The don’t-ask-don’t-tell era of racial preferences in college admissions may soon be at an end, as Abigail Fisher’s challenge to the University of Texas’s affirmative-action program makes its second appearance before the Supreme Court, which will hear the case this Wednesday.

Significantly, Ms. Fisher isn’t asking the Court to ban affirmative action. Instead, her case seeks to hold schools to the general rule that the government may employ race-based measures only as a last resort. And even then, such measures must be almost perfectly calibrated to serve a compelling interest — in this instance, achieving the educational benefits of diversity.

In the admissions context, those principles have too often been honored in the breach. And for that, blame the Court. Its 2003 decision upholding the University of Michigan Law School’s affirmative-action program combined the tough language typical of decisions reviewing race-conscious government policies with a loose and open-ended analysis of the way the program actually worked and the way it was justified.

University administrators took the decision as license to do what they pleased, never mind necessity or tailoring, so long as they stayed vague about the way their programs worked. Admissions at UT–Austin offer a case in point. In 2008, the year Ms. Fisher applied, the bulk of students (81 percent) were admitted under Texas’s Top Ten Percent law, which grants automatic admission to top students at Texas high schools. That alone made UT–Austin one of the most racially diverse campuses among elite public universities.

Nonetheless, the university layered on top of that base a race-conscious admissions program. The justification — which has changed several times over the seven years that the university has spent fighting Ms. Fisher’s lawsuit — was that the Top Ten admittees lacked what it called “qualitative diversity” or “diversity within diversity.” In other words, university officials felt, despite never having surveyed the relevant characteristics of minority students admitted under the Top Ten law, that they somehow lacked adequate diversity among themselves.

The resulting race-conscious program is called “holistic review.” The program bases admissions on a combination of academics and “personal achievement.” As part of the process, an admissions reviewer assigns each applicant a “personal achievement score,” ranging from 1 to 6, based on a laundry list of factors, including race. After the scores are assigned, applicants are selected, major by major, on the basis of grids that chart academic achievement against personal achievement. That means there is no way to know whether or how the use of race influenced any particular admissions decision.

If even this stripped-down summary sounds convoluted, there’s a reason for that: It is convoluted — and, by all appearances, deliberately so. The purpose is to obfuscate. And that’s a real problem for the university. Even putting aside whether UT can justify using race at all, given the enormous diversity it has achieved through race-neutral means, its holistic-review program is completely divorced from its rather specific “diversity within diversity” justification.

If one were trying to boost qualitative diversity, whatever exactly that may be, UT’s approach isn’t what any sane person would do. To begin with, it’s astonishingly arbitrary. Despite the enormous emphasis that admissions officials place on racial considerations, the decision of when to use race as a “plus” factor and how much weight to accord it are left entirely to the application reviewers, without specific guidance or oversight. The idea, presumably, is that they know what they’re looking for.

A sane person acting in good faith would place emphasis on transparency. UT does not. To the contrary, its holistic-review process could not have been made more opaque. Even the university has no way to oversee decisions regarding race because it has structured its process so that those decisions cannot be disentangled from the consideration of other factors.

Indeed, UT has gone to such lengths to obfuscate its use of race that it can’t even show that its application readers aren’t treating race as the defining factor in the applications they review, which would amount to a forbidden quota system.

A sane person would also focus on results. But the results of UT’s use of race are unmeasurable. The university cannot identify students admitted because of racial preferences and therefore has no ability to identify their characteristics or ascertain the impact of racial preferences on diversity at any level. In fact, UT’s admissions director conceded that he could not identify any applicant who had been admitted on the basis of race. He also didn’t see why that was a problem.

None of this is unique to UT. Many of the affirmative-action programs administered in the wake of the Court’s 2003 decision are similarly structured. Dissenting from the majority opinion in that case, Justice Anthony Kennedy warned that undue deference to schools would allow them to cite vague diversity interests as a pretext for unconstitutional discrimination. And that’s exactly what has happened.

The difference today is that the author of the Court’s 2003 decision, Justice Sandra Day O’Connor, has been replaced on the bench by Justice Samuel Alito, who hews more closely to Justice Kennedy’s views on this subject. A constitutional corrective is in order.