@UCSD: An Alumni Publication

An Alumni Publication   Archive vol1no3 Contact
 
Up Front: Letters to and from the editor
Campus Currents: UCSD Stories
Shelf Life: Books
Cliff Notes: Student life and sports
Class Notes: Alumni profiles
Campaign Update: Imagine the Future
Looking Back: Thoughts on UCSD
Credits: Staff and Contributors
Features

The Long Goodbye
Drawing on Tribal      History
Celebrating Our Sun      God
Ask Jeeves

Making Waves

Iran and Nukes
Cover Up
Tough Toucan
Alient Ant Hitchhikers
Pod People
Reef Relief

Archive

ADVERTISEMENT

ADVERTISEMENT

 

Features May 2004: Volume 1, Number 2
   

Ask Jeeves
By Kelli Anderson

 
     
 

It’s difficult to remember those dark days now, but not so long ago the Internet was a chaotic, alien world where only the bravest of ordinary citizens dared enter. Then in the mid-1990s a pioneering band of search engines appeared, offering to guide us through the wilderness. Ask Jeeves, the company founded by Garrett Gruener and David Warthen in 1996, was one of those early trailbreakers.

Garrett Gruener and David Warthen engaged the Internet long before most of us, and they knew it for the unruly, unrepentantly techie beast it was. “I could see it was going to eventually be the most important commercial structure out there,” says Gruener, who spent a lot of time in the early ’90s plying the pre-worldwide-web, bulletin board systems that provided regional and international bases for playing games and exchanging messages on the Internet. But, adds Warthen, “it wasn’t something we could see our mothers using.”

To navigate the precommercial Internet, one needed a lot of patience and a clear grasp of the Boolean logic that was behind the search protocols of the time. Gruener thought there was a better way, and he came up with an intriguing idea.

It wasn’t his first. Gruener, who had both a UCSD bachelor’s and master’s (UC-Berkeley, ’77) in political science with an emphasis on technology, had started two businesses by the early ’80s. His first had never taken off, but his second, a small communications software business in Berkeley called Virtual Microsystems, had attracted venture money and achieved modest success. He met Warthen in 1983 when the latter took a job at Virtual Microsystems, after two years working as a programmer at NCR.

“Garrett coached me a lot,” says Warthen. “I started off as a programmer, but soon I was running a project, then the software group and next the engineering department. It was an incredible learning experience.”

Though Warthen left the company after a few years to start a desktop software company and then a software outsourcing business, he and Gruener remained friends and frequently got together for lunch in Berkeley. It was at one of these sessions that Gruener first shared with Warthen his idea for making the Internet a more user-friendly place.
“I thought the Boolean search thing was way too complicated,” says Gruener, who had sold Virtual Microsystems in 1986 and spent the next few years consulting and looking for new ventures. “There needed to be a way to access the network with more natural language. And it should have personality.” He had a vision of a virtual concierge who would do one’s bidding by answering questions asked in everyday English. “I wasn’t sure most Americans would know what a concierge was, so I settled on a butler,” he says. He hadn’t read P.G. Wodehouse and wouldn’t until years later, but he came up with Jeeves— Bertie Wooster’s hyper-efficient, nearly omniscient valet in the Wodehouse novels.

In the summer of ’92, Gruener hired Gary Chevsky, a Berkeley computer science undergrad, to help write a program and develop prototypes for his question-and-answer search concept. “It wasn’t Ask Jeeves yet,” says Warthen, who supervised Chevsky’s work. “There was no character associated with it, and the technology prototype was a crude demo with basic string matching. But it showed the concept very clearly.”

Gruener pitched his idea to the venture firm that had helped finance Virtual Microsystems. The partners, however, were more interested in hiring him. Gruener took the job at what is now Alta Partners, but he was unwilling to let his butler quietly fade away. In mid-’95, during one of his lunches with Warthen, he expressed frustration about getting anyone to develop his idea. “I told him, ‘I think this is one of those ideas that if we don’t do it, ten years from now we’ll be having one of these lunches and we’ll say, ‘Remember when we had that really good idea?’” recalls Warthen. “Then I said, ‘Why don’t I take the ball and run with it?’”

