Juries and jars of jelly beans

My newly protracted commute recently led me to an iPhone app that has me thinking about juries and jelly beans.

The app is called "Waze." I log in at the start of my drive, tell it where I'm going, and start driving. The app tracks my progress and the progress of everyone else who's logged in and headed in the same direction. And it uses this aggregated information from the commuting crowd to suggest better, faster commuting routes while I'm driving. It's taken me awhile, but I've now accepted that no matter how many secret routes I think I know, the crowd always knows better.

Tapping into the wisdom of crowds is nothing new. In fact, it's quite old school. In 1906, elitist geek Sir Francis Galton happened upon a country fair, and was engrossed by a version of the how-many-jelly-beans-in-the-jar game, in which prize-seeking fair-goers guessed the weight of an ox. No one got it right, and Galton asked if he could take home the guesses to prove his snobbish theory about how far off the common folk who did the guessing were. In the end, it was Galton's ox that was gored - when he calculated the mean of all the guesses, he discovered that the crowd as a whole had been amazingly accurate. They'd guessed the correct weight within a pound. Long before the iPhone, the crowd knew better.

You can try it yourself - put a bunch of jelly beans in a jar and invite your co-workers to guess away, and you'll find that the crowd beats out the individual guessers, time and again.

It's called "emergence" - the science that explores the eerie ways in better decisions and solutions emerge from the activity of crowds. From ant colonies and Adam Smith's invisible hand to political institutions and Wikipedia, people are paying more attention to the ways in which groups of independent individuals make better decisions in the aggregate than even the experts. It's the subject of a 2004 book by James Surowiecki, "The Wisdom of Crowds." And as the title suggests, crowds are not just good at predicting things - under the right conditions, they often exercise better judgment than individuals do.

It struck me that the law has, perhaps intuitively, been at the forefront of practice emergence, in the form of the jury system. To ensure that justice is done and disputes are fairly resolved, we call upon the wisdom of the crowd, believing that the collective decisions of our peers will be better than the decisions of an individual.

Because emergence is embedded in our legal system, we might do well to pay attention to the lessons learned and insights attained through the study of emergence.

Mistakes are the engine of emergence.

One the key insights of emergence is that the mistakes of individual members help the group make better decisions. Consider an ant colony with untold numbers of ants striking out mindlessly in different directions in search of food. The one ant in 100 that finds food does so completely by mistake - she literally stumbles onto food. She then heads back to the colony, leaving a pheromone trail that other ants stumble onto by mistake and follow to the food source. Slowly, the mistakes of the other ants - the wandering that led to no food - are abandoned, and a purposeful, productive ant trail takes shape. In a sense, errors become the engine of the crowd.

My new iPhone app works in much the same way, enabling the crowd to learn from the errors of its members, and thereby make better commuting decisions. A process that allows for and harnesses the power of individual trial and error, then, is critical to fostering emergence. In emergence, errors aren't something to be avoided - errors lead to better decisions.

The bigger the crowd, the better.

A corollary to the error-as-engine aspect of emergence is that more individual decisions or guesses improve the aggregate decision. The more guesses at the jelly bean jar, the closer the average gets to accurate. The more cars out there testing different commute routes, the more likely it is that a better route will be found. Finding ways to increase the number of individual participants enhances the crowd's decisionmaking.

Diversity drives good crowd decisionmaking.

Another element that seems to improve a crowd's decisions is having different people thinking about questions in different ways. The problem with a group of specialists is that they all think alike. The value of the crowd lies in the different angles its members take on a problem, and the better results that arise out of the aggregation of those different angles. And counterintuitively, adding in people who don't know as much about an issue appears to improve crowd decisionmaking, because it brings new thinking to the table.

Safeguarding the independence of the crowd members is key.

Perhaps the most important element of emergence is the autonomy of each crowd member. A crowd that gives rise to good decisions is made of members who are less concerned about what others are doing or thinking, and more concerned with their own individual decisionmaking. Problems arise when members second-guess their own initial instincts in light of others' views, and allow themselves to be influenced by others. Following the crowd is not the point - it's the aggregated results of independent decisionmaking that lets crowds work their magic.

Aggregation is more important than discussion.

Discussion and communicative deliberation, it turns out, doesn't improve crowd decisionmaking. In fact, it can turn crowds into mobs, depriving the crowd of the independent decisionmaking and the resulting errors that drive the crowd to find better solutions, and generating conformity rather than creativity.

Our current jury system does some of this stuff well. It increases the likelihood of diversity by drawing people randomly from different walks of life. Jury instructions are designed to preserve and foster the independence of individual jurors.

But there are ways that our jury system fails to take full advantage of the wisdom of crowds. And while there are obvious limits to the application of emergence concepts to juries, emergence concepts raise questions that challenge traditional assumptions about juries. Are traditional juries of six or 12 too small to generate good decisions? Do jury deliberations as they are now practiced actually sap the wisdom of the crowd and lead to groupthink? Would juries reach better decisions if jury members didn't talk to each other at all? Is there a role for trial and error in deciding cases?

The sacred status of the jury system and its ubiquitousness and standardization make some of these questions rather scandalous. But is precisely the sheer number of courts and juries that affords opportunities for experimentation. In a handful of cases, can we make better use of technology and the Internet to grow the size of juries, and assess the results? Can we test different procedures for jury deliberations that emphasize juror independence instead over jury influence? And can we relieve juries of the burden of making a perfect collective decision by aggregating the flawed decisions of individuals?

In a prior article ("Taking the jury - and trial - to the Internet," Jan. 9), I suggested re-imagining and re-packaging the trial as a website that allowed jury members to independently take in evidence and testimony in an interactive, dynamic digital environment like the one they spend so much time in already: the Internet. The "intertubes" have proven quite a fertile environment for crowdsourcing, and taking trials to the intertubes may well be the best way to unleash the benefits of emergent decisionmaking in the trial setting.

I suppose what I'm suggesting is a process of crowdsourcing jury improvements - allowing many different individual courts to apply the science of emergence to jury trials in ways that could give rise to better jury outcomes. It's a far cry from oxen and jelly beans, but it's a means of avoiding complacency in our courts and taking advantage of trial and error in improving the quality of our trials.