Artifacts and Applications: Computational Thinking for Archaeologists

Andrea M. Berlin

(See email contacts page for the author's email address.)

I've been an archaeologist for over 30 years, and for all of that time I've resided comfortably on the far sidelines of technological approaches. In the past several months, however, I've accidentally backed into a parallel universe of deep data and the people who are fascinated by its possibilities and challenges — in other words the world of computer science — and it's changing almost everything that I think about archaeological information and publication. Here's where I suddenly find myself and how I got here.

I am now working on the further development of a website and web application that will allow scholars to easily gather detailed archaeological data and recombine it — by time period, by region, by type of political system, etc. — to address whatever issues they are interested in investigating. The eventual application would, for example, enable a political scientist to compare the patterns, intensity, and direction of trade under earlier political regimes as revealed by archaeological evidence and so gain the hard data that can lead to insight into the relationship between economies and various governmental systems. The eventual finished site will function as both an archive and a tool, enabling archaeologists and other scholars to store data but also, and even more importantly, to submit, edit, search, and compare data — of all types and periods.

This project is an outgrowth of an initiative that I began last year, to bring together researchers who work on pottery in the Levant. I myself am a specialist in pottery of the Persian, Hellenistic, and Roman eras, but I had become increasingly frustrated by the difficulties in correlating various sorts of analyses, such as petrographic or chemical, with masses of excavation pottery. I decided to begin a series of workshops for ceramicists, archaeometrists, and geologists in which we could begin speaking directly to each other to identify areas of consensus as well as subjects in need of targeted study.

I knew that workshops alone wouldn't be sufficient. I also wanted to build a website to collect the data we'd be bringing to the table. So I, along with a former software developer who is a student here at BU, built a prototype, named the Levantine Ware Encyclopedia. Almost as soon as we got it up and running we realized two things. One, even as an awkward and clumsy prototype, it was incredibly useful. Two, it was so terrible that we were going to need significant help to make it better. Hence a grant application to Boston University's Rafik B. Hariri Institute for Computing and Computational Science and Engineering.

My initial approach to the Institute put me in front of Win Treese, the Hariri's Associate Director. Win was excited about this project from the get-go, because he saw that archaeologists were sitting on an intellectual gold mine of data, deep complex inter-related data, but were unaware that techniques existed to mine it. Data mining (an actual term that computer scientists regularly invoke) is a type of computational thinking — and, I quickly learned, a different and powerful way to think about archaeological material.

I will say outright that before my introduction to the Hariri Institute, I could not have explained what computational thinking was. Had I been asked to define it, I would have said that it was just a fancy term for working with digital data, a far more efficient way to sort or search for information than my traditional manual approach of paging through books and articles. But computational thinking is not about speed (though of course speed is a component) so much as approach. It is a way of thinking that is both smaller and larger. Thinking smaller means seeing a piece of data as comprised of many specific individual attributes. Thinking larger means envisioning how different users might deploy selections of those attributes to ask and perhaps answer various other questions. Computational scientists can create applications that unbundle attributes in an organized way so as to make the information implied by those attributes more useful.

Even the best of the existing digital archaeology sites, such as that from the Athenian Agora (www.agathe.gr/) or Potsherd (potsherd.net/atlas/potsherd.html), Paul Tyers' elegant site for Roman pottery from Britain and western Europe, can't do that. They suffer from most of the drawbacks of the off-line world they seek to supplement. All use arcane language, are largely static in approach, and are rigorously oriented to fellow specialists. If a political scientist came to either of these sites with a query about the material correlates of different political systems, for example, she or he would simply turn away. Nor would many archaeologists tackle such a question, because our field's data deluge and attendant jargon is such that we tend to stay safely sequestered in specialist silos.

Let's return to that imagined political scientist for an example of how this new site would work. S/he might want to consider the relationship between political change and economic activity, a standard category of archaeological inquiry. On this site, that political scientist (or any archaeologist, for that matter) could compile material evidence for local production and long-distance trade at the end of the Late Bronze Age, in the first century BCE, and in the early medieval Mameluk Sultanate – all periods of political upheaval. All data on the website will be linked to publications and also include the name and contact information of the person who provided it, thus making it simple for researchers with similar interests to get in touch with one another.

Here's another example. An archaeologist excavates a new site, finds some unfamiliar pottery, has a petrographer do a few thin sections, and receives a report identifying key mineral components. On this website, that archaeologist could type in the names of those components and find other ceramic wares with similar profiles, as well as photographs, drawings, and information about date, location, shapes represented, etc.

I used to think of archaeological data as a great mass comprised of many discrete, separate items — whole things. Learning about computational thinking has allowed me to see those whole things as containing many individual pieces of information — parts inside the whole. With that I can now imagine a system that will make the parts as available and useful as the whole, allowing us to ask and answer new ranges of questions. It's like physics: busting the atom open in order to see and make use of the forces inside. Professor Ed Lazowksa, of the University of Washington, describes computational science as a field that is transforming observational disciplines into analytic ones. He doesn't mean that people weren't analyzing data prior to using computers, but rather that computational science provides a fundamentally new mode for acquiring, arranging, and manipulating masses of data. Computational thinking is to earlier forms of analytical thinking as hyper-drive is to a car with an automatic transmission. It is transformational. Everybody who works with data needs to understand how it works and what it can do.

-- Andrea M. Berlin