« Bad presidents and fictional schools | Main | The obligation of law schools towards law students »

Tuesday, April 19, 2016

Can An Algorithm Pick a Jury?

Right now millions of dollars are being spent to map the American electorate for political purposes.  Both political parties have invested heavily in identifying, targeting, and analyzing individual voters by name, address, party affiliation, interests, causes, and even hobbies.  The data game of political get-out-the-vote strategies is as sophisticated as it is creepy (if you care about privacy).  If you live in a swing state, your demographics, income, past voting record, not to mention race, gender, and marriage status have been crunched by data companies in what is called micro-targeting.  If you are a consumer, private data brokers know your habits, preferences, and tastes.  The FTC recently released a Report on Data Brokers that is eye-opening in the extent of personal information available. 

This data is obviously valuable to politicians, but what if the data – or any of the information from big data companies – were used for the other political right – jury service.  What if courts could pick an algorithmically pure jury pool that actually represented a fair cross section of the jurisdiction? 

Such a prospect is technically possible, but is it a good thing?  Jury selection is notoriously unrepresentative with regular lawsuits challenging the jury venire as being not an accurate reflection of the community.  Jury summons yields have reach embarrassingly low levels.  For example, jury summons no show rates have reach 85% in some jurisdictions.  The result is unrepresentative jury venires, which do not match the actual demographics of an area.  And, for all those who complain about the ineffectiveness of Batson challenges, or the lack of fulsome voir dire in federal court (and some state courts), wouldn’t a big data inspired personal dossier save a lot of time and effort in selecting juries.  This is what fancy jury consultants do in big cases with lots of money, why not adopt it as a regular practice.  Trial lawyers would know a lot more than what they can observe (race, gender), and have a far better picture of the whole individual before them.  Race and gender as rough proxies would be replaced by more sophisticated data inspired insights of big data consumerism.    

So, why not let a big data algorithm pick a jury venire and provide targeted personal information about prospective jurors?  I attempt to answer these questions in a forthcoming article “The Big Data Jury.”  I would love to hear comments. 

Posted by Andrew Guthrie Ferguson on April 19, 2016 at 09:40 AM | Permalink


I'm all for smart tech solutions, but while we can do a lot better here, I don't see that tech—as open as some would see it as being to abuse—is the first solution.

Part of the no-show issue would be fixed by a no-hassle jury duty pay at twice the minimum wage…enough that low-income people, especially small proprietors, would not suffer from being away from work (nearly so much). No, it wouldn't cover the lost income of say, higher-income self-employed types, which I would rack up as being a progressive tax on living in a civilized society.

The pay would cover the first 8 hours of service for which you had less than 24 hours of cancellation notice. I.e., in California, where you might be informed the night before that you didn't have to come in (but you'd still had to hire a babysitter, make arrangements to cover your job or arrange transportation), or when you are excused from duty because 12 jurors were found by 10am, you'd still get the next day's pay for any inconvenience.

This is not terribly unlike when you hire a private contractor to work on a project. They need comp for scheduling in advance, driving to/from your office, a minimum amount of work when they show up.

Would this affect court costs? On first blush, astronomically. But in fact, what it would do would be to shift the *incidence* of those costs, shifting them away from the middle and lower-class public, and onto the parties who incur the costs. Lawyers who settle civil cases minutes before a trial is due to start, would have to consider that they pay a premium for abusing the court process. Think of the improved productivity and lower costs for employers, who'd have fewer days to pay employees, fewer substitute teacher to hire, fewer nursing shifts to re-jigger, etc.

There's no free lunch in this—the cost of ordinary criminal proceedings would rise, and taxes with it. But this can't be much of our taxes, and again, if it cuts the wasted time, and makes the jury process more fair by getting a more representative jury pool, it'd be a net positive for the country.

Posted by: Walt French | Apr 20, 2016 2:58:11 PM

If the goal is representativeness, why not also eliminate peremptory challenges altogether (and provide for a more robust regime of for-cause challenges)?

Posted by: Donald Caster | Apr 19, 2016 2:57:23 PM

(Sent the previous message too quickly.) This means that the algorithm used to pick juries has to be designed for oversight before one can consider the outcome

Posted by: Joel Reidenberg | Apr 19, 2016 1:31:46 PM

There is an underlying issue with the assumption that "courts could pick an algorithmically pure jury." Algorithms are value laden and the oversight of algorithms for governance integrity requires new approaches. http://ssrn.com/abstract=2765268

Posted by: Joel Reidenberg | Apr 19, 2016 1:26:50 PM

The comments to this entry are closed.