Big data won't make juries better or fairer
(Bloomberg View) -- Trial by jury, among the most hallowed institutions of the U.S. justice system, can be biased and inequitable. But if you’re thinking that employing big data can make it better, think again.
Creating a jury is a fraught process on many levels. First, there’s the task of filling the jury pool with citizens: Courts tend to select people randomly from voter rolls, state IDs or drivers' licenses -- an approach that tends to under-represent minorities and the poor, who are less likely to show up in such lists and more likely to have outdated addresses or work in jobs where they aren't compensated for a missed day.
Then comes the part where lawyers craft the actual jury, using their power to eliminate members. The side with the most resources to figure out which jurors are most likely to be sympathetic -- research that can include mock trials, polling and expert consultants -- automatically has an advantage. Again, racial and other biases are common.
Now, in the age of big data, some legal experts are hoping that the reams of information available on individuals -- including data on demographics, medical history, pharmaceutical use, credit, political affiliations and media usage -- could help make juries more representative. It's a possibility that Andrew Ferguson, a law professor at the University of the District of Columbia, explores in a paper called "The Big Data Jury" -- though he notes that he's still undecided as to whether it's a good idea.
Consider jury pools. With more information, courts could summon the correct proportion of every race, class and other group, including hard-to-detect ones such as religious minorities. Ferguson, for example, suggests that using data from a legal research company such as LexisNexis could easily help solve the outdated-address issue.
Relying on a private data provider to build juries, though, could create new problems. If it were using proprietary data, for example, it could encroach on constitutional rights in ways that would be very difficult to detect. In an interview, Ferguson agreed. What if, he asked, the company's owners or executives had a political agenda? Or what if it was in charge of selecting a jury pool for a case against itself?
Other, purportedly advanced databases and techniques can yield wildly erroneous results. In this context, the 18 percent yield rate of the current flawed system doesn’t look so bad.
But what about actual jury selection? Ferguson suggests that more and easier access to information might help address the imbalance of resources, and also help lawyers look past their biases and choose the best candidates. For example, a person with a long, clean credit record -- no matter what race, gender or class -- might be more willing to be tough on a defendant who walked away from a debt.
Actually, more data requires more resources to analyze, so it's hard to see how big data would level the playing field. Demographic profiling would be easier: Big data offers myriad proxies (such as media preferences or shopping behavior) that could provide "race neutral" explanations for eliminating black jurors.
Perhaps most insidious is the idea that big data could help identify the most malleable jurors. Political campaigners have discovered that scoring people by susceptibility to persuasion, rather than by their stance on issues, can be very effective: It’s easier to talk them into things if they respond emotionally. It's a nefarious approach to politics, and it's even more disquieting to imagine such technology being deployed in the courtroom.
In the classic play Twelve Angry Men -- probably the most inspiring portrayal of the American right to trial by jury -- a single critical thinker manages to talk eleven other jurors out of conviction, explaining how the circumstantial evidence against the defendant is inconsistent. In a big data jury, I'm afraid he wouldn’t have made the cut.
(About the author: Cathy O'Neil is a mathematician who has worked as a professor, hedge-fund analyst and data scientist. She founded ORCAA, an algorithmic auditing company, and is the author of "Weapons of Math Destruction.")