“There are no solutions,” Thomas Sowell says. “There are only trade offs.” When considering policy and legal problems, tradeoffs must often be understood in light of the challenge of filtering: correctly sorting people, actions, or events into categories. The filter can be too wide or too narrow, and both flaws have costs.

　　Prominent changes in American law and policy based on the 1964 Civil Rights Act have tended to ignore the costs of broadening certain filters, particularly with respect to targeted behavior such as discrimination and sexual harassment. Thinking about the challenge of filtering clarifies how unintended consequences can emerge when we allow ourselves to think a law or policy has fixed a social problem, rather than just shifting the terms of the tradeoff.

　　Filtering, Knowledge, and Incentives

　　Say you are determining a divorce regime. You want to make it hard enough to get a divorce that marriages aren’t dissolving willy-nilly in a way that undermines social stability and human flourishing, but you want to make it easy enough that partners can get out of abusive marriages. If you make the divorce rule too strict, you end up with many people stuck in unhappy marriages where both partners are miserable, perhaps inviting further abusive and dysfunctional behavior. On the other hand, if you make the laws too loose, you end up with a bunch of divorced people whom a stricter law might have incentivized to work things out. Or, you encourage people to opt out of marriage altogether.

　　Tort reform provides another example. Presumably, we want a tort law regime that winnows the wheat from the chaff, allowing those who have legitimate claims to bring suit but discouraging frivolous lawsuits. A regime that takes all comers, with a low cost to file suit and a low cost of losing, will err on the side of incentivizing frivolous lawsuits. A tighter regime will err the other way, discouraging frivolous lawsuits but also excluding some legitimate claims. The trick is to get the criteria exactly right, such that they include all the people or behaviors that should be included and no one who shouldn’t be. That’s the filtering problem. It’s hard!

　　Filtering problems plague welfare and disability support services. How can they help those who genuinely need support, because they are just going through a rough patch or are actually disabled, without enabling others to take advantage of the system? The same holds for the question of access to guns. Red flag laws are designed to keep guns from getting in the hands of dangerous people, but not prevent law-abiding citizens from getting them. But it’s hard to know who’s dangerous before a violent crime has actually been committed. Make the laws too strict and we’ll prevent some law-abiding citizens from getting guns; make them too permissive, and we’ll let dangerous folk get their hands on deadly weapons.

　　Now say you’re designing an anti-discrimination law. You don’t want employers discriminating on the basis of race or sex when they make hiring, firing, or layoff decisions. You want them to hire on the basis of merit, their organization’s legitimate needs, and mission fit. If you make the law against discrimination too lax, you allow or invite improper discrimination à la pre-1964 Civil Rights Act America. But here’s the catch: If you make it too strict, you encourage employees to hire some people who don’t deserve the job or keep others in positions they really shouldn’t occupy to avoid the appearance of violating the law. Too-strict laws could further shape negative opinions and attitudes about members of the group the law is intended to protect—the affirmative action or diversity hire. So a too-strict legal regime, or too-broad filters, has bad consequences too—and one of them might be making improper discrimination more attractive under some conditions.

　　Anti-Discrimination Law

　　Legal scholar Richard Thomas Ford lays out the knowledge problem of anti-discrimination law, the current structure of which he critiques, in terms of a filtering problem:

　　The central problem in employment discrimination is distinguishing the victim of discrimination from the person who simply suffered an adverse employment action.

　　Under a default regime of employment at will, an adverse employment action—even an unjustified one—is not, in and of itself, a legally cognizable injury. In a large number of cases, it’s hard to tell whether the challenged employment action was justified, unjustified for nondiscriminatory reasons (e.g., the product of a personal grudge, a lapse in judgment, caprice, or a mistake), or discriminatory.

　　Ford goes on to note there is bound to be error, most of it in situations where there is an unjustified adverse employment decision, and the question is whether it was discriminatory. Liberals, he suggests, tend to accept more Type I error—falsely treating discrimination claims as valid, and conservatives more Type II error—falsely dismissing them.

　　Court rulings and subsequent civil rights legislation, building on the anti-discrimination in employment provisions in Title VII of the CRA, have broadened the filters, so to speak, accepting more Type I errors. As John Donohue writes, in Griggs v. Duke Power (1971), the Supreme Court ruled that not only “disparate treatment,” but even “disparate impact,” may be legitimate grounds to sue employers for discrimination. Formally neutral practices like requiring a high school diploma or passing a general aptitude test could be considered discriminatory if they result in statistical disparity in employment decisions. The 1991 Civil Rights Act removed the need to identify specific discriminatory practices in lawsuits dealing with disparate outcomes. The idea that a disparity in outcome is ipso facto a result of social injustice is a core premise of what Sowell describes and critiques as the “civil rights vision” in Civil Rights: Rhetoric or Reality? and the “social justice vision” in Social Justice Fallacies.

