Discrimination should be understood as action and therefore distinguished from prejudice, which is a matter of thought, attitude, or belief. Racial discrimination then would consist of social action that produces unjust allocation of valued resources, based on racial categorizations of individuals or groups (Banton; Kairys, 1996). This is the standard definition and still serves in many cases: where race is concerned, it provides basic standards for distributional justice, and it permits one to disaggregate prejudice and discrimination (Merton). What the standard definition does not provide, however, is a link between unjust action and social structure, and this presents serious difficulties both for social thought and for social justice.
Patterns and understandings of racial discrimination have undergone significant change during the years since World War II. Such long-established practices of racial favoritism as the South African apartheid system and the system of state-enforced racial segregation in the U.S. South (that is, "Jim Crow") have been overturned in a complex sociopolitical process that combined sustained social movement activity with state-based racial reform. As a result, the formal (de jure) rules and agencies of racial preference have been largely if not entirely jettisoned, and explicit practices of discrimination on the part of resource holders and gatekeepers, such as employers, landlords, service providers, and schools to name but a few, have been stigmatized, though these practices often persist in more concealed and publicly deniable fashions.
It is undoubtedly a significant issue that an individual, group, or agency unfairly withholds resources (say, mortgages, university admissions, or trade union membership cards), denying access on an equal basis to persons of a negatively valued racial identity or, conversely, granting privileged access to persons of a favored racial group. But action of this type is identifiable and sanctionable, precisely because it is action: that is, transitive and predicated in the material present. Of course, such action often is not identified and very often is not sanctioned; it remains widely accepted or at least tolerated. But it is at least recognizable and comprehensible as the work of particular subjects, whether individual or collective.
Past racial discrimination, though, is much more difficult to address directly. Because discriminatory action is accretive and cumulative over time, especially over long durations, its effects can be felt without the presence of any active agent. Prolonged, transgenerational access to socially valued resources, such as elite university degrees, high-status jobs and professions, political power, and culturally afforded honor of various kinds (not to mention wealth), is facilitated by belonging to the "right" racial group (Harris; Feagin). Conversely, impeded opportunities, low status, culturally attributed dishonor (e.g., stereotypes of laziness, ugliness, or low intelligence), and frequently impoverishment are all outcomes of prolonged, transgenerational denial of access to resources and opportunities (Oliver and Shapiro; Brown et al.).
Thus racial discrimination is more than social action; in its repeated and reiterated forms it develops into a social structure. It becomes habitual and is taken for granted in all sorts of invisible ways: in the design of neighborhoods, the making of foreign policy or tax policy, or the writing of television news; in the way one is addressed by a policeman, treated by a doctor, or expected to speak the language.
This invisible structural discrimination evades attempts to regulate it, for example, through the law. After the 1960s civil rights reforms of state racial policy in the United States, for example, and after the 1994 dismantling of the apartheid regime in South Africa as well, the social structure of discrimination still exhibited a great deal of resilience (Winant; Zegeye). Some of this persistence may be attributable to the weakness of reform policy and the reluctance of courts to intervene strongly in sanctioning discriminatory practices (Kairys, 1993; Ezorsky), but a great deal of discrimination escapes public attention (sometimes even from those it damages) because it has little or no active agent, is hardly or not at all the work of intention, and is regarded as normal practice. This "transparent" discrimination in effect has no living perpetrators; it is the legacy of such past practices as slavery and colonial rule.
The presence of structural racial discrimination also has the effect of legitimating much intentional discrimination, which can be rendered agentless simply by the refusal to intervene on behalf of principles of fairness or equality as resources of social value are allocated. In large-scale terms (such as determining housing or wage policies), in medium-scale terms (such as carrying out university admissions), and in small-scale terms (such as hailing a cab, or "driving while black"), present-tense, active racial discrimination is supported and camouflaged by structural discrimination.
Thus state regulation alone is insufficient to effectively eliminate discrimination. Of course, discrimination can be maintained by state regulation: notably in the United States but elsewhere as well, courts have demonstrated how effectively they can sustain discrimination at socially acceptable levels. For example, U.S. courts have established legal standards that turn on the demonstration of explicit intent to discriminate as the prerequisite for enforcement of civil rights laws. In response to this, explicit profession of an intention to discriminate has become quite rare, and active discriminators have learned to dissemble and mask their practices under such code words as "colorblindness" and "merit" (Brown et al.).
Far more intensive measures have been proposed to eliminate or at least seriously reduce structural discrimination. These measures would be based on a policy of redistribution of resources according to a compensatory formula, sometimes viewed as reparations for past discrimination, and would also contemplate large-scale social therapy and healing projects, such as have occurred after the fall of dictatorships and after the military defeat of repressive regimes. Examples of such intensive projects include the Reconstruction period after the U.S. Civil War (Du Bois), the promulgation of a new South African constitution after the end of apartheid in 1994, and the "denazification" and supervised democratization processes that followed the defeat of the Axis powers in 1945.
While juridical processes, such as trials of human rights abusers and state-sponsored commissions like the South African Truth and Reconciliation Commission (TRC) or the Argentine National Commission on the Disappeared (CONADEP), might be expected to play a role in these transformations (Weschler), important work also would have to be undertaken at the level of civil society, of everyday life. Such activity would require an advanced commitment to democracy. Some of the TRC's innovations, such as provision of guarantees of nonretribution to those who acknowledged their own abusive practices, may serve as an initial model. While developing the large-scale dedication needed for such intensive antidiscrimination measures would be very difficult, preliminary work done by movement activists and scholars on the implementation of reparations suggests future directions (Bittker; Brooks; Dawson and Popoff).