Metric pressure is not a defense

What it is

"It's the target from above" is not a defense against harm: pressure changes how hard it is to act, not what the act is. Metric pressure as no excuse holds that when the means of hitting a metric involve foreseeable harm, the executor's responsibility for choosing those means does not vanish because the metric exists — it merely forms a shared chain of responsibility with whoever set the target. Metrics can be questioned, negotiated, and refused; they are not laws of nature.

Why it happens

Metrics serve as moral exits through three disguises. Depersonalization: metrics describe harm as the necessary cost of numbers ("retention would drop two points"), so harm never appears in the decision as specific people and events; numbers do not cry out, and whoever writes the line never faces the sound. Necessity: management choices are presented as market laws ("every competitor does this"), as if no alternative existed — when in fact costlier or slower alternatives were actively abandoned. Dilution: the contribution to harm is spread across many small decisions, each only "slightly" worse (pre-checked boxes, one more retention step, on-by-default), individually trivial and jointly a substantial transfer of harm. Together the disguises convert an arguable value trade-off into a "system outcome" that apparently no one chose.

Where it stops holding

This does not demand costless individual defiance: refusal carries real career risk, and a responsibility ethic acknowledges risk and grades accordingly — the floor is recording dissent in writing, escalating through alternative proposals, reassignment requests, and internal escalation, with public refusal at the top; the chosen rung is personal discretion. Symmetrically, management bears the heavier responsibility for setting the metric — stressing the executor's non-immunity never lets the target-setter disappear. Metrics are also mostly proxies for legitimate goals (growth, safety, stability); the problem begins when the proxy absolutizes itself and devours the goal — goal displacement.

Applying it

• Split the proposal document into two columns: "what the metric demands" and "which means we chose"; the second column is where moral attribution lives and must not be overwritten by the first.
• Build an escalation lane for metric conflicts: when the means of hitting a target involve foreseeable harm, the decision moves up one level with a written record, never quietly digested at the execution layer.
• Translate metric language back into harm language at reviews: read "retention +2%" alongside "roughly how many people retained how many extra sessions," two lines side by side.
• Verify: sample features with harm disputes and inspect their documents — if the "means" column is answered entirely by metric requirements, the disguise is working and the process needs repair.

Related

• Same group: P4.07.1 Executors bear responsibility for foreseeable harm · P4.07.3 Refusing and recording dissent are viable professional acts
• Adjacent: P3.06 Notification-driven engagement and growth metrics · P4.14 Institutionalizing ethics review
• Search terms: goal displacement · metric fixation · responsibility under incentives