The BBC is reporting the imminent extinction of religion. This is an end result to be hoped for, which just makes me all the more critical, and I have to say up front that this is the work of mathematicians, engineers, and physicists modeling sociology. It’s interesting stuff that looks at the very biggest picture without addressing the details, and it could very well be entirely true, but I’m always going to be a little bit suspicious of academics crossing boundaries that much. Sociologists are not stupid people; I’d like to see more of them pick up on this mode of analysis, and then I’ll trust it more.

You can read the paper for yourself, it’s available on arxiv, and it’s not a piece of crackpot pseudoscience; it analyzes gross historical trends away from religious belief in diverse regions around the world, and fits a reasonable curve to the pattern using an extremely simple model of group dynamics. The simplicity of the model is the troubling part — I’m a biologist, I don’t believe in simple any more — but the fact that the model works well for at least the selected regions is a little reassuring. Here’s the short summary of what they did:

Here we use a minimal model of competition for members between social groups to explain historical census data on the growth of religious non-affiliation in 85 regions around the world. According to the model, a single parameter quantifying the perceived utility of adhering to a religion determines whether the unaffiliated group will grow in a society. The model predicts that for societies in which the perceived utility of not adhering is greater than the utility of adhering, religion will be driven toward extinction.

The data look wonderfully clean, too.


(About that rescaled time axis: the data from different regions show different rates of the deconversion process, with timescales from decades to centuries; they all fit their model with different parameters for the perceived utility of religion. The rescaling shows that the model provides a good fit to all of the data, but you can’t use this to predict the date of the worldwide Atheist Rapture — it’ll happen at different times in different regions.)

The authors also express reasonable reservations. I was wondering about these questions, myself.

Our assumption that the perceived utility of a social group remains constant may be approximately true for long stretches of time, but there may also be abrupt changes in perceived utility, a possibility that is not included in the model. We speculate that for most of human history, the perceived utility of religion was high and of non-affiliation low. Religiously non-affiliated people persisted but in small numbers. With the birth of modern secular societies, the perceived utility of adherence to religion versus non-affiliation has changed significantly in numerous countries, such as those with census data shown in Fig. 1, and the United States, where
non-affiliation is growing rapidly.

That is a real concern. Their mathematical models are built around a parameter called perceived utility, ux, which they extract from the overall data — it’s not something that can be measured directly in individuals or populations, but is derived from historical trends and then used to calculate future trends, which is a little bit circular. I’d be more confident in their prediction if perceived utility had some independent measure that could be used in the curve fitting.

And of course, as they note, it’s not at all certain that that perceived utility will remain constant — it can’t have, for one thing, or the process of deconversion would have started a long time ago, we’d be further along the curve, and we’d all be atheists now. And unfortunately, the work doesn’t address the interesting question of what caused the historical shift in the perceived utility of religion, and without that, we can’t know what kind of factors might cause it to shift back.

I’ll still hope the math is a good predictor of the fate of faith.