Leveling Down Legal Resources: Why Epistemic Arguments Fail

Abstract

The rich evade conviction more often in criminal trials than the poor. They also win more often in civil cases against the poor. Given that money buys better lawyers and better lawyers are instrumental in winning in adversarial trials, the rich have a structural advantage in laissez-faire trial systems. Such inequality is concerning. In a landmark article, Alan Wertheimer argues that we should level down legal resources in civil cases on the basis that doing so increases the adversarial system’s accuracy—that is, its chance of reaching correct decisions. In a more recent article, along similar lines, Shai Agmon also advocates that, given some constraints of adequacy, legal resources should be leveled down in both civil and criminal cases. This article aims to show that such arguments fail because leveling down legal resources could decrease a trial system’s accuracy, making it worse by Wertheimer’s or Agmon’s own criteria.

I. Introduction

The rich evade conviction more often in criminal trials than the poor. They also win more often in civil cases against the poor. The higher chances of the rich for winning court cases is a structural advantage in our legal system, given that money buys better lawyers and better lawyers are instrumental in winning in adversarial trials. This kind of inequality is concerning. In a landmark article, Alan Wertheimer argues that we should level down legal resources in civil cases to eliminate this structural advantage of the rich.¹ Wertheimer’s article has been the decisive word on the matter in the philosophical literature and remains uncontested since its publication in 1988. It was also arguably the most sophisticated treatment of the issue in the literature until it was superseded by Shai Agmon’s two recent papers, which concur with Wertheimer’s conclusions and extend them to criminal cases.^{2, 3}

Wertheimer’s and Agmon’s discussion presupposes a model according to which, in every legal controversy, one of the parties deserves to win and substantive justice requires the verdict to be in that party’s favor.⁴ For instance, Wertheimer writes that “there is an independent criterion of a just result-the party that should win is the party that deserves to win … Whatever terms we use to describe this independent criterion, it is coherent to say that a civil controversy has been settled on its merits, that there is a just or accurate resolution to a civil dispute, that one party or the other has a right to win the case or has a right to a settlement of (roughly) a certain size.”⁵ Agmon concurs, writing that, in legal trials, “roughly, a just result of the adjudicative process should be judged against the criterion of the accurate interpretation of the law, in accordance with the facts of the matter” with the implication that “those who are innocent should not be convicted.”⁶ The legal system’s accuracy—that is, its capacity to arrive at just verdicts—is then identified as its primary goal.^7, Thus, given the choice between two systems, they would have us opt for the one with higher accuracy. They then argue that we should equalize legal resources between the rich and poor by leveling them down, on the assumption that doing so will make trial systems have a “higher probability of just results” than existing laissez-faire systems.⁸

I shall argue that, given their assumptions, a laissez-faire system with unequal legal resources between rich and poor parties may have higher accuracy compared with a leveled-down alternative. Thus, leveling legal resources could decrease the system’s overall accuracy. This means that the inequality between the rich and the poor cannot be challenged by appeal to gains in accuracy, because such a gain is compatible not just with greater equality but also with greater inequality.

Before starting, let me clarify three preliminary points. First, in the existing discourse, underlying the idea of a court’s accuracy lies the assumption that decisions have a binary value (correct or incorrect), that decisions are made on single issues, and that the decisions are of roughly equal moral significance. All these are clearly simplifying assumptions that limit the application of this discourse to court reform in the real world.⁹ However, I will not pursue these practical points. My agenda, though perhaps of practical significance, is to show that Wertheimer and Agmon lose at their theoretical game.

Another assumption in the existing discourse is that by paying more, the rich can secure better legal resources. This assumption may be outright false, because the rich may pay much more for lawyers who promise the stars but deliver very little. Equally, the poor may avail themselves of public defenders or legal aid organizations whose services by far surpass many expensive lawyers in quality. Thus, despite paying more for their legal resources, the rich may not have better legal resources. But nothing of substance hangs on this potentially false assumption, given that we can simply drop any talk of rich versus poor and adopt, instead using Galanter’s distinction between “haves” and “have-nots” (this distinction simply tracks those who have on average better legal resources and those who do not).¹⁰ If having better legal resources amounts to better prospects, those who have them will have better prospects than those who do not. But I think the talk of rich vs. poor does some rhetorical work, especially in pumping the intuitions against me, and in favor of leveling down, so I shall continue using it.

