442 AMERICAN BEHAVIORAL SCIENTIST

infraction)? Should there be a presumption against using such techniques, or certain forms of them just as there is with the use of weapons or Fourth Amendment searches, except under special circumstances and when no other practical means are available? How does systematic data searching compare to other means of obtaining information on low-visibility offenses such as undercover tactics and efforts to increase citizen reporting?28.

As in so many other areas of contemporary life, rapid technological development has outpaced the establishment of ethical and legal standards for their use. The important Federal Privacy Act of 1974 does not address many of the issues raised by recent computer developments. Less than one-fifth of the states have laws requiring written standards for the collection, maintenance, and dissemination of person information, though this number is growing. Of course, as time passes and problems are identified the quality of computer use in the areas considered above will no doubt improve. But this is likely to be offset by problems associated with the continuing expansion of computers to new untested areas.

SOME THEORETICAL IMPLICATIONS

The significance of systematic data searching goes beyond the public policy questions considered above. It also has implications for understanding society and the nature of social control. The use of computers as informants is but a small part of a broad social process of rationalization.

The recent growth of matching and profiling is part of a more general process of rationalization that began in the nineteenth century. The same broad social forces affecting the economy touch criminal justice (Spitzer, 1979). In a rational effort to control the environment, policy has become more systematic and routinized. Social control has sought greater effectiveness, efficiency, certainty, and predictability.

Rather than having to rely on what citizens happen to report or police accidentally discover, control agents are taking greater initiative. This may bring greater equity as police seek independence from the biases a citizen-based reporting system may entail. With a skeptical and scientific ethos and a broad data base that can be inexpensively screened, it becomes prudent to consider everyone a possible suspect initially. Analysis rather than tradition becomes the basis for action.

Marx, Reichman / COMPUTERS AS INFORMANTS 443

Eliminating the traditional temporal distinction between locating an offense and searching for an offender may yield greater efficiency. Some systematic data searches collapse these processes as offense and offender are discovered simultaneously.

Yet, just as Mark Twain observed that claims of his death were greatly exaggerated, so too many claims about the efficacy of a rationalized criminal justice system be overoptimistic. In the case of systematic data searching, for example, if it does not contain within it the seeds of its own destruction, it at least contains an ironic vulnerability to its own neutralization (Marx, 1981). In any setting of strategic conflict, efforts at systematization (unless kept secret) can be exploited by skilled adversaries."

The certainty such techniques seem to offer may be illusive. Their advantages may be temporary or may result in a skewed population of apprehended offenders. Routinizing discovery procedures usually involves focusing attention on a limited number of indicators. These may be invested with far more predictive power than they warrant. Focusing attention on specific indicators implicitly diverts attention from other indicators and can result in tunnel vision.30 The indicators chosen can easily come to be treated in a ritualized way. Enforcement agents may be held accountable for following correct procedures, rather than for the results of following those procedures. Only superficial concern may be given to whether or not indicators are valid or have been obtained or presented properly.

While deterring or discovering some offenders, routinization can offer an almost guaranteed means of unauthorized access to others, who gain knowledge of the system and take actions to neutralize it. Altheide (1975) has illustrated how security operations designed to restrict territorial access also can serve as a means for facilitating unofficial entry. The same holds for access to the benefits that systematic data searching is designed to control.

By learning what prompts a hit or a red flag, knowledgeable violators may take steps to avoid them. Some variables used in matching and profiling can be manipulated or avoided easily. For example, the well-publicized match of welfare and bank records in Massachusetts no doubt led some persons to hide money in banks outside the state, to entrust it to others, or to convert assets to a different form.

A different type of neutralization lies in the use of false names and identification numbers. Basic to some contemporary matching is discovering the same name, identification number, address, and the

444 AMERICAN BEHAVIORAL SCIENTIST

like on lists that should be mutually exclusive. This can be avoided through the use of false identification." The name, identification, or record presented may be valid but may simply not belong to the person presenting it. A record check may attest to the validity of the record, but it is unlikely to discover that it does not legitimately belong to the person presenting it.

Publication of the characteristics used to profile arsonists or skyjackers may offer such persons a way to avoid detection. The likelihood of the discovery of an arson pattern through the Property Insurance Loss Registry described above is reduced if each property is in a different and unrelated name. In response to five skyjackings to Cuba in a two-month period, the Federal Aviation Administration is considering changing its behavior profile (New York Times, July 7, 1983).

