Fragmentation in timelines

Nota

To learn more about the fundamentals of dates and timelines in Security Force Monitor’s approach read Claims with dates first.

Answering the question “who was where when?” is central for investigations into allegations of human rights abuse(s). Because of this perhaps one of the most defining, and complicaticating, features of the Security Force Monitor’s data is that almost everything we research is connected to time including:

  • Existence of units

  • Parent relationships between units

  • Location of units

  • Areas of operation for units

  • Membership/participation of units of in multi-unit operations

  • Positions held by people

While attaching time to data points aids our mission to support human rights investigations and advocacy, it raises methodological challenges and questions such as:

  • Why the Monitor would (or would not) connect two bits of data through time

  • How the Monitor handles gaps in the public record

  • Questions analysts run through while reviewing time based information

In an ideal world the Monitor would have a source from every day of the year stating where a unit was located or conducting operations. Barring that, having multiple sources regularly making statements like “since X date this unit has been based in this city” would be tremendously helpful. Unfortunately, neither scenario currently occurs, or is likely to occur in the near future, making it necessary to develop a robust way of thinking through time.

Broadly speaking the Security Force Monitor uses agreement among sources to build up details on security force units and individuals. Most of the Monitor’s sources, like government press releases and newspaper articles, can be used to link a value, such as the location of a unit, to a specific date (usually the date of publication). As we collect more sources we need to determine what agreement among sources means for time based values, like the location of a unit.

Example: the Monitor finds Source A published on 1 July 2012 stating that the 1 Battalion is based in Lagos. If Source B published on 3 August 2012 also states that the 1 Battalion is based in Lagos we have a decision point about what claim we should make.

Using sources A and B we have two options which can be expressed in text:

  1. Separate claims: “As of 1 July 2012 the 1 Battalion was based in Lagos and as of 3 August 2012 the 1 Battalion was based in Lagos, the Monitor does not know where the battalion was based between those two points in time.”

  2. Contiguity claim: “From at least 1 July 2012 to at least 3 August 2012 the 1 Battalion was based in Lagos.”

Thus, whenever the Monitor gets a new source of information we have to decide whether to make a “separate” or “contiguity” claim. Based on the example of the 1 Battalion above the Monitor would run through a series of questions to determine which claim (if any) to make:

  • In general, how do other battalions operate, are they sedentary, or highly mobile?

  • How has the 1 Battalion acted in the past, has it been sedentary or highly mobile?

  • Are there other sources disputing these claims (i.e. 1 Battalion being based solely in another city)?

  • Are there any sources indicating the 1 Battalion was in Lagos in July and/or August as part of a “special”, “emergency” or otherwise temporary posting?

  • Are there sources that indicate the 1 Battalion moved in between these two points of time and thus these should be treated as separate deployments to Lagos?

  • Is there anything related to the 1 Battalion’s parent or child units that may impact where it was based?

  • Are there any other mitigating sources (i.e. major restructuring of the military, constitutional changes, etc.) which may impact the basing of the unit?

  • Is more research needed before the Monitor can make any claim?

An argument could be the Monitor should always make “separate claims” as that would be more faithful to the sources. However, the result likely mean an almost incomprehensible amount of detail in the records of people and units, which would obscure when changes really did occur, for instance when a person changed positions or a unit ends operations in an area. To some extent, our claim-based system (see Claims with dates) allows us to manage this complexity more easily. It gives us the ability to check specific underlying dates and data points, to plot timelines that have both claimed and inferred date ranges, and to make choices about continutiy as described in this article.

Perhaps the most important point is that even though data points can be continuous through time, it should never be assumed that those types of features remain consistent between two or more sources. Time is a constant challenge, but given that is a key element in identifying perpetrators of human rights abuses it is necessary to get it right.