Mind · Strategy
CorrectAsYouGrow
When the total investment is not known in advance, add corrective information as mass is added, from more than one model
Definition
Mechanism
Because of MassAmplifiedError, a frozen plan needs initial accuracy matched to the total mass it will carry, which is only possible when that total is known (the concrete for a building). When it is not known (how much money will ride on a royalty line, how far a product will be used), switch from one up-front decision to a cadence of corrections: add mass in increments and check shape between them, like a delivery drone that flies slowly enough to re-plot its course, or a royalties team that re-checks data as more money is allocated to it. Concentrate checking where mass is accumulating rather than everywhere at once. To avoid SelfCheckingBias, the corrections must draw on overlapping, non-identical models, i.e. on MetaModularity, not only on fresh data for the original model.
Preconditions
- Mass can be added incrementally with points to correct in between
- The final amount of mass is uncertain or open-ended
Invariants
- The correction cadence must keep pace with the rate at which mass is added
- Corrections are focused where mass and therefore potential error is accumulating
- At least some corrections come from a model other than the one that created the shape
Risks
Failure modes
Fixing quality standards up front for an investment whose final size is unknown
Correcting only with the original model and so missing systematic error
Applying it where some things must be right from the start and cannot be corrected later (the lean of a foundation)
Relationships
Builds on
References
A fixed error rate in a frozen model grows into a larger absolute error the more is invested in that model
A body plan that holds many overlapping but non-identical models of the same problem, ready to instantiate whichever fits
Correcting as you go still fails when the same model that produced the shape is the one checking it
Nothing in this library builds on this pattern yet.