The issue of affirmative action continues to dominate the racial justice landscape. It frequently dictates the conceptual and policy framework for increasing access to education and employment. Its preservation has largely defined the strategy and goals of many civil rights organizations. Its operation often substitutes for a more comprehensive effort to address the fundamental structural distortions in the way we allocate educational and employment opportunities for everyone.

“The Future of Affirmative Action” is an effort to shift this preoccupation with affirmative action as the organizing framework for addressing access to education and employment, both generally and for women and people of color. Affirmative action is, in our view, a program, rather than a vision. It may be an appropriate short-term program, but it is not an adequately conceptualized long-term strategy for pursuing equity and efficacy in education and employment. Especially when affirmative action is the primary strategy for racial justice, it offers a narrow, at-the-margins response to exclusion, which deflects attention from more central problems with the current system for allocating access. Our goal is not, as Stephen Steinberg presumes, to replace “affirmative action as we know it,” or to provide a substitute because courts have begun to invalidate traditional affirmative action programs. It is, instead, to shift the overall framework of the debate. We strive to focus attention on developing a long-term, normative vision of racial justice that corresponds to present circumstance, and on generating experiments in institutional redesign that can address problems at this more structural level.

The responses to our essay reveal the challenges we have yet to meet in this effort. Perhaps the most intractable challenge, although the easiest conceptually, is to dislodge the prevailing commitment to short-term definitions of efficiency and merit. For example, quibbling with the validity of tests for predicting first-year performance in school largely misses the point. The central problem with tests’ role in selection is how they are used–to rank-order applicants at the margins, to exclude applicants who could do as well, and to stand in for defining values and institutional goals. As Peter Sacks says, “the cult of measurement” and the “prevailing and entrenched hegemony of test scores” triumph over “one’s actual achievement” in the real world. Like Sacks, our concern is who gets excluded despite their capacity to succeed, rather than whether those who perform well on tests are also screened for other traits. It is how our institutions define value in ways that, as Howard Gardner puts it, are heavily skewed toward a few measurable “end states,” often at the expense of more important values.

Unlike Ward Connerly, we are not confident that those who do less well on high-stakes tests in fact do worse in school or on the job. Both anecdotal and quantitative evidence suggests that many tests exclude applicants who could in fact perform successfully. For example, preliminary evidence on the impact of the Texas Ten Percent Plan, which admits the top 10 percent of the graduating class from every high school in the state, suggests that SAT-type tests exclude candidates who would be able to succeed. Peter Cappelli acknowledges this general claim–that high school grades have a less discriminatory impact–and suggests that alternative admission practices are a sensible option. His confidence is not misplaced, based on data from the first two years of the Texas program. Indeed, those admitted pursuant to the 10 percent standard, based on high school grades alone, outperform peers who were admitted based on traditional SAT-type tests. The ten percenters have a higher freshman grade point average than those admitted using conventional criteria, apparently because the ten percenters see themselves as successful, have drive and self-confidence, and are willing to seek help and ask questions when needed.

A second challenge that we face based on the responses to our essay is to move the public conversation and practice beyond the traditional dichotomies, such as formal/informal, public/private, normative/instrumental, racial justice/merit, that narrowly construct our understanding of the universe of choices available for pursuing racial justice.

For example, Stephen Steinberg and Paul Osterman argue that moving beyond dominant reliance on tests necessarily leads back to the systems of informal, subjective, and biased decision-making that tests were in part developed to prevent. This concern is well founded, but in our view, unduly static and reactive. The approach to selection need not simply reflect a choice between these polar alternatives. We are urging active engagement and experimentation with reconfiguring institutions to permit more accountable and transparent decision-making, which will make the inevitable exercise of discretion more fair and functional.

Such institutionally based responses are increasingly crucial. The dynamics of “second generation” racial and gender exclusion–bias resulting from patterns and structures of interaction–are often complex, culturally embedded, and subtle. They do not necessarily emerge from the conduct that produced the first generation of inequality and activism. We appreciate Michael Piore’s concern that unless we organize our remedial strategies around the original normative account of racial inequality, which was based on systematically and deliberately subordinating people of color and women, we necessarily abandon any principle of racial justice. But this account would place the more structural, second-generation forms of bias beyond the scope of either normative or legal inquiry. Complex conditions, such as those producing continuing racial and gender inequities, require structural solutions with multiple normative justifications. The assumption that we must choose a single normative account based on the conditions that prevailed when the civil rights laws were enacted, in our view, unnecessarily polarizes and constrains the choices facing those concerned with racial justice.