　　In Rights Gone Wrong: How Law Corrupts the Struggle for Equality, Ford describes how civil rights law pertaining to employment has spread to become available to almost everyone as a tool to claim compensation and damages, with the strange result that civil rights claims have risen since the 1970s, but civil rights litigants often lose. (That is a broad brush on a complicated and debated area of law. For detailed empirical analysis and some completed explanatory analysis, see Theodore Eisenbergs workhereand Michael Selmis workhere). There is some evidence that claimants tend to be the worst employees, because, compared to more productive employees, they have an easier time showing they don’t have other employment options and have a larger incentive to invest in litigation. What’s more, anti-discrimination laws have not achieved the eradication of disparities related to employment, though this is ostensibly their purpose. Part of the reason may be that broadly trained anti-discrimination laws, especially those focused on protections against discriminatory termination, inadvertently incentivize discrimination against protected groups at the hiring stage. Some studies have demonstrated the validity of this concern.

　　Bryan O’Keefe and Richard Vedder even argue that the 1971 Griggs decision and subsequent legislation reaffirming its provisions contributed to the emergence of a segmented labor market between the college-educated and those with only high school degrees or less. In the past, a simple aptitude test sufficed for employers, benefiting everyone who could pass the test. But since the law forbade employers from using high school equivalence or general intelligence tests and testing increased the risk of litigation, they turned to college degrees as indicators of general aptitude—a much costlier, and more debt-inducing way of signaling general intelligence and aptitude. As O’Keefe and Vedder conclude—admittedly in a speculative way—employers’ increased expectation of college degrees and the attendant demand and rising costs of higher education have likely backfired in terms of the goal of helping members of minority groups and the less well-off:

　　This increase in college tuition has disproportionately harmed minorities and the poor, as many members of these socio-economic groups are unable to afford college today, even with financial aid. Thus, in spite of the outspoken goals of improving minority employment that presumably motivated the long series of cases and political lobbying described in this paper, it is possible that the Griggs decision has made it harder for some minority candidates to secure jobs.

　　Drawing on O’Keefe and Vedder’s study, George Will highlighted Griggs and its implications as a prime example of the “Law of Unintended Consequences.”

　　What lessons should be taken from this experiment? Broad filtering has no doubt remedied many instances of discrimination. Anti-discrimination laws at the state and national level in the 1940s, ’50s, and ’60s contributed to significant gains in employment for black Americans, for example. But since the 1970s, a further broadening of filters seems to have left us paying high costs for modest benefits: employers must buffer against and engage in litigation, and society must respond to new incentives for signaling, with limited benefit in terms of addressing employment disparities.

　　Solomon successfully filters out the true mother using God-given wisdom, not traditional law. In many cases, traditional law may not be the best tool to serve justice.

　　In the same way, Title IX of the 1972 Education Amendments to the Civil Rights Act has been successful in meeting its core objective of promoting access to education for women. However, it expanded to include a broad mandate for cultural change on gender and a broadly filtered attempt to eradicate sexual misconduct amidst a sexual climate in which shared norms have collapsed.

　　Title IX and the Burden of Proof

　　In criminal law, the filtering problem is relevant to determining the proper burden of proof standard for various offenses. We want our filter to make it likely we only convict the guilty and none of the innocent. If the burden of proof is too high, we risk not convicting the guilty (false negative), but if it is too low, we risk convicting the innocent (false positive). For criminal law, our system ostensibly errs on the side of protecting the innocent, employing the “beyond a reasonable doubt” standard for burden of proof. That means a higher risk of false negatives, which also has social costs. English jurist William Blackstone acknowledged the tradeoff in his famous “ratio”: “Better that ten guilty persons escape than that one innocent suffer.”

　　This becomes important in the debate on the burden of proof for sexual assault cases under Title IX. The Obama Administration required colleges and universities to adopt “preponderance of evidence” as the standard, while the Trump administration required a “clear and convincing evidence” standard, in between preponderance of evidence and reasonable doubt. The Biden administration has returned to the Obama-era rules.

