Interaction

May 17, 2006

Legal technicalities

L.A. Cicero

Roland Vogl, left, and Nat Love are co-founders of the CodeX Project at the Law School.

An impressive group of people stared out from the cover photo of a recent issue of Stanford Lawyer. Together, they provide legal advice to the leaders of a host of top high-tech companies in Silicon Valley and beyond, companies such as Microsoft, Google, Cisco, eBay, Yahoo, Qualcomm, Autodesk and Oracle.

But Stanford's legal expertise does not just guide tech companies in the law; it also explores ways in which technology itself can help navigate the law.

The law school's Program in Law, Science and Technology (LST), launched by former Dean Kathleen Sullivan is an umbrella for several centers that address the ramifications of such issues as intellectual property on the Internet, bioscience and ethics, and the intersection of law and computers in consumer society.

Intellectual property is an obvious example of a field that has been utterly transformed by the Internet. On the one hand the technology unleashed massive copyright infringement; on the other, through mutual exchange, that same technology has the potential to vastly increase the possibilities for human creativity.

In a slightly different vein, computer scientists and legal scholars under the auspices of LST, have established a center called CodeX that aims to harness technology to help people make their way through thickets of codes and regulations.

"The legal field is one that in many ways operates the same way it did hundreds of years ago, " said Roland Vogl, executive director of LST and one of the forces behind the center. "CodeX, and the computational-legal methodology behind it, is designed to increase the efficiency of certain legal procedures such as copyright clearance." Basically, he said, software could translate code along the lines of what TurboTax does.

One of Vogl's partners at CodeX, Professor Michael Genesereth, who leads the Logic Group at the Computer Science

L. A. Cicero

Michael Genesereth leads the Logic Group in the Computer Science Department.

Department, is particularly enthusiastic about the possibilities in managed health. Genesereth, who said he once spent several days lodged in limbo between an HMO and a hospital, figured this was a problem that could be fixed.

"This is very much a computational law problem," he said. Although lawyers certainly stand to gain from such innovations, the CodeX people are especially concerned with helping people affected by the law. "It's legal empowerment through informational technology," Genesereth said.

The goal is a world in which consumers could point and click their way through the world of health insurance, reading the relevant rules, regulations and statutes as they go. What each party could and could not do would be clearly spelled out; the facts of the case would not be in question. Law, essentially, would be brought to the site of conflict, where code would be made manifest to physicians, pharmacists, patients and HMOs.

Genesereth said he was bothered by what he saw as widespread cynicism toward the law, the result of excessive complexity and unequal access. So, with money in hand, thanks to his own successful start-up during the tech boom, he turned his attention to issues of social value.

He enlisted the help of CommerceNet, a research outfit advocating the principled consumer use of the Internet. (http://www.commerce.net/)

The involvement of an outside firm is not surprising; there are a multitude of links between the legal and technology communities off-campus and their professional counterparts at the university. The route between the law school and the computer science department in fact was to some degree mediated by Joshua Walker, an intellectual property associate at the Menlo Park office of the law firm of Orrick, Herrington and Sutcliffe. He joined Genesereth's Logic Group, which then approached Vogl, which in turn led to the establishment of CodeX.

"We all decided we wanted to create a new center with folks from many disciplines, and work at the boundaries," Genesereth said. "We'll do tech, they'll do law, but together we'll be able to do things we couldn't do alone."

It was years ago that such a mutual need was first recognized by Professor Margaret Jane Radin, of the Law School. She had a feeling that the intersection of law and cyberspace was a field begging to be figured out, so in 1995 she asked her dean, Paul Brest, if she could teach a course on it. He agreed — and hers was the very first course of its kind at Stanford, Radin noted proudly.

Radin, whom Vogl described as a "pioneer," eventually established and directed the law school's E-Commerce Center, which, like CodeX, is part of the Program in Law, Science and Technology.

"I could see there was going to be a revolution in intellectual property and contracts," she said recently, speaking by phone from New York University, where she is a visiting professor. Then Genesereth made his way to the law school, possibly directed her way by a graduate student. Their conversation led to a cross-listed course, with Radin doing guest-lecture duty.

"I thought it was wonderful," she remembered. "He was thinking about these problems from the other side, but in similar ways. It was incredibly enlightening, a really good collaboration. People in Computer Science, though, didn't understand there could be a law professor interested in what they do. So we had to tell them I was a bit atypical!"

Radin's commitment to bridging the divide between law and the technical sciences prompted her to speak up strongly in favor of allowing engineering and law students to sit in the same classroom. As a matter of fact, next year she will hold a fellowship in Law and Public Affairs at Princeton, where she'll be teaching patent law to engineering students, something she said she had always wanted to do.

