Worth A Thousand Words: Mapping Mass Murder

Prototype heat map designed by author in ARCGIS. Even though scraping the Death Lists provides the historian with unprecedented - and personal - detail on the geography of killing in Afghanistan in 1979, depicting that information is a task of its own. Important is finding a balance between vivid data, sensitivity to place and topography, and the stories of the victims themselves.

If, as the cliche goes, a picture is worth a thousand words, then part of the reason why may be that it can take much, much longer to put together the former than the latter. Writing prose may not always be easy, but mapping data in a way that actually communicates the desired information to the user can take hours, as we discovered in the process of putting together Shattering Afghanistan.

Particularly with biographically and personally sensitive information like that contained in the Death Lists, it is of crucial importance that historians take the care to think about their mapping choices, and to explain and justify the choices they do make. They also must remain aware of the paths not taken and the particular narrative emphases that "mere" different visual codes may communicate. In this section of Shattering Afghanistan, before delving into the maps derived from the Death Lists, a few words on how some of these maps were generated bears discussion.

A simple Fusion Tables map of all of the places of origin of those registered in the Death Lists. Note the dots in Peshawar (used as a proxy for "Pashtunistan") and Tehran (used, however imperfectly, as a proxy for "Iran").

Back to Basics

Once I had compilated all of the data from the Death Lists - itself a time-consuming task - the first task on my mind was to compose a "simple" map of the places of origin of all of those killed by the PDPA in 1979 as reflected in the sources. (Again, it bears emphasizing that, as best we understand, the lists reflect the place of origin of those killed rather than the location where they were murdered). Whether operating in ARCMap or Google's Fusion Table application, this proved relatively simple - deceptively simple, even. As the chart to the right shows, the victims of the PDPA's terror during 1979 came from all over Afghanistan. Strikingly - and underscoring the extent of trans-national mobility in Central Asia during this period - nineteen victims came from Iran or Pakistan (one and eighteen, respectively). 

Yet these preliminary maps arguably obscured as much as they revealed. Within Afghanistan itself, a basic analysis of the place of origin data failed to "stack" the data in a way that emphasized the over-representation of places like Kabul or Behsud. As for those originating from outside of Afghanistan, the geographical vocabulary of the PDPA's record-keepers is suggestive. The man coming from Iran is listed as merely coming from "Iran," while several of those who came from Pakistan are listed as being from the "Free Frontier Area" (probably the Pakistani FATA) or "Pashtunistan." This raises questions of geographical imagination that are arguably beyond the reach of GIS specialists, at least with the (relatively) small amount of data we have in the Death Lists. Even as Pakistan was a separate country recognized by almost all of the international community, it bears remembering that to this day no government in Kabul has ever recognized the legitimacy of the Durand Line. Territories that register in the non-Afghan mind as being "Pakistani" may in fact belong to an inchoately imagined idea of Pashtunistan that still weight powerfully on the minds of many in the region.

The first draft of the point maps used to depict the Death List data in this exhibit, produced using ARCMap. Even the "decision" to use a point map raises just as many questions as it solves. Particularly when handling personal information like that in the Death Lists, all sorts of questions on what GIS programs call the style of points emerge. Should we depict deaths as circles? In what color? Should we use more personalized symbols - a human body, for instance? In cases like these, historians need to be both scarecrows and tin men.

Central Asian Approaches

Trying to expand my skills in several mapping applications, but also faced with the very real pressures of time and deadlines, I eventually settled on two programs to do the mapping work for Shattering Afghanistan, ARCMap and Google Fusion Tables. The former was used primarily for generating some of the animated maps that appear later in the exhibit, as well as most of the maps dealing with emigration from Afghanistan into Pakistan (in the next section of this exhibit); Fusion Tables was used to generate almost all of the "heat maps" you see in this exhibit. (While it is possible to generate heat maps in ARCMap, the relatively small data sample size and technical ignorance on the part of the author made Fusion Tables a better choice, at least for this first iteration of Shattering Afghanistan).

To depict the Death Lists, I settled on a combination of using point maps (depicting the number of people killed from a given location on a given day) produced in ARCMap and heat maps, produced in Fusion Tables; an earlier draft of the former kind of map can be found to the right. Why include both kinds of maps? To answer this question, it is important to understand the difference between the heat maps and the point maps. While both maps depict the density and/or quantity of data at a given geographical location, and while heat maps are understandable more vivid to the human eye, it also bears remembering that heat maps depict "heat" relative to the rest of the data presented on the map. Point maps, meanwhile, present absolute values that don't shift. Put in other, more concrete terms, this makes a difference when mapping data like the Death Lists, where so many people who were murdered came from Kabul. Drawing a heat map that deliberately exlcudes people from Kabul Province (containing the capital as well as outlying villages and towns) may give us a more precise picture of the topography of killing, as we see in the image on the right.

Sometimes, when working with heat maps, excluding the major "hot spots" can lend a more nuanced picture of data outside of the center, since heat maps depict relative rather than absolute frequencies. Here, in this map, produced using Google Fusion Tables, we can see that outside of Kabul most victims came from five places in the east and north of the country: Logar, Behsud (home to Shi'a Hazaras), the Panjshir Valley, Ghazni, and Baghlan, an industrial and cotton-picking center just north of the Hindu Kush.

Entering the Prison

Before we look into the maps of the Death Lists, however, it may be useful to learn something more about the context within which these documents were produced, and the specific conditions under which the innocent were murdered. That means investigating the history of Pul-i Charkhi Prison, the national prison in Afghanistan that was turned into a hellish warren of torture and arbitrary murder by the KhAD under Asadullah Sarwari and Asadullah Amin, the heads of the KhAD during the period in question.