　　Preponderance of evidence is the typical standard for a civil case, requiring only that a proposition is more likely true than not. The clear and convincing evidentiary standard ups the threshold, requiring that the evidence suggests a high probability that an accusation is true. Proponents of the “preponderance of the evidence” standard for sexual harassment in colleges and universities argue that the burden of proof should be low enough to account for universities’ limited investigatory apparatus. Others argue that a lower burden of proof presents the flipside of Blackstone’s ratio, as John Villasenor has explained: “Victims’ rights advocates correctly argue that a lower burden of proof makes it easier to ensure that the guilty are punished. But there is also a mathematically inevitable corollary: a lower burden of proof increases the probability of concluding that the innocent are guilty.”

　　Writing in Law and Liberty, Elizabeth Kaufer-Busch describes the costs to universities and to both complainants and defendants of lax due process standards—broadened filters—for conviction in terms of justice, wellbeing, and litigation costs:

　　Collectively, these due process changes make it easier for an accused student to be found guilty. However, if an increase in guilty verdicts is owing to inadequate due process protections, accused students, whether guilty or innocent, will sue their universities fordefamationand the mishandling of sexual assault complaints, and they will often win.

　　A campus justice system that fails to provide adequate due process harms both the complainants and the accused. Under the 2024 guidelines, innocent students may falsely be accused or punished, leading to tragic consequences such assuicide. Victims/survivors may not see their attackers punished. Improperly handled allegations have resulted inmillionsof dollars awarded to accused students. Such victories for accused students retraumatize the complainant, who may not receive justice.

　　The sexual harassment and assault issue in general is a thorny one. On the one hand, too-narrow filters—high burdens of proof and strong due process protections—likely put many women and men who are sexually harassed at a disadvantage and deny them justice. Evidence in these cases often reduces to a “he said, she said,” problem, unlikely to reach even a clear and convincing evidentiary standard, much less the reasonable doubt standard for criminal cases. Narrow filters may mean powerful people inclined to take advantage of those in subordinate positions face little opposition or incentive to think twice. On the other hand, broadening the filters means sweeping up innocents into the net of persons stigmatized, prosecuted, and convicted for sexual harassment, further incentivizing false accusers who may take advantage of the system. Further complicating all this in the college campus setting is the presence of alcohol and drugs, along with promiscuous sexual norms that cause emotional harm and confusion in situations that result in accusations of sexual misconduct.

　　Goldilocks and Solomon

　　There are no solutions here, only tradeoffs. Social policy problems often require filtering, and policy “solutions” to obvious problems are often just changes in the breadth of our filters. These can have unintended consequences, leaving new problems in their wake.

　　Can thinking in terms of filtering problems help us make improvements? It will help to keep in mind the ideal, Goldilocks solution to a filtering problem: establishing criteria broad enough to include all the people, actions, or events that belong in a category, but narrow enough to exclude those that do not belong. There are no perfect systems, which means that there will always be undeserving people who take advantage of overly broad filtering criteria, and deserving folks left out by too narrow filtering criteria. But understanding that tragic reality does help us think a bit more clearly about social problems and how law and policy affect them. Legal remedies won’t be perfect, and there are inevitable costs to broadening the filters to catch more bad behavior. Continual appeals for renewed civil rights activism create unrealistic expectations about what law can prudently do to perfect society and prevent wrongs.

　　As discrimination becomes less overt and subtler, and persistent economic and social disparities come to be less reflective of overt, legal discrimination and more reflective of other causes, the costs of extending the civil rights vision begin to outweigh the benefits of extending it. No doubt, law helps shape culture—though law is also reflective of culture. But, there are limits to what the law can do and how far the “civil rights vision” can go to addressing the disparities and injustices its proponents seek to mitigate.

　　Recall the story in 1 Kings 3:16–28 where King Solomon successfully navigates a “she said, she said” situation requiring some sifting to determine which claimant is being genuine. When one mother claims the other rolled over on her child and killed it, then stole the living child, Solomon contrives his famous solution of cutting the living baby in half. The false mother is fine with the solution, but the claimant, the true mother of course prefers to give up her child, and the case is solved.

　　Solomon successfully filters out the true mother by his ingenious “solution.” He relies on God-given wisdom, not traditional law. In many cases, traditional law may not be the best tool to serve justice. As criminal justice professor Laura Bazelon, who has represented students accused of sexual misconduct, suggests, the institution of restorative justice practices holds promise in dealing with ambiguous cases. The achievements of laws against discrimination employment and Title IX should not blind us to their costs and the fact that they have not solved the problems of discrimination and sexual harassment, but shifted the terms of tradeoffs that are always present in public policy.