Daniel Martin Katz
Daniel Martin Katz,
an assistant law professor at Michigan State University, has been studying how advances in computer technology and analytical software can aid lawyers in decision making. He spoke with California Lawyer
about the new field of quantitative legal prediction.
What is quantitative legal prediction?
Right now, if you go to a lawyer and ask, "Do I have a case?" human judgment will be the centerpiece of his answer. Your lawyer is basing his assessment on his own knowledge and experience. With the rise of legal-tech companies that use data analysis to guide decisions by lawyers or general counsel, we're seeing a shift where data is supplanting or supplementing some of that decision making.
How is it changing the way law is practiced?
One example is a RAND Corporation study that notes discovery can account for up to half the cost of a given piece of litigation. Historically, you needed to get someone to scan or read every single document. The first generation of e-discovery tools use a search term-based approach, where lawyers use search tools that comb electronic records and emails for key terms relevant to a particular legal matter. Though not yet uniformly accepted, a significantly superior technology leverages machine learning to algorithmically identify relevant documents. It may take a while, but courts will ultimately mandate these "predictive coding" approaches because they are both superior and cheaper.
Lawyers often like to say, "I can't be replaced by a machine." What should lawyers and law students be doing to ensure they can thrive in this new world?
The lawyer of the future will need to become data literate. The pathology of "I didn't go to law school to do math" is not going to cut it. Substantive legal knowledge will, of course, continue to be important, but lawyers need to combine it with understanding quantitative methods and analysis.
During the recession, companies and general counsels had to find ways to reduce expenses. Once [clients] learn they can pay less and get more, they'll never pay more again. You can do more with less by leveraging technology and
What will be the role of human reasoning in a world of quantitative legal prediction?
Blending the two streams of experience will typically provide better results than either alone. We still need humans on the front end and back end for quality control. For more complex questions, it will be like in [the baseball drama] Moneyball
, where information from human talent scouts is blended appropriately with a data-driven approach.
A number of companies are doing various pieces of the puzzle right now. TyMetrix has collected data on more than $40 billion in billings from many of the nation's largest legal clients. General counsels can use this data to conduct bill analytics and save money. Lex Machina takes data on every patent case filed since January 1, 2000, and predicts outcomes in patent disputes using an algorithm. If you use technology and analytics, you will do a better job and make fewer mistakes in the long run.