The year of 2012 has been a good one for Social Evolution and Cliodynamics. We had a wonderful meeting in Frankfurt that celebrated the new stature of Cultural Evolution. The Social Evolution Forum has been developing very nicely (in the next blog I will provide more information). And the science of Cliodynamics has been adding both breadth and depth. We have just published a new issue of Cliodynamics: The Journal of Theoretical and Mathematical History:
http://escholarship.org/uc/irows_cliodynamics
This blog takes off from the Editor’s Column with which the latest issue of Cliodynamics starts, but then I go in directions that would not be quite as appropriate for an academic publication. Note that when I refer to the journal, I use italics (Cliodynamics) and when to science just plain font.
Although the subtitle of the journal emphasizes theoretical and mathematical approaches to history, the overall goal of Cliodynamics is to contribute to developing a mature theory, theory in the broad sense, which integrates conceptual elements with mathematical models and, very importantly, data. It would be ideal to publish articles that do both models and empirical tests, but in practice it is difficult to do both well in the framework of a single article.
A division between modeling and empirically-oriented articles is probably inevitable. For example, in physics, which is often thought as the Queen of Sciences, there is a very clear differentiation. Some specialize as theoreticians and others as experimentalists. The two sub-professions can and do talk to each other. In fact, my impression is that there is more interchange between mathematical and experimental physicists, than between mathematical and empirical economists.
In general, physics’ reputation as a mature science is well deserved. But don’t accuse me of physics envy! If I wanted, I could always become a physicist, but I find the challenge of historical and social sciences much more – well, challenging. Basically, physics succeeded because physicists craftily chose very easy problems to solve. What could be more simple than a solar system? As a planet serenely sails through vacuum, the problem of describing and predicting its motion can be reduced to a couple of equations.
This is easy, compared to much messier things, such as an ecosystem, or even worse, a social system. It’s amazing that history is not “just one damn thing after another,” that there are actually general principles governing the dynamics of human societies.
But to get back to the current issue of Cliodynamics. It illustrates the theoretical/empirical divide quite nicely, because the first two articles focus on models, and the next two articles on data. Bret Beheim and Ryan Baldini develop a rigorous quantitative framework for studying cultural evolution, while Radek Szulga builds model of pre-Neolithic economy. Although both papers are primarily theoretical, this is not sterile, abstruse theorizing.
I remember that when I began to be interested in mathematical approaches to history, I was quite amazed to discover that there is a number of journals devoted to mathematical economics. I eagerly started to read articles in them, but I was severely disappointed. Maybe my impression was too hasty, but here it is: the field of mathematical economics is largely a ‘Glass Bead Game‘, in which its practitioners write papers having nothing to do with reality and for consumption only by a very small group of initiates. If this is an unfair assessment, I would be happy to be rebutted.
In the journal, in contrast, we make a great effort to enforce a connection to reality (on the part of mathematical papers), as well as aconnection to theory (on the part of empirical papers). While it may be too much to get articles that do both math and data well, we can certainly stay away from both mindless empiricism and sterile theorizing.
You may say, very well, physicists can do experiments, but experiments are impossible in historical sciences. True, manipulative experiments to test theories of history are impractical, or unethical, or both. There are exceptions, but I will leave this topic for future. Largely, this is true.
However, notice ‘manipulative.’ There are other kinds of experiments that are quite possible in history (and other historical sciences: astrophysics, geology, evolutionary biology, and historical linguistics). The core of scientific method is to butt two (or more) rival theories against each other, with data as the ultimate judge. There is no requirement to set up a pure experiment for this to work (although if it’s possible to do, so much the better). It can be done in our messy world.
There is one important point, however: do we use known or (as yet) unknown data to test predictions of theories? Logically, it shouldn’t matter, but we all know that if the data are already known they will ‘contaminate’ theory development. In other words, theories that we develop are already influenced by known data, so ‘testing’ them on those data suffers from the problem of circularity. Not a good approach.
So what we need to do in order to really test theories is to contrast their predictions with data that were not known when theories were constructed. That’s really why we value predicting the future so much, because future is unknown – and perhaps unknowable. So if a theory makes reasonably good predictions about the future, that’s a very strong test, and correspondingly our confidence that the theory is correct, or at least managed to capture some slice of reality correctly, is correspondingly enhanced.
If we deal with long-term processes, then we have to wait for a long time for that future to arrive and see whether predictions of the theory are supported or not. That’s not a recipe for rapid-discovery science.
Fortunately, waiting for the future is not the only way to test theories with unknown data. My favorite dream is to get two theories to make predictions about data that are buried several feet below the surface. Then we recruit archaeologists, and tell them to dig here. Once they get to the cultural layer in question, we find out which theory is right.
Perhaps we will be able to do this, but for now, I am putting my efforts into creating a historical database. In many ways, historical data are as unknown as archaeological data that hasn’t yet been unearthed. Yes, these data are known – to narrow specialists of the region and period in question. But it is dispersed across many peoples brains, and a specialist in medieval Burma, for example, has no idea about the Songhai Empire.
We can ‘unearth’ such data by getting specialists from Burma and Sahel to agree to answer our questions, so that their knowledge can be codified in a massive historical database of cultural evolution. And then we can mine the database for answers.
And we can use such a database to test theories, especially those that have been developed prior to the database construction. So putting together databases should be one of our priorities.
The current issue of Cliodynamics introduces a new section, Databases. One article, on which I am the first author, describes the on-going project, which I have already mentioned in a previous blog. Our paper presents no analysis and no results, and most academic journals would not even consider it for publication. But I feel it is critical to publish such papers before there are any results and conclusions, because how will you, otherwise, get out the predictions in print?
Although prediction (not just about the future, but in the sense I just described) is at the core of doing science, surprisingly few scientists ever publish predictions ahead of time. But it’s actually a good thing to do. I’ve done it before in my previous life as an ecologist (and it resulted in a publication in Science), and in our paper in this issue of Cliodynamics we did it again.
And, unlike in a casino, when playing at the game of scientific prediction you can set it up as a win-win situation (well, in the casino it is the owners who always win). For example, in our paper we give an example of a specific test of two hypotheses against each other. Once we have the data and analyze them, either one or the other hypothesis wins. In either case we have made an advance in our understanding. Just about the only way we can lose is if we are unable to recruit enough specialist historians to help us build the database. But judging from the initial response by historians, it doesn’t seem likely. It will just need a lot of work. Nothing worthwhile happens without getting the proverbial sweat on one’s brow.