The pragmatic, useful intersection of law and technology being studied and championed by scholars today is the most recent development in a long history of collaboration at Stanford. In the 1970s, the university already was in the vanguard of artificial intelligence and computational logic, which essentially ask, can computers think or reason? And if so, how? From the start, legal and computer scholars wondered if the legal system could be streamlined with the aid of computers, if information retrieval could be improved and if the theories being developed in artificial intelligence could help in developing models of legal argumentation and reasoning. In fact, the nation's very first AI and Law course was taught here, in 1984. (The International Conference of AI and Law, which meets every two years, will hold its 2007 meeting at Stanford Law School.)

But from the start the new field had snags.

For one thing, the U.S. legal system has historically been based on case law, which is practically impossible to systematize as it develops. (That is why CodeX is concerned with regulations or rules rather than with cases.) Perhaps for that reason, the field is stronger in Europe, which has a tradition of civil law, not common law. Adding to the difficulty is the fact that law uses natural language, as opposed to a constructed language. Furthermore, statutes are not necessarily inert or neutral; judges and lawyers examine them from different perspectives to draw different conclusions, and legislators can be intentionally ambiguous when writing them.

"Codifying case law is almost an oxymoron, and representing statutes in a computer also has serious pitfalls" admitted Anne Gardner, who has both a Ph.D. (1984) in Computer Science and a JD from Stanford. Her revised Ph.D. dissertation (An Artificial Intelligence Approach to Legal Reasoning, under the direction of Terry Winograd) was published in 1987, one of the first books in the field.

Undaunted, she and her colleagues are committed to the notion that AI can help uncover the bases for how we argue and what we mean by rationality.

As an example, Gardner's first project was to work on a Supreme Court case with Stanford law professor (and associate dean) J. Keith Mann. In the course of analyzing, constructing and writing legal arguments, Gardner said, she was struck by "the huge distance between our simple computational models and the complexity of the real thing."

"It's important to be accurate in describing the facts, the law and the arguments, and there was much more room for error than I had realized. A tool I'd like to see would help keep this all straight - an argument-checker, we might call it - that would raise flags if you make unsupported assertions or contradict yourself or engage in wishful thinking about what a case holds."

That idea has something in common with one of the projects on the desk of Deborah McGuinness, a senior research scientist and co-director of the Knowledge Systems AI Lab, which is about to join up with Genesereth's Logic Group. The project, funded by the National Science Foundation's Cybertrust program, concerns the Privacy Act and the way information can and cannot be used.

Despite its ambiguity, "law is very amenable to AI" McGuinness said recently in her office at the Gates Computer Science Building. "We can code for ambiguity," she explained. An expression like "routine use," for example, can be partially encoded by representing some ways of determining examples. Not all cases will be covered, but most of them can, and human beings can fill in the gaps.

And law is a lot easier than medicine, she noted: "How do you encode a human hand?"

Lawyers who use a system such as those designed by McGuinness must be able to believe in it, she said. "My job is to leave an audit trail of sources and reasoning" to ensure that explanation and trust is present every step of the way.

"You have to tell people what you're doing and how you reasoned, and create tools so people can follow the crumbs and dig down when they need to."

For Radin, it makes more sense to reach for what she called the "low-hanging fruit," the more standardized legal obstacles that we expect can be overcome with computers.

Vogl thinks in similar terms. He has an intellectual property project under way, for example, thanks to initial funding from Media X, to construct a software system to facilitate copyright clearance. The project aims to use computational logic and semantic web technologies to help creators find and obtain permission to use content and build it into new works which then could be released and earn royalties. For example, a documentary filmmaker using the system would be able to go online to access and get permission to use clips from other films. Or, the author of a work would be able to charge others a sliding scale depending upon the use to which the work was put, for example classroom, nonprofit or commercial. Thus advanced e-commerce technology, adapted in novel ways, could help ordinary people.

The first year of the project was mostly devoted to legal research by law students into such issues as copyright licensing terms, automated transactions and contracts, and online trust and dispute resolution. The second phase, assuming the group gets funding, will bring in the Computer Science people, who will go to work on developing a prototype.

With all the back-and-forth between the Law School and Computer Science, a joint degree might be in order, and in fact Professor Jeff Strnad, who is spearheading the law school's drive to expand the number of dual and joint degrees, said an agreement with the Computer Science department is near the top of his wish list. The groundwork appears to have been well laid by Vogl's program.

One person who arrived on the scene too early to get a joint degree is Nat Love, a CodeX co-founder. When he began working on the project, he was a Ph.D. student in Computer Science. He will graduate in June, and in fall will start life anew as a law student.