Problem This Solves
History is full of recurring patterns -- crises, collapses, civil wars -- yet neither specialist historians (too narrow to separate general principles from specific circumstances) nor amateur "cyclic history" theorists (who cherry-pick and force data into fixed cycles) can provide the understanding needed to navigate current crises. "Ninety-nine percent of 'cyclic history' suffers from one or both of these problems." Cliodynamics fills this gap by treating history as a science: translating theoretical ideas into explicit dynamical models whose predictions are tested against data.
Key Principle
Cliodynamics rests on several methodological pillars:
- Seshat Global History Databank: Launched in 2011, this collaborative database systematically converts historians' knowledge into analyzable data using a three-tier collection process (trained research assistants, PhD-level supervisors, expert scholars). It covers roughly 300 crisis cases from the Neolithic to the present across all major continents.
- CrisisDB: An offshoot of Seshat focused specifically on societal crises and their outcomes -- not just collapses but also successful exits. About 100 of 300 cases are fully coded, enough to discern main patterns.
- Lanchester's Square Law: Mathematical proof that quantitative models work for history. Numerical advantage translates to a combat advantage equal to the square of the ratio -- a 4:1 population advantage becomes a 16:1 warfare advantage. This explained the North's inevitable Civil War victory despite Southern qualitative advantages.
- Factor X (Tolstoy's "Spirit of an Army"): Even "squishy" variables like morale can be quantified through systematic empirical analysis. Trevor Dupuy operationalized this by analyzing 81 WWII engagements, finding German combat efficiency was 1.45x that of British forces.
- The Structural-Dynamic Approach: Identify interest groups in a society, assess their relative power and organization, and track how these change over time. Societies are "groups of groups of groups." Build mathematical models, test predictions against data.
- Materialist Assumption: Groups pursue the material interests of their members. Those claiming prosocial behavior "need to go an extra mile to show that they are not feeding us a load of bullshit." When inside information is unavailable, deduce agenda from consequences of actions.
- Cliodynamics as Intervention Design: "Prediction is overrated. What we really should be striving for, with our social science, is ability to bring about desirable outcomes and to avoid unwanted outcomes."
Good Examples
- Lanchester model of the Civil War: The North had a 4x population advantage (22M vs 5.5M white Southerners) and a 32:1 advantage in rifle production. Even granting the South a generous 4x skill/morale advantage, the North's squared numerical advantage (16x) was overwhelming. The model correctly predicted the grinding four-year outcome.
- The 2010 Nature prediction: Turchin published in 2010 (blogged September 3, 2012): "I feel quite safe making the prediction that there will be a peak of political violence in 2020 (plus/minus a few years)." He also made the metaprediction that nobody would listen. Both proved correct.
- Seshat state formation model: A relatively simple model of Old World state formation (1500 BC - 1500 AD) did "a remarkable job of predicting where and when 'macrostates' formed and how they spread."
- Seshat multi-proxy methodology: Using pollen cores, dendrochronology, potsherds, skeletal height, skeletal trauma markers, and parish records to reconstruct historical dynamics. Different proxies have different biases that can be cross-checked.
Bad Examples
- Cherry-picking: Selecting only historical examples that fit a pet theory while ignoring counter-examples.
- The Bed of Procrustes: Stretching or cutting historical examples to force them into fixed cycles. Turchin avoids the word "cycle" in professional articles because of negative baggage, using "oscillations" or "boom-bust dynamics" instead.
- Amateur cycle theory: Grand historical narratives built on convenient examples without systematic sampling or falsifiable predictions.
- Theory-free big data: The structural-dynamic approach requires theoretical models, not just data accumulation. Individual historian narratives are too narrow; aggregation through formal models is necessary.
Key Quotes
"Making precise predictions about events in human societies decades or centuries in the future is pure science fiction." (Chapter A1) Cliodynamics accepts chaos but finds that trajectories still trace recognizable attractors.
"Social breakdown and internal warfare kill people, wreck economies, and roll back human achievement. We must develop a clear-eyed understanding of why it happens so that we can avoid the endless cycle of recurrent waves of instability and violence." (Chapter A3) The moral imperative behind cliodynamics as a discipline.
"Prediction is overrated. What we really should be striving for, with our social science, is ability to bring about desirable outcomes and to avoid unwanted outcomes." (Chapter A1) Cliodynamics is intervention design, not fortune-telling.
"The mind of another is an enigma wrapped in darkness." (Chapter A3) Why cliodynamics analyzes group-level material interests rather than individual motivations.
Rules of Thumb
- Model complexity: "Start with the simplest possible design and then add 'stuff' to it. It's like cooking a soup." Once you reach the right level of complexity, adding more makes the model worse.
- Default to material interests: Assume a group pursues material self-interest. Require evidence for claims of prosocial or propaganda-driven behavior.
- Organization beats individual power: "A disciplined, well-structured army will always defeat an unorganized mob of individually powerful warriors." When assessing group power, weigh cohesion and organizational structure alongside individual member attributes.
- Multi-proxy cross-checking: Use multiple independent data sources because each proxy has its own biases.
- Chaos means agency, not impossibility: Individual free will can have macro-level consequences -- "we are all 'Mules'" -- but structural dynamics still trace recognizable patterns. Cooperation with others is the realistic path to positive results.
- Separate general from specific: To learn from history, distinguish principles applicable across societies from circumstances peculiar to one case. Don't mechanically transfer lessons without this separation.
Related References
- The Structural-Demographic Theory of Political Disintegration - The theory these methods support
- Historical Cycles and Patterns - The patterns discovered through this methodology
- Multipath Forecasting and Future Scenarios - Applied cliodynamics for future scenarios