Thought experiment
A structured mental simulation to test ideas when real experiments are costly, slow, or impossible.
author
Various (Galileo, Einstein, Parfit)
Model type
About
Used across philosophy and science to probe mechanisms, expose contradictions, and locate decision boundaries. Good thought experiments isolate the essence of a claim and ask what must follow if it were true.
How it works
Frame the claim: define the proposition and the minimal assumptions.
Construct the scenario: strip to essentials; push to an extreme or edge case.
Derive predictions: what must happen if the claim holds?
Check for contradiction: clash with facts, logic, or comparable cases.
Refine: adjust assumptions or abandon the claim.
Construct the scenario: strip to essentials; push to an extreme or edge case.
Derive predictions: what must happen if the claim holds?
Check for contradiction: clash with facts, logic, or comparable cases.
Refine: adjust assumptions or abandon the claim.
Common types
Reductio – drive an assumption to absurd consequences.
Edge-case – test behaviour at limits (zero cost, infinite demand).
Principle test – hold everything constant to see if a rule still works.
Intuition pump – clarify values and trade-offs (e.g. trolley problems).
Edge-case – test behaviour at limits (zero cost, infinite demand).
Principle test – hold everything constant to see if a rule still works.
Intuition pump – clarify values and trade-offs (e.g. trolley problems).
Use cases
Strategy – “What if a rival prices at marginal cost?” “What if switching costs drop to zero?”
Product – design for the user with none of today’s crutches; eliminate steps.
Risk – pre-mortems and kill-switch criteria before launch.
Policy/ethics – reason about rules under clear, contrasting cases.
Product – design for the user with none of today’s crutches; eliminate steps.
Risk – pre-mortems and kill-switch criteria before launch.
Policy/ethics – reason about rules under clear, contrasting cases.
How to apply
State the hypothesis in one sentence.
List the fewest assumptions needed to run it.
Pick an extreme that isolates the mechanism (zero, infinity, perfect info, no frictions).
Work the consequences; write them down.
Decide: keep, refine, or reject the hypothesis.
List the fewest assumptions needed to run it.
Pick an extreme that isolates the mechanism (zero, infinity, perfect info, no frictions).
Work the consequences; write them down.
Decide: keep, refine, or reject the hypothesis.
pitfalls and cautions
Loaded premises – smuggling your conclusion into the setup.
Fanciful worlds – scenarios too detached from constraints to inform action.
Over-indexing on intuition – treat results as prompts for testing, not proof.
Fanciful worlds – scenarios too detached from constraints to inform action.
Over-indexing on intuition – treat results as prompts for testing, not proof.