By Steven Edwards

The ASOR Annual Meeting is one of the highlights of my academic year. Every November, I travel to a different city for the chance to catch up with friends and colleagues, mingle at a few social receptions, and attend papers. Lots of papers. What does this tell us about intellectual and social trends in ancient Near Eastern studies?

In fact, over the last ten years, the total number of papers at the Annual Meeting has increased by almost 250 percent. I know this because over the past year I’ve been poring through the program books from the last ten Annual Meetings looking for patterns in the way that conference participants interact.

Right from the start, one thing was clear: the Annual Meeting has gotten big! There are more papers being read at the Annual Meeting than ever before. There are more authors of those papers too. A lot more. From the 2009 Annual Meeting in New Orleans to the 2017 conference last year in Boston, the total number of authors has increased by almost 300 percent. But changes in co-authorship are interesting because they also reflect the way that research is being conducted by the ASOR community. The data show that collaboration is increasing at a rapid rate.

Co-authorship is also interesting because it stands as a proxy for interaction. By looking at who is authoring papers with whom, I can reconstruct a hypothetical social network for the ASOR Annual Meeting community. And that is exactly what I did. The results of this effort were presented at the Annual Meeting in Denver in November 2018. The main takeaway: a few key people are co-authoring a lot of papers with a lot of different people, and this means that there are a few really well-connected scholars in the ASOR community.

Inspiration for my study actually came from the field of mathematics, and particularly the Hungarian mathematician Paul Erdős.

Paul Erdős. (https://upload.wikimedia.org/wikipedia/commons/2/2f/Erdos_budapest_fall_1992_%28cropped%29.jpg)

Erdős was arguably the most prolific mathematician of the 20th century, having published more than 1,500 peer-reviewed articles over his long career. Interestingly, in producing these papers, Erdős collaborated with more than 500 unique authors. This incredible output put Erdős at the center of a very large, worldwide network of mathematics researchers. His legacy has largely been cemented by penchant for collaboration, such that scholars actively seeks out ways to measure just how close they are to Erdős, at least in terms of co-authorship. Erdős numbers are a measure of collaborative distance between Erdős himself and his many co-authors, such that those who personally co-authored papers with Erdős have an Erdős number of one, while those who only co-authored papers with Erdős’ co-authors have an Erdős number of two, and so on.

Armed with this concept of collaborative distance, I sought to determine if there was a Paul Erdős-like figure in our network of researchers at the ASOR Annual Meetings. Looking at both co-authorship and participation in academic sessions, I generated a social network of the ASOR community and then ran a series of statistical tests to determine who might be considered the “best connected” person at the conference. The answer did not seem to surprise anyone: Thomas Levy, a Near Eastern archaeologist.

Co-authorship network of the Annual Meeting of the American Schools of Oriental Research. This network graph contains 699 nodes, representing scholars who have co-authored a paper at the conference at least once from 2009-2018. Red nodes correspond to men, while blue nodes correspond to women. Node size is proportional to betweenness centrality, one of the metrics used to identify the most important people in the network. (Courtesy S. Edwards, 2018)

Branch of the co-authorship network occupied by Thomas Levy and his UCSD “cluster”. Levy links a group of researchers working predominantly as part of the Madaba Plains Project (labelled as MPP) to the main body of the co-authorship network. (Courtesy S. Edwards, 2018)

Tom Levy. (http://anthro.ucsd.edu/people/faculty/faculty-profiles/thomas-levy.html)

Levy heads an active lab of researchers at the University of California, San Diego, and they have been well-represented at the Annual Meetings over the last decade. In fact, Levy is listed as an author or co-author more than any other ASOR member during this time period. It is perhaps not surprising, then, that he comes out on top as the “best connected” person at the conference. But there’s a little more to it than that.

Measuring how connected a person is within a social network can be challenging. What does it mean to be connected anyway? It might seem reasonable to assume that the person who co-authors the most papers would be the best-connected node in the network, but that’s not necessarily the case. Think about it this way: If Person A co-authors ten papers with the same two people over a ten-year span, they are really only connected to those two other people. On the other hand, if Person B gives just five papers over that same span, but with five different co-authors they would still have more than twice as many unique connections as Person A.

That a person who co-authors with a lot of different people would be well-connected is intuitive. Unfortunately, it gets even more complicated than this. What about the connections of one’s co-authors? If Person A’s two co-authors also happened to co-author separate papers with a large number of people, those connections would only be one step removed from Person A. This should be accounted for when calculating one’s connectedness—or, to put it in network terms, one’s centrality. Centrality is a big topic in network analysis, and the various approaches to calculating centrality are as numerous as they are complicated.

As a non-mathematician, I opted for a classic method for determining centrality called betweenness. Betweenness is a measure of how frequently a given nodes lies on the shortest path between every pair of nodes in a network. It is not a perfect method for measuring connectedness, but it is more robust than simplistically counting how many links a node has. When crunching the numbers in the co-authorship and session participation networks, I came upon a pretty clear result: Thomas Levy has the highest betweenness centrality not only because he co-authors a lot of papers, but because his co-authors also tend to co-author a lot of papers with different people. Looking at the network graph, this seems almost too obvious. Levy acts as a bridge between various parts of the network. This privileged position means that Levy is well-placed to connect disparate clusters of researchers.

Apart from Levy and his “UCSD cluster,” there were some other surprising names on the list of “best connected” people at the Annual Meeting. For example, ranking as the third most connected person in the co-authorship network, according to their betweenness centrality, is Deirdre Fulton, professor of Bible at Baylor University. Though Fulton has co-authored fewer papers than many scholars lower down on the list, she nevertheless serves as a hub between clusters of researchers. I think one of the reasons behind Fulton’s high score has something to do with her being a zooarchaeology specialist as well as a Biblical scholar. Having a specialized skillset means that Fulton is more likely to collaborate with a wider range of people working on different periods and in different places. By co-authoring papers with these different groups of people, Fulton acts as a bridge between clusters of researchers that would otherwise be disconnected.

Such connectedness is not trivial: information, ideas, innovations and other resources tend to flow through hubs like Levy and Fulton. But how did this network structure develop in the first place? Levy suggests that his early adoption of the “science model” of publication might have played a factor in his high betweenness ranking. In this model, publications often list large numbers of co-authors, including those that may only have provided minor contributions. Whether or not one accepts this model of publication, it has led at least in part to Levy developing a wide neighborhood of connections to other researchers at the ASOR Annual Meetings, linked archaeologists working in many periods to text specialists, including Biblical scholars, and scientists in still further fields, including remote sensing and archaeometallurgy.

While Paul Erdős remains unchallenged as the most prolific mathematician, perhaps we can begin to refer to Levy numbers in the field of Near Eastern archaeology to measure our own collaborative distances from similarly influential scholars.

Future work will continue to examine this social network in an effort to determine how ideas and theories propagate through the ASOR community. The starting point has been to determine the general structure of the Annual Meetings through the lens of co-authorship and session participation, and to identify the key players in this network. The next steps will be to investigate the influence of these key players on the ASOR social network across a range of domains. Does being linked to a well-connected scholar directly impact one’s ability to publish in top tier journals, acquire grants, or get a tenure-track job? This is the question at the heart of my research into the structure of the ASOR community. I hope to present an answer to this question at this year’s ASOR Annual Meeting in San Diego, in November.

Steven Edwards completed his Ph.D. at the University of Toronto.