Gruener agreed. Warthen reconnected with Chevsky, who was working at Informix at the time, and convinced him to put together a first version of the Jeeves software suite, in his spare time. “David was really a technology visionary,” says Chevsky. “He captured the moment and realized that the time was finally right for this concept.”

By mid-1996, the infrastructure was strong enough to start a company. “We had a working demonstrable prototype and it seemed like the next step was to buckle down and build the technology and the knowledge base to the point that it was really deployable,” says Warthen. Before incorporating in June of that year, Warthen lured Chevsky away from Informix and hired three content editors. Warthen installed them all in the squalid suite of offices on the second floor of the University Avenue building in Berkeley where he had been running his outsourcing company. “There were roaches, and there was a constant stink from the Chinese Longlife Vegie House downstairs,” recalls Chevsky. “The mesh-cage elevator was so ancient it required a human operator.” Gruener and Warthen paid for everything, including salaries, out of their own pockets. “David was very generous,” says Chevsky with a laugh. “He got me a folding bed just in case I decided to work all night.”

Everyone, especially Warthen, worked long hours building the service’s infrastructure. “Garrett refers to me as the technical founder,” Warthen says. “But in truth I wound up being the entire management team, half the engineering staff and the entire IT department for 15 months. My pager would go off when the server crashed. I was the first salesperson, too.”

Editors Needed

The content editors’ mission was to build a knowledge base of questions on pre-selected topics such as sports and celebrities. “We created these little capsules of knowledge that we called question templates,” says Warthen. “An editor would decide, ‘Here’s a question that a lot of people are going to ask. I’m going to go and figure out how best to answer it.’ Then she’d put that answer in a little capsule of knowledge that users could reuse over and over.”

Meanwhile Warthen built a meta-search engine that would aggregate search results from other traditional search engines in the event that someone asked a question that the content editors hadn’t anticipated. (A meta search is a query submitted to more than one search engine or directory). “The topics we focused on changed once we went live and got real-world feedback about what people were really interested in,” say Warthen. “That allowed us to focus more on the frequently asked questions.”

Ask Jeeves went live in April 1997, with a knowledge base of several thousand answerable questions. At the time Yahoo, Excite, Infoseek, Lycos and Alta Vista were operational but the competition went far beyond that: One directory at the time listed over 250 search engines. Ask Jeeves, however, “was something new,” says Chris Sherman, senior editor of searchenginewatch.com, a search industry news and analysis site. “Search in the early days wasn’t very good. There were basically two approaches. One was the search engine, where a computer would try to find everything possible and create a big index. The other was the directory, where human beings would look at web pages and write brief descriptions. Ask Jeeves really tried to be something in the middle.”

The question-and-answer format was not the only difference between Ask Jeeves and the other search engines on the web. Instead of search results, a user got back a set of confirmation questions that helped narrow his search. “If a user asked, ‘What’s the best place to eat in San Francisco?’ he’d get back a list of questions that basically told him that we know the answers to these questions, such as, ‘What are the highest rated restaurants in the Bay Area?’” says Chevsky. “He’d then get a pull-down menu of different types of restaurants in, say, the East Bay. The other thing that was different was that the answers or search results were editorially selected. People who would construct questions and answers in the area of food would really be experts in that area.”

Because of its novel presentation and a few lucky breaks—it was Yahoo’s “Pick of the Week” in its first week online—Ask Jeeves quickly gained a cult following. “Jeeves appealed to a segment of the population that was just learning to use the Internet,” says Dan Miller of the Roda Group, the venture firm that provided Ask Jeeves with angel financing in late 1997. “These people were scared of getting 250,000 results, which was what was happening at the time. People would hit you over the head if you tried to take Jeeves away from them. It was the same kind of loyalty you see with Macintosh computers. That’s one of the reasons we decided to invest, even though most people thought the search space was already over.”

PAGE2

RELATED LINKS

Where's Jeeves?
VIEW

Ask.com
VIEW

"In December 1999 the stock price reached $190.50 a share. Life was good...It would bottom out at 86 cents in June 2002."