1.2 Why use a formal system for information management?

If a single person were doing an investigation, her notes might be cryptic and brief, sufficient only to jog her memory. If this investigator were collaborating with only a few other people working on a few cases, then the group might meet periodically to exchange all the relevant ideas and information. However, these simple and informal ways to manage information are likely to be inadequate in larger projects. For example, if, during the course of the investigations, some people leave the group and new people join, there must be a way for the new people to learn everything that the departing people knew. If the group consists of many people who may be working in widely-dispersed areas, it may be impractical for each member to teach every other member everything she has learned since the last meeting. If the organization is investigating hundreds or thousands of cases, there must be a systematic way to file information with the relevant cases as the information comes to different members of the organization in tiny increments. In short, in all but the smallest, briefest, and most informal investigations, there must be a way for the group to remember as a group what individual members have discovered.

This is the difference between people working alone and people working together in an organization: an organization has a memory which persists and extends beyond the participation of any particular member. An organization "remembers" by designing and using a good information management system. The benefits from such a system almost all derive from the intellectual work that people in an organization do classifying information according to definitions that the organization establishes. The information management system gives the organization a way to accumulate many individuals' systematic efforts. Thus the organization's memory can slowly grow to be greater than any of its member's memories.

The general rule -- garbage in, garbage out -- is particularly important for a human rights database. However, the design of the information management system has implications that go far beyond computing. For example, if the organization conceptualizes a human rights event as a one kind of violence committed by a single perpetra-tor against a single victim, the organization has already ruled out any analysis of violence that happens to many people at the same time. And, in fact, much political violence is directed against groups -- whole villages, trade unions, and peasant organizations are frequently targets of repression. Massacres, wholesale detentions, and destructive searches of organizations' offices are common kinds of attacks. To understand mass events, the victims cannot be parcelled out into unrelated individual cases. Instead, attacks against all the named or unnamed victims must be maintained as part of a single conceptual event while at the same time all the individual stories are kept organ-ized according to the person who gave them. If we fail to keep the representations of violations that happened to victims in a single event together, then later analysis of this event will show many unrelated individual cases. We will have lost the coherence of the mass event, and therefore, we will have eliminated the possibility of analyzing the difference between targeted versus indiscriminate violence.

Similarly, information about each of the violent acts committed against each victim must be maintained in a meaningfully related sequence. If a person was detained and tortured, the information about how she was tortured must be kept in some relation to the information about the detention.

But the garbage in-garbage out rule still relates to the database. If the information inputted to the database is already confused or improperly organized by bad collection techniques, having a very sophisticated database will not help much. The converse is also true: if an organization collects a mass of very detailed interview and investigation data, and then they put the data into an improperly designed database, the computer-generated reports will reflect errors in the database rather than trends in the data. Designing an information management system means figuring out how, exactly, all of this information is kept bundled together during interviews with witnesses and victims, in data processing and coding, in the database, and then in the analytical reports. The hardest part of the design is ensuring that each part of this system is coordinated with the other parts.