The Army has been a model of successful racial integration. It has promoted large numbers of African-Americans—Colin L. Powell, for one—and other minorities into the once lily-white officer corps. It appears to have done so without much clamor about reverse discrimination by passed-over whites. The white, black, Hispanic, and Asian soldiers now fighting side by side in Afghanistan appear largely free from the racial tensions that alienated many black foot soldiers from their white officers in Vietnam three decades ago and sapped the Army's effectiveness as a fighting force.

So it came as a shock on March 4 when U.S. District Judge Royce Lamberth struck down as unconstitutional what he called the Army's "use of racial and gender classifications in its promotion policy." His decision hands a very hot potato to the Bush Administration, which will anger its conservative base if it brings an appeal defending racial preferences, and may inflame liberals and the media—while disappointing many Army leaders—if it does not.

My first reaction on learning of Lamberth's ruling was: Why did he have to go and do that? The Army has a good thing going. I don't know exactly how they do it, but it seems to work, and it hasn't caused a big public fuss.

But it turns out that the Army's affirmative action policies have embittered more white males than I had imagined. Some of them have sued, including Raymond Saunders, a retired lieutenant colonel, who claims that the Army's denials of his bids for promotion to colonel in 1996 and 1997 were tainted by racial and gender discrimination. Lamberth rejected the Army's motion to dismiss Saunders's lawsuit and kept his claim for monetary damages alive. The judge's 68-page opinion makes a strong case that the detailed "equal-opportunity instructions" provided to the Army "selection board" that evaluated Saunders for promotion amounted to a "statement of preference for minorities and females" and that they "clearly imply that disproportionate promotion [of whites or males] is in some way a disfavored result." (The current instructions, revised in 1999, were not directly at issue.)

The equal-opportunity instructions did not in so many words mandate proportionate or supraproportionate promotions for minorities and women. But it's hard to miss the message. One provision stated: "Your goal is to achieve a selection rate in each minority and gender group ... that is not less than the selection rate for all [eligible] officers." Another provision required selection boards to "identify" and "explain" any "situation where a particular minority-gender subgroup did not fare well in comparison to the overall population." Still other provisions repeatedly stressed the need to "be alert to the possibility of past personal or institutional discrimination" against minorities and women. There was no need for selection boards to explain, on the other hand, if white males did not "fare well" or win a proportionate share of promotions, and no need to consider the possibility of past discrimination against them. And while the instructions did caution against using a "quota," Lamberth held that "a defendant may not cleanse a policy of an impermissible preference merely by disclaiming that preference."

Lamberth also ruled that the evidence does not establish the kind of recent discrimination that is necessary, under the relevant Supreme Court precedents, to justify the drastic remedy of racial or gender preference. The judge found no evidence that the Army's legacy of discrimination continued into the 1990s, when the equal-opportunity instructions were adopted. Disproportionate promotion rates do not necessarily prove discrimination; they may be attributable to racial disparities in average educational qualifications, for example. In any event, Lamberth noted, "the promotion rate of black officers in the five years preceding the adoption of the 1993 policy ... suggests that such officers were promoted at rates comparable to, and sometimes greater than, their white colleagues." Women had also won a proportionate share of promotions in that period.

As these numbers suggest, Lamberth's decision (even if upheld in any appeal) seems unlikely to reduce drastically the percentage of promotions going to African-Americans and women or to doom the Army's hope for a more diverse officer corps. The main reason appears to be that a healthy desire to overcome our legacy of discrimination, and to increase the number of minorities and women in leadership positions, has been woven into the culture of the Army and into that of many other institutions. This cultural reality helps explain why the abolition of governmental racial preferences in California (by ballot initiative) and in Texas (by federal court decision) in 1996 has not led to the "resegregation" of the University of California or the University of Texas, as was feared.

No doubt this culture of affirmative action injects a subtle, often-undetectable element of racial or gender preference into the mind-sets and subjective judgments of many of the people who allocate promotions and other coveted positions. But this phenomenon produces a lot less of the bitterness among white males—and sends less of a stigmatizing, you-might-not-make-it-on-merit-alone message to others—that are the toxic byproducts of formalized, governmentally mandated preferences.

Such overt, in-your-face counting by race can, as the Supreme Court stressed in 1989 in Richmond v. J.A. Croson Co., "promote notions of racial inferiority and lead to the politics of racial hostility." Even in the Army, it appears. Although "officers don't like to air their dirty laundry," asserts Christopher Sterbenz, who was the lawyer for Saunders and 11 other military men who have brought similar lawsuits, "within the officer ranks, this has actually become a matter of bitter debate."

This distaste for systematic, group-based preferences is not confined to disappointed white male Army officers and conservative judges. It also pervades the general public, across racial lines. Consider a joint Washington Post-Harvard University-Henry J. Kaiser Family Foundation poll of various racial groups last year. When asked whether "race or ethnicity should be a factor when deciding who is hired, promoted, or admitted to college," or whether such decisions should be "based strictly on merit," one group chose "strictly on merit" by a whopping 86 percent to 12 percent. And that was the African-American group. The margin among whites was 94 percent to 3 percent.

The vague, euphemistic phrase "affirmative action" gets much more support in polls. And the culture of affirmative action has become a background reality that people have learned to live with, just as short men and plain-looking women have learned to live with the subtle advantages that our culture confers upon tall men and good-looking women. We want the rules of our games to be fair; we know that life never will be.

As for formal, written affirmative-action policies, nothing in Judge Lamberth's decision would (in my view) prevent the Army from stressing—in race-neutral terms—that "the diverse backgrounds, ideas, and insights offered by soldiers and citizens of all races and of both sexes are a great source for our nation and our Army," that selection boards "must be alert to the possibility of past personal or institutional discrimination," and that they should demonstrate "that race, ethnic background, and gender are not impediments."

The language quoted above is, in fact, the heart of the equal-opportunity instructions currently used by "centralized" Army selection boards. These instructions, revised in 1999, also specify that the "foregoing guidance shall not be interpreted as requiring or authorizing you to extend any preference of any sort to any officer or group of officers solely on the basis of race, ethnicity, or gender."

Judge Lamberth has suggested that the current promotion policy may still be unconstitutional. And some provisions of the Army's overall affirmative action plan, to the extent that they call for proportional representation by race and gender, may have to go. But the current equal-opportunity instructions would be constitutional, in my view, if one word—"solely"—were deleted from that last sentence. They would also, I'd bet, look constitutional to the Supreme Court. So the Army may already be on the way to solving its constitutional problem.

The solution—race-neutral affirmative action—may not (and I hope will not) have a dramatic impact on promotion rates. Indeed, the difference may be mostly one of appearance. But it is the appearance of a formal double standard, more than the culturally embedded impulse to look for ways to increase diversity, that causes most of the racial conflict and bitterness associated with affirmative action. That's why most institutions that systematically use racial double standards deny doing so.

And that's why some of us who admire the Army's objective of diversifying its officer corps are troubled to see it using methods that smack of a double standard. Formal preferences may have been necessary years ago. The optimistic view is that they are on their way out—and that we don't need them anymore.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.