search engines such as Lycos and InfoSeek conducted searches that were wide ranging but not very selective. (They also weren’t bringing in much revenue.) These engines relied more on matching the words in a search string to the words on a page, with no way to sort out which sites might be the most relevant. It wasn’t until Google started using algorithms to weight the relevance of its search results—and posting clickable, text-only ad links on its quick-loading pages—that search engines began to show big revenue potential.

Neils and Johnson decided that getting the technology to market would be easier if its industrious creator could be kept nearby, so they hired Schultz as a consultant while they tried to figure a way to sell it. (They gave it the name ePrécis—combining the word “précis,” meaning summary, with the techie prefix “e”—late last year.) They believe that search engines are now ready to make use of Schultz’s program.