BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

Stanford-Bred Startup Uses Moneyball Stats To Handicap Judges, Lawyers

Following
This article is more than 9 years old.

If you're being sued for patent infringement before U.S. District Judge Lucy Koh in the heart of California's Silicon Valley, there's something you ought to know. Koh is tough on defendants and only grants 18% of motions for summary judgment, less than half the national average. So your lawyers had better be on their game before they try to convince Koh to chuck out the case, and you'd better have a fallback strategy in case she doesn't.

These stats are just a sample of what a venture-funded startup with roots in a project funded by Cisco Systems and Apple is doing to bring mathematical analysis to the arcane world of litigation. Where Lexis and Westlaw tell attorneys what the law is, Lex Machina tells them what actually happens in the courtroom.

The system, focusing for now on the litigation-intense world of patent law, has compiled an exhaustive database of patent cases going back to 2000 so companies and lawyers can determine how many times a patent has been the subject of litigation and how the lawsuits were resolved. Lex Machina also handicaps law firms based on their win-loss records before specific judges with specific procedural maneuvers, so in-house attorneys can determine who to hire.

Lex Machina's founder, Josh Becker, calls it "Moneyball law," after the Michael Lewis book detailing how Oakland A's General Manager Billy Beane used statistics to win against better-funded teams.

Lex Machina starts by trolling through the Pacer database of federal lawsuits, pulling data and compiling it with the help of machine-learning systems. Subscribers, some of whom are paying more than $100,000 a year, then construct specialized searches to determine how specific judges or courts handle motions and which law firms have the best records on certain types of lawsuits or legal procedures.

Becker got the idea while pursuing a combined law and business degree at Stanford. It started as a public interest project funded by Cisco and Apple, who were beset with patent litigation and wanted to get a better sense of who was doing the suing and how different courts handled the cases.

Becker, 45, had already started a career-services company in the 1990s after graduating from Williams College and worked his way through graduate school in technology, where he was also associated with Stanford Angels and Entrepreneurs, a startup organization. He realized there was a market for legal intelligence, but that conventional machine-learning systems had a difficult time deciphering legal prose.

"The data is out there, and you could hire an army of people and actually hand-code those pages," he said. "We asked: Can you use technology to crawl thru those hundreds of thousands of pages and get the information you need? It turns out to be a really difficult problem."

This being Stanford, he went across the campus to the computer science department to see if anybody was interested. The department said no, but Becker was able to hire one of the top students and spent several years developing a machine learning system to interpret legal filings. He hired Owen Byrd, 52, a University of Chicag0-trained lawyer who'd been involved in the Obama campaign's data-crunching effort, to help build the system and sell it to law firms and corporations.

Lex Machina drew $8 million venture funding from Yahoo ! co-founder Jerry Yang, Boston's Cue Ball Capital and other firms and launched its platform last November. Becker claims more than 100 customers including Google, eBay and Microsoft, and a third of the 100 biggest U.S. law firms.

Lex Machina focuses on patent litigation right now, since it lends itself particularly well to statistical analysis. Lexis and Westlaw, the dominent legal-research platforms, are fantastic at documenting case law and pulling up citations helpful to constructing a brief. But they aren't so good at predicting what any given judge will do with that brief.

"We know more about these judges than they do about themselves," Byrd told me. "We know to the second decimal point how they behave."

Behaved, past tense, not necessarily how they will behave. But if your patent case gets assigned to Judge Sue L. Robinson in Delaware, you might want to try move it to Chicago. She's only dismissed three cases on summary judgment since 2000, according to Lex Machina, out of more than 1,000 cases heard.

"You still have to write the brief," Byrd said.  But then you can press a button and see the 20 cases Judge Robinson has transferred out of her courtroom, to determine which arguments seem to be the most effective.

The process isn't entirely computerized. Lex Machina has 30 employees including a data team that hand-codes whether a case was won or lost and the damages awarded. (Surprisingly enough, it often takes some legal judgment to determine who won a case -- other than the lawyers on both sides, of course.)

"It sounds fluffy, but I really do believe analytics holds the promise of improving all areas of the law," Byrd said.

Lex Machina plans to expand into copyright and trademark law next, then into the rest of federal civil litigation. It also gives the service away to federal judges.

"Every time I get in front of a judge, the first question is, `what have you got on me?'" Byrd says.