Finally, and relatedly, the premise that having disproportionally better legal resources amounts to better prospects may also prove false, though it is empirically supported.¹¹ But this too should not worry us, because our question is not whether, in reality, the rich can win more often. Instead, our question is whether the system’s accuracy is necessarily reduced, if they indeed do (or what comes to the same, namely whether leveling down legal resources would necessarily increase the system’s accuracy if the rich indeed do win more often).

Given the differences between civil and criminal trial systems, I shall address them separately and proceed as follows: in Part II, I address civil systems, which are Wertheimer’s only subject and (due to their competitive nature) Agmon’s greater focus. In Part III, I turn to criminal systems. With respect to both civil and criminal trials, my aim will be to show that leveling down legal resources may not just increase the trial system’s accuracy but could also decrease it. This means that we cannot advocate for leveling legal resources down on the basis of promised gains in accuracy. In Part IV, I conclude by briefly discussing the implications of my analysis for egalitarian concerns about laissez-faire trial systems in the status quo.

II. Civil Systems

Considering civil trial systems first, we can divide cases into the following two mutually exclusive groups: those that the poor (hereafter “P”) should win and those that rich (hereafter “R”) should win. If all cases are decided correctly, P wins all the cases it should win and R wins all the cases it should win, with the following ratios obtaining:

Cases P wins/ Cases P should win = 1

Cases P wins/ Cases R should win = 0

Cases R wins/ Cases R should win = 1

Cases R wins/ Cases P should win = 0

In reality, however, each party only wins a fraction of the cases it should win due partly to the lack of sufficient legal resources. Thus, in reality some cases will be decided incorrectly, yielding the following actual ratios:

Cases P wins/ Cases P should win = <1

Cases P wins/ Cases R should win = >0

Cases R wins/ Cases R should win = <1

Cases R wins/ Cases P should win = >0

Civil cases are decided incorrectly when P wins cases that R should have won or vice versa. Thus, cases that are incorrectly decided can be so decided either to the detriment of P or R.

Let us then construct an example on the basis of these givens. Suppose that a civil trial system’s accuracy rate is eighty percent, that is, in eighty percent of the cases, the system reaches correct decisions. A fair assumption is that rich and poor parties each deserve to win fifty percent of all cases. Thus, absent any disparity, the rich and the poor have each a fifty percent chance of winning every case, though of the fifty percent of the cases they win, they each only deserve to win forty percent of the cases. The other ten percent of the cases that they win, they win due to errors that happen to be to the detriment of their opponents. Thus, absent any disparity, trial statistics for both the rich and the poor will be as follows:

Fraction of cases deserved to win: 50%

Fraction of cases won: 50%

Cases justly won: 40%

Cases won in error: 10%

Fraction of cases lost: 50%

Cases justly lost: 40%

Cases lost in error: 10%

Given that we have assumed absence of any disparity in legal resources, these trial statistics would match those of existing legal systems after legal resources are leveled down by policy interventions. In order to evaluate such policy interventions, we must compare these trial statistics with those of laissez-faire trial systems. In order to do that, we must now analyze how the inequality of legal resources complicates this picture.

The starting premise of our discourse is that people with better legal resources (who we assume are the rich) have a greater chance of winning, meaning they will have a greater share of total wins. Given the context of a civil trial, and the fact that each win for the rich is a loss for the poor, this means that the poor will have a smaller share of total wins. Accordingly, suppose that the inequality of legal resources perturbs the parties’ fifty–fifty chances into sixty–forty chances in favor of the rich. The new statistics for the rich will be as follows:

Fraction of cases deserved to win: 50%

Fraction of cases won: 60%

Fraction of cases lost: 40%

As compared with the new statistics for the poor:

Fraction of cases deserved to win: 50%

Fraction of cases won: 40%

Fraction of cases lost: 60%

The main point of this article is that this sort of disparity, assuming it does arise, would be compatible with several possible distributions of correct and incorrect decisions.