Awareness of this neutralization potential raises questions about who is likely to get caught in a routinized discovery system. Clearly, not all potential offenders can acquire the knowledge, or have the skill, sufficient to neutralize the system. However, over time it seems likely that these systems will disproportionately net the marginal, amateur, occasional, or opportunistic violator, rather than those who are more systematic, repetitive, skilled, or professional in their rule breaking. The latter ironically may be granted a kind of license to steal, even while headlines hail the effectiveness of control agents using new techniques. 32 To be sure, where costly violation of the public trust or serious crimes are involved, any apprehension may be desirable. But the routinization of discovery does raise a type of equity issue rarely heard. The question is not the familiar one of how authorities use their discretion in deciding what laws to enforce or who to go after, but, given the means they use, what kinds of cases they are likely to discover. 33

Beyond questions of equity, efficiency, and the cyclic and dynamic nature of rule enforcement and violation, there is a broader question about the reach of social control. Observers such as Foucault (1977) view an irreversible continuing historical process of more intensive and extensive social control. The capacity of the modern state to gather information and to punish is seen to extend ever deeper into the social fabric. Control is based on "observation, surveillance, and inspection" rather than primarily on physical coercion. Conformity is

Marx, Reichman / COMPUTERS AS INFORMANTS 445

thought to emerge out of fear of a pervasive and omnipresent panoptic eye. The net has widened and the mesh thinned (Cohen, 1979). While computer matching and profiling may seem to be relatively pale and benign variants of this, variants they are.

How far do we want those in authority to go in their power to discover infraction? In a time of strong citizen concern over crime and the increased prevalence of low-visibility offenses, there is a great deal to be said for enhancing this ability. The proportion of offenses discovered by police relative to those reported by citizens is increasing.

Yet there is another side as well. A different version of the equity problem may appear when there is a gap between the knowledge of violation and the ability to sanction. While ignorance is not bliss, there is a certain wisdom to the inability of the three monkeys to see evil when action cannot be taken with respect to it. Powerful new discovery means may overload the system. Authorities may discover far more violations than they can prosecute or process. This overabundance can lead to the misuse of discretion and demoralization. Charges of corruption and favoritism may appear and the system may be perceived as unfair.

If this were all that was at stake, awareness of the potential problems and well-conceived policy for structuring choices might suffice. But there is a more onimous side. Paradoxically, both repression and equal law enforcement may be inhibited when authorities lack information. As Selznick (1948: 84) observes:

Do we need or want agencies of control so efficient and so impartial that every actual offense has an equal chance of being known and processed?... I am concerned that we do not respond too eagerly and too well to the apparent need for more effective mechanisms of social control. In the administration of justice, if anywhere, we need to guard human values and forestall the creation of mindless machines for handling cases according to set routines. Here vigilance consists in careful study of actual operations so that we may know what will be lost or gained.

Systematic data searching, along with the new citizen reporting programs, undercover police practices, electronic surveillance, and other technical means, offers compelling and little understood arenas for such study.

446 AMERICAN BEHAVIORAL SCIENTIST

NOTES

1. We are defining "police" very broadly. By "police" we refer to those charged with the policing function, regardless of what the formal title is. All persons who enforce rules must confront issues around the discovery of their violation.

2. Such programs do generate information. For example, a Baltimore cail-in radio program. “Report a Pusher," led to 91 arrests on drug charges. During the 4-hour program, police appealed to citizens for information on drug trafficking. Off the air, detectives took calls and recorded names, license numbers, and other information about persons callers suspected of being involved in narcotics transactions (New York Times, November 7, 1982). In Michigan, $1000 is offered for information leading to the arrest and conviction of arsonists. From the inception of this reward system in 1975 to 1981, 26 payments were made (Arson News, 1981). What is not usually considered is how much of the information provided would have been forthcoming even in the absence of such programs.

3. Thus federal and in many places state legislation and judicial decisions have offered new protections for whistle-blowers. The Federal Witness Protection Program provides relocation and a new identity to informants (see Montanino, this issue). Legislation has also introduced negative sanctions for not reporting things such as child abuse and certain hazardous working or environmental conditions.

4. The methods are not mutually exclusive. For example, a lead generated by a hotline or a computer search may lead to an undercover operation. Computers, of course, are part of a broader family of rapidly developing technological means, including electronic surveillance and forensic science, also used to enhance discovery.

5. For example, it contrasts with a New York City program called “CATCH" (computer-assisted terminal criminal hunt) designed to streamline the identification of suspects. CATCH is a computerized "mug book" permitting quick retrieval of names of suspects who fit the description fed into the system. Computers have simply improved upon a traditional tactic (Computerworld, April 7, 1980).

In focusing on the discovery of offenses we are also referring to something beyond merely checking a second data base to find a person's address (such as the Selective Service's use of IRS data to locate people suspected of failing to register for the draft) or using that data base for sanctioning purposes, as with the state's garnishment of income tax refunds due to fathers who default on child support payments.

6. In 1959 entitlement programs accounted for 15% of the federal budget; in 1970 such expenditures had increased to one-third of the $62 billion budget; by 1981 they were $300 billion-almost half the budget.

7. See Katz (1978). Vaughan (1980), and Altheide and Johnson (1980) for discussions of the way task differentiation and bureaucratic organization can shield deviance and neutralize control.

8. Matching across data bases, which is one of our concerns here, shares much with the more traditional and common searching of a single data base. At an abstract level the correlation of distinct information involves the same logic of inquiry. But the former raises questions of privacy and data compatibility (which may have implications for errors and misinterpretations) not found when a single data source