Popper's Removability Test
Judge a system by how cheaply it removes bad rulers and policies without violence.
- Difficulty
- Easy
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 68%
Plato framed democracy as a question about who should rule — philosopher kings or the citizens — and worried that a mob would vote away a minority's rights. Popper rejected the question. Democracy has nothing to do with who rules; it is the system that lets you remove policies and rulers most efficiently, without violence. That reframing makes democratic systems comparable: France, England, the United States and Canada can be ranked by how quickly, cheaply and peacefully they eject the people and policies that are not working. It is the same move as in science. No method of science delivers the true theory; science is an error-correcting mechanism whose job is getting rid of bad ideas. Applied to any institution, the test becomes an audit: trace the removal path for a real decision, count what it costs, and shorten it.
Origin
Karl Popper's reframing of democracy, contrasted with Plato's who-should-rule question and explicitly linked to science as error correction in this episode of Naval.
Core principles
- 01The question is not who should rule but how bad rulers are removed.
- 02No selection method reliably produces good rulers, just as no method produces true theories.
- 03A system is good to the degree it deletes mistakes quickly, cheaply and without force.
- 04Removability is measurable; wisdom of leadership is not.
- 05Systems that make reversal expensive stop correcting and start ossifying.
How to run it
- 1
Drop the who-should-rule question
Stop trying to design a selection process that guarantees good leaders or good policies. No such process exists, in governance or in science.
Pro tip Notice how much of your governance debate is really about selection — that is the budget you can redeploy to removal.
- 2
Pick a live decision and trace its removal path
Take a policy or an appointment that currently exists and write out every step required to reverse it: who must agree, what evidence is needed, what the sequence is.
Pro tip Use a decision people already privately think is wrong — the path will be honest rather than theoretical.
Watch out Do not audit a hypothetical policy; hypotheticals always look easy to reverse.
- 3
Measure time, steps and cost
Put numbers on the removal path: elapsed time, number of approvals, political or financial cost to the person who initiates it. These numbers are the score.
Pro tip Track who personally pays the cost of initiating a reversal — if it is one individual, removals will not happen.
- 4
Check for force, exit and schism
Ask what actually happens when removal is blocked. If the answer is that people leave, fork, or fight, the system is not error-correcting, it is deferring the correction.
Pro tip Forks and resignations are removal attempts routed around a broken path — read them as data.
Watch out A system with no visible conflict may simply have made dissent too expensive to express.
- 5
Rank alternatives on removal cost
Compare candidate designs on how efficiently each ejects mistakes rather than on how likely each is to pick well in the first place.
Pro tip Score two designs side by side on the same real decision to keep the comparison concrete.
- 6
Shorten the most expensive reversal
Redesign the single removal path that costs the most — a sunset clause, a named owner with authority to revoke, a scheduled review with a default-off.
Pro tip Attaching an expiry date to a new policy converts removal from an action someone must fight for into the default.
Watch out Making everything trivially reversible has its own cost; target the paths that are blocking correction, not all of them.
In the wild
Plato asked who should rule and concluded a mob would vote away a minority's rights. Popper asked instead how a country gets rid of the people it does not want. That converts a philosophical argument into a comparison: France, England, the United States and Canada are all called democracies, but they differ in how quickly, efficiently and peacefully they remove rulers and policies. On that axis they can genuinely be judged better or worse — which is impossible if the criterion is whether the system produces wise leaders.
→ An unanswerable question about ideal rulers becomes a measurable comparison between real systems.
A company adds a spending approval step after one bad purchase. Two years later everyone agrees it slows the team down, but nobody removes it: reversal requires the finance lead to admit the control was excessive, plus sign-off from two executives who did not introduce it. The removal path is traced, timed at roughly four months of effort with reputational cost falling on one person, and redesigned — every new control now ships with a six-month expiry and must be actively renewed.
→ Reversal becomes the default rather than a fight, and dead controls expire without anyone having to lose an argument.
Common mistakes
Optimising selection instead of removal
Endless effort goes into better hiring, better voting and better appointment processes, none of which fix the systems where a wrong choice cannot be undone.
Reading calm as health
A system with no removals may be working perfectly or may have made removal so costly that nobody attempts it; only tracing the path tells you which.
Treating irreversibility as commitment
Hard-to-undo decisions get praised as conviction, but they remove the error-correction mechanism that would have caught the ones that were wrong.
Is it for you?
Best for
Anyone evaluating or designing a governance system — a board, a company's decision rights, a DAO, a policy regime, or a country.
Not ideal for
Situations demanding deliberate irreversibility, such as constitutional protections or long-horizon commitments that only work if they cannot be undone on a whim.
From the transcript
“Democracy is the system which allows you to remove policies and rulers most efficiently without violence.”
“But Popper questioned this whole idea of looking at what democracy was.”
“Science is an error correcting mechanism. All we can hope for is to get rid of the bad ideas.”
From the episode
The Beginning of Infinity, Part 2