What development research can learn from Asimov’s psychohistory

[Original post from 19/3/2914. Reposted for the fun factor]

It has been more than a year and a half since I received my Ph.D. after writing and defending a lengthy dissertation that I liked to think of as “policy-relevant social science”. Thirteen months into my current job, researching and networking with the same aid organizations and actors that populated my dissertation, I have come to realize that social science and development policy are two entirely different beasts, and that reconciling them in any meaningful way is a challenge far beyond the skills of even the most imaginative Ph.D. candidate.

Two things drew me to study the political economy of development: first, a genuine – if naive – belief that development problems could be solved if only we thought harder about them; second, an enthusiastic – if misguided – fascination with the possibility of using models of social behavior as tools for changing reality. For my naivete I have no-one else to blame; for my misguided scientism, however, I can point to a clear culprit: Isaac Asimov.

Isaac Asimov (1920-1992), writer and science communicator extraordinaire.

The Foundation Series of novels by Isaac Asimov is easily one of the chief achievements of science fiction in the last 100 years. Asimov is not the most evocative writer, nor the most imaginative builder of worlds. But he had a good eye for politics, and a passion for reason as a force for good. That is why he asked, in his parable about the fall of the Galactic Empire, whether science could help limit chaos and anarchy through its influence over policy. His hero, Hari Seldon, is a mathematician who conceives an entirely new branch of science, “psychohistory“, first as a purely theoretical postulate and then as an empirically-based science of large-scale social prediction. Through the use of psychohistory, we are told in Foundation (1951), Hari Seldon foresees the imminent collapse of the Galactic Empire, determining that the ensuing dark ages will last 30,000 years. By creating a haven for reason, a scientific “Foundation”, Seldon calculates that he can shorten the descent into barbarism down to 1,000 years.

Most social scientists who emphasize the “science” part of the expression are following in Hari Seldon’s footsteps. Economists in particular like to fantasize that their models can predict the future, even if real life keeps disproving them with impolitic frequency. Many political scientists and development researchers have jumped on this bandwagon. Bruce Bueno de Mesquita has developed computer models that can “predict” policy choices by studying the preferences and relative power of key decision-makers. Researchers at the MIT Poverty Action Lab believe that randomized control trials (RCTs) hold the key to effective development policy by identifying and modelling why poor people behave the way they do. And “the science of delivery” has become the World Bank‘s mantra under its new president. “Predicting” as the chief object of science is generating methodological hierarchies everywhere, from traditionally qualitative political science departments like Cornell to traditionally innovative and risk-accepting development agencies like DFID.

If you read the novel Foundation and stop there, then pychohistory looks indeed like the ultimate social science, the remedy to all political and social maladies and definitive proof that scientists can improve policy. If you read the following novels in the series, however, things get murkier (caution: spoilers ahead).

An old white man telling everyone what to do. Science fiction indeed…

Foundation and Empire (1952), the second novel in the series, posits the fundamental challenge of predictive social science: What happens to the model when an unpredictable variable appears? In the novel, the birth of a single man with psychic abilities – something that Hari Seldon never foresaw – threatens to derail the entire plan to avoid 30,000 years of barbarism. What is more worrying, to the bitter end the citizens of the Foundation hold on to the desperate belief that their wise founder must have predicted this, and therefore that he must have accounted for it in his plans. But he didn’t. And eventually the mutant takes over the entire Foundation.

The rise of “the Mule”, as Seldon’s psychic antagonist is known, is a valuable cautionary tale for researchers and development experts enamored with the promise of social science as a force for good policy: it tells us that the best model is only good as long as the assumptions that sustain it remain true. Among the military – an expert community long committed to counting things and planning the future – there is a well-known aphorism for this problem: a military plan rarely survives contact with the enemy. Carl von Calusewitz called it “friction“: the messy and fundamentally unpredictable dynamics of war due to human ingenuity, emotion, and unintended consequences. Development studies does not have a similar term, even if we constantly face a similar problem: “friction” seems as good a concept as any for the tenacious frequency with which developing country politics derails development interventions.

Beyond a skepticism of social science as a policy guide, and about people’s ability to recognize its limitations, the Foundation Series has one more lesson to teach us.

What does it take for psychohistory to work

The appearence of an unpredictable factor, and the total derailment of Hari Seldon’s plans, forced a few smart citizens of the Foundation to wonder whether a “Second Foundation” had existed all along together with the first one: an organization “at the other end of the galaxy”. It eventually turns out that such “end” was not a physical or geographical one, but actually an epistemological one: the Second Foundation was built not on science and openness, but on mind control and secrecy, tasked with the crucial mission of priming and nudging the minds of key individuals to ensure that the Seldon plan actually comes true. Starting with Second Foundation (1953), the Foundation Series becomes an exploration of the idea of control – subtle, constant, and above all secret – as the only way to ensure that humanity will not descend into barbarism.

This is a powerful corollary to the cautionary tale of psychohistory: predictive models can only be fully realized through absolute control over its subjects, an idea that resonates with Karl Popper‘s critique of “historicism” as the use of scientific prediction for public policy, which in his mind could only lead to totalitarianism. We need not take this corollary at face value, given that – for all we know – mind control does not exist and we are not mere puppets in a Seldon-like master plan. But at a more abstract level the Foundation Series posits the idea that the best science of human affairs can only guide actual policy through the purposeful and adaptive agency of policy-makers.

Development policy, from this perspective, cannot be reduced solely to a methodology for predicting results, whether through computer models or RCTs. It will always be subject to stochastic factors, and it will always be filtered through the agency of people at all levels. Given that it will be up to policy-makers to grapple with uncertainty and manage complex interactions, development may turn out to be more art than science.

That is the lesson that I failed to learn during my Ph.D., when I operated under the assumption – reinforced by graduate school and the professional norms of political science – that we could model reality in a meaningful way. It is dealing with policy-makers and analyzing policy issues that I have finally come to understand that the true divide between academics and policy-makers arises from a fundamental misconception: they keep asking for von Clausewitz, and we keep giving them Hari Seldon.

[Originally posted on 21 October 2013]