Consider that our starting premise only says that R earns a greater share of wins (which corresponds to P earning a smaller share). This greater share of R’s wins could come from either the 40 percent of cases that R justly loses (P justly wins) or from the 10 percent of cases that R unjustly loses (that P unjustly wins). In other words, the premise by itself is ambivalent as to which of the following two ratios would rise:

Cases R wins/ Cases P should win

Cases R wins/ Cases R should win

If the first fraction were to rise, the additional cases that R wins would be compensated by a decrease in the cases that R justly loses (which P justly wins). Such a change would decrease the overall probability that cases are decided correctly. If this happens in our example, then we should further specify our new statistics, which will be as follows for the rich:

Fraction of cases deserved to win: 50%

Fraction of cases won: 60%

Cases justly won: 40%

Cases won in error: 20%

Fraction of cases lost: 40%

Cases justly lost: 30%

Cases lost in error: 10%

And for the poor:

Fraction of cases deserved to win: 50%

Fraction of cases won: 40%

Cases justly won: 30%

Cases won in error: 10%

Fraction of cases lost: 60%

Cases justly lost: 40%

Cases lost in error: 20%

Here, the overall probability that cases are decided correctly will be dropped from the initial 80% to 70% (call this possibility 1).

But if the second fraction were to rise, the additional cases that R wins would be compensated by a decrease in the cases that R unjustly loses (which P unjustly wins). Such a change will increase the overall probability that cases are decided correctly. If this happens in our example, then we should further specify our new statistics for the rich as follows:

Fraction of cases deserved to win: 50%

Fraction of cases won: 60%

Cases justly won: 50%

Cases won in error: 10%

Fraction of cases lost: 40%

Cases justly lost: 40%

Cases lost in error: 0%

And for the poor as follows:

Fraction of cases deserved to win: 50%

Fraction of cases won: 40%

Cases justly won: 40%

Cases won in error: 0%

Fraction of cases lost: 60%

Cases justly lost: 50%

Cases lost in error: 10%

Here, the overall probability that cases are decided justly will be raised from the initial 80% to 90% (call this possibility 2).

Finally, it is possible for both the first and the second fractions to rise. For instance, in our example, the 10% greater share of R’s wins may be compensated partly by a decrease in the 10% that R unjustly loses (that P unjustly wins) and partly by a decrease in the 40% that R justly loses (P justly wins). Such a change could either decrease or increase the overall probability that cases are decided correctly, depending on which fraction were to rise more. If this happens in our example, we will have to further specify the new statistics in one of the two ways.

First, suppose that of the 10% greater share of R’s wins, a smaller part—say, 1%—comes from the 10% that R unjustly loses (that P unjustly wins), while a greater part—say, 9%—comes from a decrease in the 40% that R justly loses (P justly wins). The further specification of our new statistics will be for the rich:

Fraction of cases deserved to win: 50%

Fraction of cases won: 60%

Cases justly won: 41%

Cases won in error: 19%

Fraction of cases lost: 40%

Cases justly lost: 31%

Cases lost in error: 9%

And for the poor:

Fraction of cases deserved to win: 50%

Fraction of cases won: 40%

Cases justly won: 31%

Cases won in error: 9%

Fraction of cases lost: 60%

Cases justly lost: 41%

Cases lost in error: 19%

Here, the overall probability that cases are decided justly will be dropped from the initial 80% to 72% (call this possibility 3).

Alternatively, suppose that, of the 10% greater share of R’s wins, a greater part—say, 9%—comes from the 10% that R unjustly loses (that P unjustly wins), while a smaller part—say, 1%—comes from a decrease in the 40% that R justly loses (P justly wins). The further specification of our new statistics will be for the rich:

Fraction of cases deserved to win: 50%

Fraction of cases won: 60%

Cases justly won: 49%

Cases won in error: 11%

Fraction of cases lost: 40%

Cases justly lost: 39%

Cases lost in error: 1%

And for the poor:

Fraction of cases deserved to win: 50%

Fraction of cases won: 40%

Cases justly won: 39%

Cases won in error: 1%

Fraction of cases lost: 60%

Cases justly lost: 49%

Cases lost in error: 11%

Here, the overall probability that cases are decided justly will be raised from the initial 80% to 88% (call this possibility 4).