For this reason, we do not take the position that we should “end affirmative action as we know it.” Indeed, because our goal was to shift public discourse, we did not explicitly take a position in the current affirmative action debate. We agree with the comments of Mary Waters, Carolyn Boyes-Watson, Deborah Kolb, and Maureen Scully, who suggest that racial and gender inclusion is a both/and proposition. Institutions that cannot, or will not, address the problem of racial and gender inclusion more structurally may need some form of affirmative action at least as a stopgap. Yet, narrowly defined affirmative action programs are inadequate long-term solutions to problems of access and participation. Sustainable solutions will require an approach that links issues of racial and gender justice with changing institutions, experimenting with new pedagogy, redefining long-term goals, and clarifying notions of democratic participation.

The comments of Waters, Boyes-Watson, Kolb, Scully, and Howard Gardner highlight a third important challenge for subsequent work: identifying and encouraging the development of workable ideas about how to allocate educational and employment opportunities. Waters and Boyes-Watson propose a weighted lottery for the top 1 percent of high school students. Kolb and Scully demonstrate the promise of experimentation by transforming our performance-based test of Bernice into a new form of collaboration and teamwork in the workplace. Howard Gardner proposes using apprenticeships and technology to create opportunities to learn, and then to base future opportunities on how students make use of those opportunities.

Our current work has taken the next step to document and extrapolate from these emerging experiments with structural approaches to racial and gender inclusion.1 This work shows that forward-looking companies, non-governmental organizations, advocates, and courts have begun to develop more structural approaches that connect issues of racial and gender justice with broader institutional transformation. These examples establish that effective, legitimate, and accountable processes can emerge, and offer a way to develop contingent criteria that could assist in the evaluation of future processes.

For example, Deloitte and Touche, America’s third-largest accounting, tax, and management-consulting firm, implemented a major Women’s Initiative, which dramatically increased women’s advancement in the company and reduced the turnover rate for women in particular and employees in general. The firm accomplished this by forming ongoing, participatory task forces, and giving them the responsibility to determine the nature and cause of problems, make recommendations and develop systems in order to address them, and then monitor the results. The task force recommendations were implemented through ongoing data gathering and analysis, operational change through line management, and accountability in relation to benchmarks. This approach offers a structured set of opportunities for collective action by women’s groups oriented around addressing problems of immediate and direct concern. The Women’s Initiative produced swift and observable results, both in women’s participation and in the firm’s overall retention rate. The combination of increased communication and programmatic change contributed to what many called a culture change. Flexible work has become acceptable at Deloitte, for women and men.

Home Depot also faced the problem of how to minimize the expression of bias in a highly discretionary process of hiring and promotion, in a company that was dynamic, decentralized, and entrepreneurial. The solution was to achieve accountability through technology, information systems, and systematizing discretion, rather than through rules. The keystone of the new system is an automated hiring and promotion system, called the Job Preference Program. This process virtually eliminates the possibility for managers to steer applicants to particular roles based on stereotypes, expands the pool of applicants for every position, and opens up avenues for advancement that applicants themselves may not otherwise have considered.2 As a result, rates of participation by women have risen and employee turnover rates have dropped across the company. People of color are participating at higher rates. And, the company has begun to track and use information from its Open Door Dispute Resolution system as a problem-solving tool.

Both examples illustrate the promise of, and need to learn from, experiments that use institutional innovation to pursue racial and gender equity as a catalyst for more comprehensive institutional change. We mention them not as solutions, but as examples of the possibility of change–in how race and gender issues can be addressed to permit the structured and accountable exercise of discretion; in how institutions can actively define long-term goals and institute systems that enable those goals to be pursued and continually re-evaluated; and in how racial and gender inclusion is connected to questions of institutional efficacy on one hand, and basic values of democratic participation on the other. We are confident that others will come forward to elaborate more fully the concrete possibilities, to help reframe the debate, and reclaim affirmative action’s innovative ideal.


1 See Lani Guinier and Gerald Torres, The Miner’s Canary; Susan Sturm, “Second Generation Employment Discrimination: A Structural Approach,” Columbia Law Review 101.1 (2001).

2 These examples, and their implications for the regulation of workplace bias, are developed more fully in Sturm, “Second Generation Employment Discrimination.”