Can we ever model sociology like we model climate?

Earlier today, in a feature on the science behind gun policies, I told you about how difficult it is to get reliable answers that pinpoint exactly what helps society and what hurts it. Models — computer algorithms that help us understand how complex systems work — play a role in this, but the ones used for gun research aren't very good yet. In fact, that's true about a lot of sociology fields, write the editors of the Get Stats blog. In general, our knowledge of how society works lags far behind our knowledge of the natural world. Can that ever be fixed? Some scientists think so.


  1. I think lack of understanding by laymen, and psychologists and social scientists, of the difference between the social sciences and the natural sciences is at the root of a lot of problems right now.  In short, most people who aren’t natural scientists tend to give results from the social sciences too much credit, and those from the natural sciences too little.  Psychology in particular is riddled to its core with human exceptionalism and Judeo-Christian/Cartesian assumptions about the nature of mind.  Economics is pretty much a system some people made up and then pretended was universal. All of the social sciences except anthropology, but especially psychology, tend to drastically underestimate the amount of our behavior that comes from culture.  

    Even a natural science like climate science, which uses models that aren’t entirely accurate, has well understood principles that can make very accurate predictions underlying anything they put in their models.  Eg.  it’s pretty easy to calculate the change in equilibrium temperature brought about by adding a certain amount of CO2 to the atmosphere.  That’s a trivial  problem in statistical physics.  Figuring out how long it will take is harder.  Figuring out the effects of all the other physical processes that will be set in motion by that change, which will either amplify it or somewhat negate it is much much harder, but we understand all of the physics behind each of those processes.  Figuring out the biological influences is much much harder than that, but even in that case, we have a far better descriptive and predictive understanding of the processes involved than we do about anything in any of the social sciences.   I’m not saying its impossible, but we still don’t have the level of understanding in any of the social sciences that would make this possible, and I think most of them will probably need to scrap their current paradigms before we get there.

  2. Wellll….since climate models are *never* accurate…sure, we can model social interactions just as well.   Put together a model that overpredicts and overestimates everything in society and BAM…climate model, meet sociology model…

    I’m writing this from my home at the bottom of the ocean, where my former beach-front house now resides after all the coastal flooding from the ice caps melting away….

    1. Climate models are not precise, because there’s a whole lot of factors that go into them. This means that there is a fairly wide, but known,  margin of error. There is no reason to assume that they are inaccurate, meaning that the actual results would fall outside of that margin.  I don’t think any of the models ever predicted that you would be underwater now.  Maybe in a few decades if you live in Bangladesh. The greenhouse effect itself is a trivial result of statistical physics, which we understand very well, have derived from first principles, and which provides extremely accurate predictions in an incredibly wide range of endeavors (eg, we wouldn’t be able to make the predictions necessary to figure out how to put billions of transistors on a tiny microchip if statistical physics were not incredibly accurate.)   The uncertainty mostly comes from secondary effects that propagate from it.

  3. Base ten is arbitrary. A bell curve literally wouldn’t look that shape to most animals. These don’t really mean anything, by themselves, but I’m sure the argument could be made for confirmation bias.    

    1. A bell curve (otherwise known as a gaussian) looks qualitatively the same no matter what base you plot it in.   Since we only really see them when we plot fairly abstract stuff, I don’t see how we’d know how it would look to another animal.

      1. The base thing had nothing to to with the curve, and you know for a fact the bell is SCALED to be visually mnemonic/shorthand for humans. Doesn’t matter whether an animal has two eyes or ten – the significance of a bell curve isn’t innate. I know exactly what you’re going to say – but MATH is. We’re not talking about math. We’re talking about people drawing conclusions about a complex system from a series of APPROXIMATIONS like bell curves.

  4. Society changes a lot faster than much of what we study in the natural world.
    Society is probably not very conducive to modelling because it’s predominantly based on learned behaviour which varies widely and is continuously altered.

  5. Girard’s model of competition, conflict and violence is one I found quite insightful.

    I also like Victor Turner’s model of liminal social drama.

Comments are closed.