As we saw, the literature, in line with lay intuition, advocates leveling down legal resources in civil trial systems under the assumption that doing so will necessarily increase the trial system’s accuracy. For instance, Wertheimer argues that we should level resources down because doing so “will result in a higher probability of just results than the present laissez-faire system.”¹² It is, of course, entirely reasonable to want to increase the probability of just results in our trial system. But if this is what we wanted to do, we should only level down legal resources if possibilities 1 and 3 were the only possibilities. Given possibilities 2 and 4, leveling down legal resources could dent the laissez-faire system’s accuracy from 90% or 88% down to 80%.

Wertheimer’s analysis does not even show awareness of these other possibilities. It just assumes that increasing a party’s chances of winning must mean increasing their chances of winning more of the cases they don’t deserve to win. Agmon shares this assumption. His analysis suggests that what at least partly motivates this assumption is the competitive nature of adversarial civil trials.¹³ But the competitive nature of the trials alone does not guarantee possibilities 1 and 3. Of course, given the fact that in adversarial civil trials the parties compete, the resources of one side can be used to lower the chances of the other side, not just to improve each party’s chances separately. Concretely, by spending more than you, I can not only increase my chances, but also decrease yours. But all that follows from this is that I will have more and you will have less chances to win. And as we saw just now, this is compatible with me winning more of the cases that I deserve to win, or more of the cases that you deserve to win, or more of both. More wins therefore doesn’t mean more (or even any) unjust wins, not even in a competitive context.

In a competition, you lose by default, even if you deserve to win. This holds true in both civil and criminal cases. You can prevent losing by default if you put up a good defense. Even then you may lose, but the chances that you lose unjustly will be significantly diminished. And the better the defense you put up, the lower your chances of losing unjustly become. And it is precisely this that can increase your chances of winning. But it does not increase your chances of winning unjustly—unless, of course, your opponent (be it another private party or the people) fails to put up a good enough case. This is all that follows from the competitive nature of adversarial trials (both civil and criminal). The rich may be able to better prevent losing by default in both civil and criminal trials. And by doing so, they may be increasing the probability of just outcomes in both kinds of trials.

The poor, on the other hand, may not manage to do the same as often. They may be susceptible to losing by default in both civil and criminal trials. Hence, they may suffer from unjust outcomes that could have been prevented with sufficient legal resources. But these unjust outcomes don’t arise from the greater legal resources of the rich, even in civil disputes between the rich and the poor. In such cases, what is responsible for the poor unjustly losing to rich adversaries will not be the rich’s additional resources. It will be the poor’s lack of sufficient resources. The poor may just as often unjustly lose to the poor. The only difference may be that they also just as often unjustly win against the poor. Meanwhile, they don’t nearly as often unjustly win against the rich. And that’s where the disparity of chances between the rich and poor could come from in a competitive context.

Possibilities 2 and 4 undermine epistemic arguments in favor of leveling down because, in these cases, leveling down decreases accuracy. This means that the inequality between the rich and the poor cannot be challenged by appeal to gains in accuracy because such a gain is, at least in theory, compatible not just with greater equality but also with greater inequality.

What about in practice? Could the proponents of leveling down concede that gains in accuracy are conceptually compatible with greater inequality, but argue that in practice possibilities 2 and 4 will not obtain? This line of argument is open, but evaluating it would require empirical evidence about what is more likely for the rich to be able to achieve, given their disproportionally greater legal resources. Is it more likely for them to be able to win more of the cases they don’t deserve to win—that is, in essence, trick the court into reaching unjust outcomes to their favor? Or is it likely for them to be able to win more of the cases they deserve to win—that is, prevent the court from erring to their detriment? Or both, and if so, in what proportion?

Intuitively, it seems more plausible for the rich to be able to prevent the court from erring to their detriment more so than tricking it into winning unjustly. Obviously, the rich may be able to fabricate false evidence or otherwise defraud the court. But it would be odd to associate such abilities with their additional legal expenditure, given such interventions are—well, illegal. In corrupt legal systems, the rich may even be able to bribe the court. And they may even do so with the help of their lawyers. But we would not describe this as the rich expending disproportionate legal resources.

