Big Data predicts Riots in 2020

Remember Psychohistory — the fictional science in Asimov’s FOUNDATION saga, in which Hari Seldon used mathematical models to predict the overall behavior of large human populations?  Well, it’s a little bit real…

Have you ever noticed that the United States has significant outbreaks of violence every 50 years or so?  Remember the Civil War (~1870)?  The Depression (~1920)? Civil Rights protests (~1970)?  Here’s what the data shows:

(Credit: Peter Turchin –

Peter Turchin at the University of Connecticut has combined 3 key ingredients:

  • Ecosystem dynamics (the well-established science of modeling how biological populations interact with one another and their environment.  It’s what Fish & Wildlife Services use to manage wild herds of deer, for example, helping them calculate the number of deer licenses to sell to hunters each year).
  • Terabytes of new data about mundane historical facts — crop reports, birth announcements, death counts, etc. — a data set which continues to grow as archivists digitize more and more old newspapers and put them online.
  • Big-Data tools (the same kind of tools Google uses to know when you’re going to get a cold, so it knows when to show you NyQuil ads).

This combination is leading to a new field of study — what he calls “cliodynamics“(after Clio, the muse of history) — predicting the future with data from the past.  Sounds like Asimov to me!

HistoricalCyclesIn ecosystem dynamics, animal population levels and food levels dance around each other in a predictable boom-and-bust cycle.  The deer have plenty of grass, so their population grows, the population gets so big that it consumes grass faster than grass can grow, so the population level goes down due to starvation, and then the grass can grow all it wants, producing high food levels, and the cycle starts over.

Humans are certainly complex creatures, but we follow similar patterns.  When populations increase, competition over resources increases, and the results are often violent. Our social structures (and our ways of distributing resources) are more complex, and our resource pools are more complicated than a field of grass, but our population levels do have a predictable relationship with our resource levels.

It’s a fascinating attempt at looking at history from the perspective of math and the modeling of complex systems.  We’re scientists here, so let’s experiment.  Let’s see what the 2020’s looks like, and then we’ll meet back here to discuss.

And why do I care about this?

1.  Predicting the future is fun.

2.  Big Data is cool.

3.  The prophets of my spiritual tradition (specifically Jesus and Daniel) seem to find theological significance in the predictable rise and fall of civilizations.

So, what do you think?  Fascinating development, or just plain off the deep end?


Leave a Reply