# Metric Ethics

> Use when a product will publish numbers that score work - quality scores, productivity dashboards, SLAs, agent evaluations: anti-goals enforced by architecture rather than prose, a counter-metric blended into every pressurable metric, confidence carried with every score, human-state heuristics kept outside the composite, and the artifact rather than the person as the unit of evaluation.

- Skill: `konradcinkusz/metric-ethics` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add konradcinkusz/metric-ethics`
- Raw SKILL.md: https://api.skillmd.com/api/skills/konradcinkusz/metric-ethics/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: konradcinkusz (https://skillmd.com/u/konradcinkusz)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/konradcinkusz/metric-ethics

---


# Metric ethics: measuring work without measuring people

**Read [`references/METRIC-ETHICS.md`](references/METRIC-ETHICS.md) before applying any of this.**
That file is the standard; everything below it is a summary to help you decide
whether this skill applies and to check your work afterwards.

Reference-architecture principles: P14.

## What this standard covers

- Anti-goals go at the top, and are enforced by architecture
- Every pressurable metric carries its counter-metric
- No number leaves without its confidence
- Heuristics about people are report-only
- The unit of evaluation is the artifact, never the person

## Failure modes

| Symptom | Cause |
|---|---|
| A score intended as a diagnostic appears in a performance conversation | The unit of evaluation is the person, or a per-person view exists for someone to sort (§5, §1) |
| A metric improves steadily while the underlying outcome does not | The pressurable metric shipped without a counter-metric, so the cheapest way to move it is the degenerate one (§2) |
| A counter-metric exists and nobody looks at it | It is reported beside rather than blended into the same component, so consulting it is optional (§2) |
| Two scores are compared and the comparison is meaningless | Confidence or sample size is not carried with the number (§3) |
| A sentiment heuristic ends up driving a decision | It was inside the composite at a low weight rather than outside it entirely — and a weight is one config change from non-zero (§4) |
| The README's anti-goals are cited in an argument and lose | They were prose with nothing in the architecture behind them (§1) |

## Checklist

- [ ] Anti-goals are stated near the top of the README, not in a footer
- [ ] At least one anti-goal is enforced by an architectural absence — a view that does not exist — rather than by policy alone
- [ ] Every pressurable metric has a named degenerate strategy and a counter-metric that catches it
- [ ] The counter-metric is blended into the same composite component, not reported beside it
- [ ] Confidence or sample size travels with every exported score, in the same payload
- [ ] The documentation says how to read a score/confidence pair
- [ ] Heuristics about human state live outside the scoring engine entirely, not inside at zero weight
- [ ] The unit of evaluation is an artifact — session, request, change, run — and never a person

---

Generated from [`docs/guides/METRIC-ETHICS.md`](https://github.com/konradcinkusz/architecture-standards/blob/main/docs/guides/METRIC-ETHICS.md) by `scripts/build-marketplace.mjs`. Do not edit this file: change the source document, or its entry in `catalog/marketplace.catalog.json`, and re-run the generator.