To summarize, accuracy can be increased either by only fixing for errors or by fixing for more errors than are introduced. Thus, accuracy can increase by either increasing correct decisions or increasing a mix of correct and incorrect decisions when the increase in correct decisions is greater. And even practically speaking, these possibilities—especially the latter—intuitively seem more likely to obtain. This isn’t to say that our intuitions about how the distributions of error pan out in practice couldn’t be proved wrong. But the proponents of leveling down must challenge these intuitions with hard empirical facts.

III. Criminal Systems

In criminal trials, the analysis will be similar. Suppose again that our trial system’s accuracy rate is 80%. To simplify matters, let’s also assume that only 50% of all defendants are guilty. A further fair assumption is that this 50% ratio distributes evenly among the rich and poor. If so, rich and poor defendants deserve to win against the people 50% of the time. But given the 80% accuracy, they only win 40% of the time. The other 10% of the time will be false negatives—that is, wrongful acquittals. Meanwhile, each party deserves to lose 50% of the time. But once again, given the 80% accuracy, they only lose 40% of the time. The other 10% will be false positives – that is, wrongful convictions. Thus, absent any disparity, trial statistics for both the rich and the poor will be:

Fraction of cases deserved to win: 50%

Fraction of cases won: 50%

Of which are rightful acquittals: 40%

Of which are wrongful acquittals: 10%

Fraction of cases lost: 50%

Of which are rightful convictions: 40%

Of which are wrongful convictions: 10%

Given that we have assumed the absence of any disparity in legal resources, these trial statistics would match those of existing legal systems after legal resources are leveled down by policy interventions. In order to evaluate such policy interventions, we must once again compare these trial statistics with laissez-faire trial systems. In order to do that, we must turn to how the inequality of legal resources complicates this picture.

Recall that our starting premise only says that in criminal trials R has a greater chance than P of evading conviction. Because R and P don’t face each other in trial but face the people, their chances of winning can vary independently from one another. Accordingly, suppose that the inequality of legal resources perturbs the fifty–fifty chances of the rich into a sixty–forty chances in their favor, while leaving the fifty–fifty chances of the poor intact. The new statistics for the rich will be as follows:

Fraction of cases deserved to win: 50%

Fraction of cases won: 60%

Fraction of cases lost: 40%

The greater share of R’s wins against the people could come from either the 40% of cases that R justly loses (rightful convictions) or from the 10% of cases that R unjustly loses (wrongful convictions), or a mix of both. If R’s additional wins come from reversing rightful convictions, we should further specify the new trial statistics for the rich as follows:

Fraction of cases deserved to win: 50%

Fraction of cases won: 60%

Rate of rightful acquittals: 40%

Rate of wrongful acquittals: 20%

Fraction of cases lost: 40%

Rate of rightful convictions: 30%

Rate of wrongful convictions: 10%

Here, the overall probability that People v. R. cases are decided correctly will drop from the initial 80% to 70% (the system’s overall accuracy will drop to 75%) (call this possibility 1).

If R’s additional wins come from reversing wrongful convictions, we should further specify new trial statistics for the rich as follows:

Fraction of cases deserved to win: 50%

Fraction of cases won: 60%

Rate of rightful acquittals: 50%

Rate of wrongful acquittals: 10%

Fraction of cases lost: 40%

Rate of rightful convictions: 40%

Rate of wrongful convictions: 0%

Here, the overall probability that People v. R. cases are decided correctly will increase from the initial 80% to 90% (the system’s overall accuracy will jump to 85%) (call this possibility 2).

Now for mixed cases. If of the 10% greater share of R’s wins, a smaller part—say, 1%—comes from reversing wrongful convictions, we should further specify the new trial statistics for the rich as follows:

Fraction of cases deserved to win: 50%

Fraction of cases won: 60%

Rate of rightful acquittals: 41%

Rate of wrongful acquittals:19%

Fraction of cases lost: 40%

Rate of rightful convictions: 31%

Rate of wrongful convictions: 9%

Here, the overall probability that People v. R. cases are decided justly will drop from the initial 80% to 72% (the system’s overall accuracy will drop to 76%) (call this possibility 3).

