Great Filter Constraint Analysis
Locate the missing step by forcing incompatible beliefs to compete
- Difficulty
- Advanced
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 94%
Great Filter analysis starts from a discrepancy: billions of possible origins for life exist, yet humanity sees no extraterrestrial arrivals. Something in the chain from a habitable planet to a spacefaring civilization must be extremely difficult or systematically suppressed. Map the candidate transitions, then test each against available evidence. A step that occurred independently several times on Earth, such as advanced cognition across different lineages, is less credible as the dominant filter. A uniquely slow or unobserved transition remains more plausible. The method also exposes incompatible belief bundles: easy intelligence, feasible colonization, diverse motives, and zero visible civilizations cannot all be true together. At least one assumption must change.
Origin
Bostrom applies the Fermi paradox and the Great Filter idea to constrain beliefs about life, intelligence, and space colonization.
Core principles
- 01A large input and zero observed output imply a strong filter
- 02Repeated independent emergence makes a candidate filter less plausible
- 03Incompatible beliefs cannot all remain untouched
- 04Observation selection can explain why observers see rare prerequisites
How to run it
- 1
Name the discrepancy
Set out the large number of plausible starting points and the unexpectedly small number of observed endpoints.
Pro tip Use orders of magnitude before arguing about fine precision.
Watch out Confirm that the endpoint would actually be detectable.
- 2
Map the transition chain
List the major steps required to move from origin to observable success.
Pro tip Separate biological, technological, motivational, and expansion steps.
Watch out Do not hide multiple difficult transitions inside one vague stage.
- 3
Test candidate rarity
Use elapsed time, independent repetitions, and missing observations to assess where extreme improbability could plausibly sit.
Pro tip Independent recurrence is especially useful evidence against a candidate bottleneck.
Watch out A long duration alone does not prove low probability.
- 4
Force the belief trade-off
Write down the set of assumptions that jointly predict an endpoint you do not observe, then identify which assumption must yield.
Pro tip Make the contradiction explicit in one sentence.
Watch out Do not preserve every favored assumption by inventing an untestable exception.
- 5
Update the filter location
Re-rank candidate bottlenecks as new evidence appears, especially evidence about transitions already passed.
Pro tip Distinguish filters behind us from filters potentially ahead of us.
Watch out The analysis constrains possibilities; it rarely identifies one certain cause.
In the wild
A marketplace has thousands of qualified visitors but no completed purchases. The operator maps discovery, product view, checkout start, payment, and fulfillment. Analytics show many checkout starts but no payment confirmations, while earlier stages repeat successfully across channels. The Great Filter structure directs investigation toward payment rather than vague claims that demand is absent.
→ The hidden bottleneck is narrowed to the transition that best explains abundant starts and zero endpoints.
Common mistakes
Keeping incompatible assumptions
If the assumptions predict many visible endpoints and none appear, at least one assumption must be revised.
Picking a familiar bottleneck
A candidate feels intuitive but should fall in rank when the transition has occurred independently several times.
Is it for you?
Best for
It is best for strategic systems with a long chain of prerequisites and a surprising absence of expected results.
Not ideal for
It is not ideal when the endpoint is poorly observable or the starting population cannot be estimated at all.
From the transcript
“There has to be some great filter that you start with billions of germination points and you end up with a net total of zero…”
“if it happens several times independently on Earth then it can't be that unlikely”
“something has to give and it gives us clues”
From the episode
Nick Bostrom: How Entrepreneurs Can Win in an AI-Dominated World
Nick Bostrom