Error Rate Management
also Reliability standards · Tolerable error rate
Setting the reliability an idea or skill must reach in proportion to how often and how deeply it will be reused, since errors in heavily-reused building blocks multiply through everything built on top of them.
Error rate management is CF’s claim that the reliability standard for any idea or skill should scale with how heavily it will be reused, not be a flat figure like the schoolroom’s 95%. The governing metaphor is a building: the more layers you stack on a block, the more weight it bears, so the stronger it must be. An idea with a 1% error rate used once a year, under conscious attention, is fine — you will notice the slip and fix it. The same 1% rate in a low-level subconscious skill invoked hundreds of times a day produces multiple errors daily, often where you cannot see what or where the fault is. CF draws the parallel to software: a single call to a high-level function can trigger thousands of calls to low-level ones, so those low-level functions need extraordinary reliability or building twenty layers above them becomes impossible.
The practical upshot is that foundational, frequently-invoked building blocks need a success rate like 99.9999%, not 95% or 99%, before you treat them as settled and let other knowledge depend on them. This is why CF couples error rate management with integration and mastery: you keep practicing past mere conscious competence until a skill is automatic and near-flawless, precisely because it will become load-bearing.
CF ties this to its broader picture of a weighted error rate (quantity times severity) that must stay below your error-correction capacity, with a buffer for statistical fluctuations. Demanding adequate reliability before stacking new layers is the discipline that prevents overreaching — attempting work whose error rate outruns what you can correct.