Finally, if of the 10% greater share of R’s wins, a smaller part—say, 1%—comes from reversing rightful convictions, we should further specify the new trial statistics for the rich as follows:

Fraction of cases deserved to win: 50%

Fraction of cases won: 60%

Rate of rightful acquittals: 49%

Rate of wrongful acquittals:11%

Fraction of cases lost: 40%

Rate of rightful convictions: 39%

Rate of wrongful convictions: 1%

Here, the overall probability that People v. R. cases are decided justly will jump from the initial 80% to 88% (the system’s overall accuracy will jump to 84%) (call this possibility 4).

To summarize, the disparity in legal resources results in higher chances of acquittal for R, which could come from reversing rightful convictions, wrongful convictions, or a mix of both. If R’s higher chances of acquittal comes from evading rightful convictions alone, it will decrease the system’s accuracy (in line with possibility 1). If it comes from evading wrongful convictions alone, it will increase the system’s accuracy (in line with possibility 2). If it comes from evading a mix of wrongful and rightful convictions, it will decrease accuracy when more rightful convictions than wrongful convictions are evaded (in line with possibility 3) and will increase accuracy otherwise (in line with possibility 4).

In criminal trial systems too, therefore, leveling down legal resources is compatible not just with increasing their accuracy but also decreasing it. Specifically, possibilities 2 and 4 undermine epistemic arguments in favor of leveling down, because in their cases, leveling down would decrease accuracy. This means that the inequality between the rich and the poor cannot be challenged by appeal to gains in accuracy, because—at least in theory—such a gain is compatible not just with greater equality but also with greater inequality.

What about in practice? Could the proponents of leveling down concede that gains in accuracy are conceptually compatible with greater inequality, but argue that in practice, possibilities 2 and 4 won’t obtain? This line of argument would once again require empirical evidence about what is more likely for the rich to be able to achieve with their disproportionally greater legal resources. Is it more likely for them to be able to reverse rightful convictions—that is, tricking the court into reaching unjust outcomes to their favor? Or is it likely for them to be able to reverse wrongful convictions—that is, preventing the court from erring to their detriment? Or both, and if so, in what proportion?

Intuitively, it seems that legal resources would be more effective in preventing wrongful convictions than in misleading the court. This makes me think that, with more legal resources, although the rich can evade some rightful conviction, most of the time they evade wrongful conviction. Although this introduces some new error into the system, it fixes proportionately more errors and thereby increases the overall accuracy of the trial system. Once again, this is not to say that these intuitions could not be proved wrong. But it is hard to see how proponents of leveling down could dismiss possibilities 2 and 4 without hard empirical facts.

IV. Conclusion

The chief worry of the discourse to which Wertheimer and Agmon contribute is about the system’s inequality rather than its inaccuracy. Wertheimer and Agmon seem to share this worry—that is, they seem vexed by the disparity in chances of winning between the rich and the poor, which resonates with lay egalitarian intuitions. Their commitment to boosting accuracy thus seems only instrumental to me and in the service of their egalitarian agenda. So, I think they only appeal to accuracy because they think they can condemn the disparity in the chances of winning between the rich and the poor in laissez-faire systems on grounds of lesser accuracy. I hope to have shown that challenging said disparity by appeal to accuracy could backfire. For it is possible that thanks to permitting inequality in legal resources, laissez-faire systems enjoy higher rates of accuracy compared with alternative systems where legal resources are leveled down. This means that leveling down legal resources could make trial systems worse with respect to accuracy. This is not to say that laissez-faire systems are good, nor even that they are overall better than systems where legal resources are leveled down. To the contrary, laissez-faire systems may prove wholly unjustifiable, or overall worse than leveled-down counterparts, notwithstanding their potential for greater accuracy. The point is only that, contrary to what Wertheimer and Agmon purport, laissez-faire systems are not unjustifiable on grounds of greater inaccuracy compared with systems where legal resources are leveled down. In conclusion, Wertheimer’s and Agmon’s egalitarian agenda may well be worth pursuing, but it is an agenda that must be pursued without appeal to the value of accuracy.