Mind · Reasoning

Fermi

Estimation via decomposition

Definition

Mechanism

Decomposed Estimation: break unknown quantity into factors you can Estimate. Multiply factors. Accept order-of-magnitude accuracy. Example: "pianos in Chicago" = population × household fraction × piano-owning fraction. Errors often cancel across factors. It invokes Decompose to break an unknown quantity into estimable sub-factors.

Invariants

  1. Error Cancellation: Overestimation in sub-factors tends to cancel underestimation
  2. Geometric Mean: Final estimate is geometric mean of bounds

Risks

Failure modes

  1. Correlated Error Stacking

    If all sub-estimates are biased in the same direction (e.g., all optimistic), the errors compound rather than cancel.

Relationships

Builds on

Composes with

Decompose

Dividing complexity into manageable units

Nothing in this library builds on this pattern yet.

Reference

Parameters

confidence_interval Percentage, [50%, 95%]
Target coverage of true value
decomposition_depth Integer, [2, 6]
Estimation chain length

Identity

Ref
Fermi#f206
Sema ID
sema:Fermi#mh:SHA-256:f206e74f19af3eb55825b02c2192779a627ddff7b1c465a09d73431b863232db
Ring
2
Tier
2
Updated
2026-11-11
Supersedes
  • sema:Fermi#mh:SHA-256:1e06d04bfff04954ad09852a7a4b1b583acc76c431dc926b4afe3cce5abfe210
  • sema:Fermi#mh:SHA-256:128b7131810fba03748bcc5dce657248d5547beeb23e87750cd76e6e59a80bd5
  • sema:Fermi#mh:SHA-256:964a2d4f8166deefb544d347da4f4c36b2c32885bb24f2f3e36cd3361638c9c2
  • sema:Fermi#mh:SHA-256:688559c783f162aee4fc61fd380bd30094219d76a844a6a9efce9ae1feeaf622