Bystander effect in large organizations

Child note of bystander-effect: how the effect is countered in large organizations, where process is affordable and the failure mode is diffusion itself.

Countermeasures, reactive (diffusion already exists). (1) One named owner per item (DRI) — a name, never "the team"; two owners is zero owners. (2) On-call rotation — the roster decides who responds first. (3) Handoff by explicit acceptance — ownership transfers only when the next person says "I have it", never by the clock. (4) Active incidents stay — a responder on a live incident stays until it is mitigated or explicitly relieved. (5) Escalation clocks that do not reset on handoff — dragging makes the problem louder, not quieter. (6) Visible holding time — who held an item and for how long is on the record.

Countermeasures, preventive (diffusion never gets a chance). (1) Route to one person, never broadcast — a message to a group is addressed to nobody; @all is the largest known diffusion device. (2) Owner mandatory at creation — every service, ticket, and document is born with one owner name; orphan items are not allowed to exist. (3) Name the person — direct address is the one countermeasure verified in the lab tradition itself: a victim who singles out one bystander ("you, call an ambulance") restores helping. (4) Automate first response — auto-rollback, auto-ticket-and-assign; a machine does not wait for someone else to move.

The on-call hole. Rotation alone invites end-of-shift dragging: hold the problem until the next shift inherits it. The fix is one sentence — the roster decides who responds first, not when you may drop it. Ownership ends at resolution or accepted handoff, never at the clock.

Transfer to team ownership. "Everyone owns it" is diffusion by design. A team that talks about ownership without naming one owner per item has recreated the smoke-filled room as an org chart: everyone waiting for someone else to move first.

Parent: bystander-effect · small-teams · open-teams