instagram pinterest linkedin facebook twitter goodreads

Thoughts and Hallucinations

Me, a modeler?

I worked for a few days with a group of medical researchers in Texas around the question of how to improve primary health care teams. They’d asked me last year to help design an agent based model out of the intensive ethnographic-type work they’d done. I only helped them get started, but it led to a model and paper and article and proposal to extend the ideas. So they invited me back to lift the lid off the research and climb down into more of the details and fatten up the model with some reality. It was interesting and worthwhile work, though being a New Mexican now for some years I do see Texas in different ways. For instance, to pump up the emotions in the recent state election for Governor, the Democratic candidate called the Republican candidate, Susana Martinez, “Susana Tejana (Susan from Texas), since she was born there. Didn’t work, though, she won the election anyway.

It’s funny, though, my role in this health care project. I’m called a “modeler.” “Model” is one of those words that mean a lot of things. I built plastic airplane models when I was a kid. I modeled clothes for a charity event when I was a returning high-school exchange student, a pretty exotic status back in the dark ages in my home town. I’ve been told I had to be a model, like a role-model, as a college instructor. I learned how to put numerical data into a graphing calculator to see what kind of mathematical model came back out. But none of that describes the kind of modeler I was supposed to be in Texas.

This time I was an “agent based modeler.” This means, look at a system as the result, probably surprising, of a bunch of autonomous but interconnected system elements (agents) changing through time on the basis of their equally dynamic local information. So that kind of modeler meant, simplify a social world (a teaching hospital clinic) into one grouping of their agents (teams of physician, resident, interns, and medical students), and then figure out how those agent groupings can vary, and then look and see what kinds of overall patient outcomes the variations on the themes produce. It was good and interesting work with moral value with a group of dedicated people.

It made me think of why I wasn’t really a “modeler,” but rather an ethnographer who happened to know a computer language called Netlogo. I care about ethnography, not about modeling. At the end of an ethnographic bit of work, I wind up with some concluding arguments, like “people do such and such in situations and it produces a social world like this.” Now how do I generalize this? Actually the Language and Communication article downloadable to the left has something to do with this question. The Netlogo language is a formal language about agents interacting through time and producing systems not reducible to the characteristics of the agents themselves. It is, in social-theory-speak, a thought experiment laboratory for issues of structure and agency.

It does a few things. First of all, an effort to translate a concluding argument into Netlogo forces me to clarify implicit assumptions in the argument I want to make. Second, by its very nature the language is a generalization of whatever specific argument I’m translating. And that generalization enables explorations of the argument, by running it several times with the same initial conditions to see the variations it can produce, by changing its parameters or contingencies to see what if any difference it would make in how the argument turns out, by following a single agent on the ground to see the variety in a particular agent biography.

I’m not a modeler. I’m an ethnographic clarifier, generalizer, and player on the field of the rhetoric of alternative scientific arguments.
Be the first to comment