Skip to main content

Compounding Standards

How does a good decision become the floor for the next cycle?

A standard is persisted state: a decision that survived evidence, became an interface, and lets the next participant begin from a higher baseline.

Standards Are Fittings

Pipes, threads, voltages, ports, schemas, and protocols reduce custom negotiation because their interfaces fit. Business work compounds the same way. A useful standard:

  • makes inputs, outputs, authority, and evidence fit across people and systems;
  • removes routine naming and handoff judgment;
  • makes outcomes comparable and auditable;
  • reduces bespoke glue on the next run.

Standardization is not paperwork. It is the stable interface that lets value flow.

The Onboarding Test

Give an unfamiliar person the artifact and the will to do right. Can they contribute safely with minimal hidden training? If good will and attention still permit predictable error, the standard has a gap. Quality should depend on the process, not a hero remembering unwritten context.

The Feedback Loop

PLAN → DO → CHECK → ACT
↑ ↓
└──── better baseline
StepOutput
PlanSetpoint, owner, limits, and expected evidence
DoA measured run under the declared protocol
CheckExpected-versus-actual variance
ActA correction retained in the rightful standard

Quality Tiers

Use the strongest relevant reference: a formal obligation, an industry consensus, or a declared convention. A higher-status source does not remove the need to show that the artifact passes its own quality criteria.

Use one commissioning ladder:

  • L0 — absent: the artifact does not exist.
  • L1 — drafted: its intended job and shape are visible.
  • L2 — built: it passes structural checks.
  • L3 — used: a real case produced feedback.
  • L4 — transferred: every required gate passes and an independent person in the served role validates it.

Do not average mandatory gates. One failed gate means commissioning is incomplete.

Limits

A standard becomes bureaucracy when it cannot name the variance it prevents, the interface it stabilizes, or the evidence that should change it. Retire controls whose cost exceeds their proven protection.

Changes my mind: repeated work performs better without the interface, or the standard blocks safe adaptation without reducing meaningful variance.

Context

Questions

Next question: which interface still depends on hidden training or heroic memory?