"The perfect," Voltaire wrote, "is the enemy of the good." This is frequently taken to mean that, since perfection is an unattainable ideal, its pursuit would be wasteful and, ultimately, futile. Instead, we should be satisfied with some suboptimal state of affairs. But what if perfection were attainable, and what if achieving it didn't require ruinous levels of expenditure—could it be possible that we would nevertheless still settle for, or even prefer, imperfection? It is exactly this argument that Professor Anup Malani advances in his paper Does Accuracy Improve the Information Value of Trials?, delivered during last Thursday's WIP talk. Written with Professor Scott Baker of Washington University in St. Louis Law School, the paper suggests that, from a social welfare perspective, we should tolerate somewhat inaccurate trials, even if it were costlessly feasible to improve courts to a level of perfect accuracy.
From the litigants' perspective, perfect trial accuracy would seem to be an inarguably good thing, as innocent parties will always be found innocent and guilty parties will always be found guilty. Furthermore, from a broader societal perspective, the threat of appearing before a perfectly accurate court seems like it would increase the ex ante incentive to comply with the law. Yet trials do more than vindicate private rights and provide a deterrent for wrongdoing—they also generate valuable information that parties not directly involved in litigation use to make important decisions. It is this function that, in Professor Malani's view, would be significantly undermined by hypothetically perfect courts.
The reasoning behind this argument is deceptively simple. Imagine a world in which there are "good" firms that only make safe products and "bad" firms that only make unsafe products, and further imagine a world in which courts can perfectly distinguish between the two. In such a world, it stands to reason that only good firms will ever go to trial. Along this dimension, then, perfectly accurate courts have increased the amount of information available to consumers about the identity of good firms. Counteracting this informational gain, however, is the fact that in such a world bad firms would never go to trial—instead, they will always settle rather than have their true nature conclusively revealed in court. At first blush, this would seem to divide the universe of firms into two distinct classes—those that go to trial and those that don't—and one might conclude that consumers could infer the quality of the firm based on its inclusion in one or the other class. But because settlements are almost universally accompanied by a non-disclosure agreement, and because they are far less likely to garner news coverage than guilty verdicts, consumers will be unequipped to differentiate bad firms that always settle from good firms that have never been sued—especially if the assumption that not all good firms can go to trial is true, which it clearly is in a world without advisory opinions and a limited number of judges and juries.
With imperfect courts, on the other hand, bad types will risk going to trial with some positive probability in the hopes of being falsely exonerated. Even given the looming chance of false exonerations, imperfect courts will still be generating useful information so long as they correctly convict bad firms more often than they wrongly convict good ones. In this way, consumers can only benefit from the information generated by trials so long as they are willing to tolerate a certain level of mistakes.
This paradoxical result leads to a number of normative implications, many of which were expounded upon during the Q&A session following Professor Malani's talk. First, courts should be more tolerant of inaccuracy the more important victim precaution is relative to precautions taken by the liable firm. Professor Malani points out that this is analogous to the effects of strict liability versus no liability on the level of victim precautions as discussed by, among others, Judge Posner, though with a key difference. Whereas potential victims will increase their level of care as the level of firm liability decreases, there may be a point at which lower trial accuracy so reduces the information available to consumers that they are no longer able to take rational precautions. Because of this, Professor Malani made clear that he does not advocate less accuracy, only that the optimal level of accuracy is somewhere short of perfection. Second, Professor Malani suggests that a rule either banning settlement or mandating the disclosure of the terms of settlement would improve social welfare, and would even be preferred by producers before they learned their type. Finally, the information value of trials, Professor Malani argues, offers a novel reason for supporting the law's historically greater tolerance for mistaken exonerations than mistaken convictions that does not heavily rely on contentious value judgments.
The hypothetical world of perfectly accurate courts raises some interesting questions about the world of imperfect courts that we actually inhabit. For instance, how would more accurate courts influence the incentive of plaintiffs to litigate? Would this change offset the increased incentive accurate courts give defendants to quietly settle? How can one tell, on the margin, whether a new procedural rule meant to improve accuracy actually increases or decreases the amount of information available regarding the identity of bad firms? And how do different liability and damage rules affect the return to accuracy—that is, does the change in probability of suit due to legal rules or damages measures, and the extent to which bad types get sued more than good types, change the relationship between accuracy and total information produced by litigation?
Professor Malani is likely to tackle these and other questions in subsequent articles. The power of the argument that inaccurate courts are preferable to perfectly accurate ones is that in can be extended to scenarios beyond just products liability litigation. Indeed, Professor Malani contends that it can be applied to any legal case where third parties rely on trial outcomes to make better-informed decisions, such as licensors of patented technology and potential employers of convicted felons. Furthermore, the model can be extended to scenarios where individuals learn from voluntary audits by third parties where the third party lacks the capacity to monitor everyone, such as SEC investigations of securities fraud, or FTC consumer product reviews. Thanks to the model's wide applicability, Professor Malani plans to further refine this powerful argument in future scholarship.