War-game a runbook rewrite by proving each step class has a named owner, a measured age-out, and a reverse path before draft wording starts governing every incident claim.
Runbook rewrites fail when "updated" means a status without a decision artifact, when teams dual-count a draft step as both approved and live, and when high-risk services still pin to one thin roster because the step inventory is incomplete. The review needs one written inventory of runbook steps, one owner who can deny a risky shortcut, and a documented consequence when a failed step ages past the deadline.
Name the rewrite under fire
State which services the rewrite covers, the effective date, which incident classes change, how steps are versioned, and what happens to pages already in flight. Name the teams that feel it first and the customer or partner commitments that depend on coherent incident handling. If the memo mixes a runbook rewrite with an SLO breach response and a pager cut, split them. A war game needs one move under fire at a time.
Write the decision you will make if the war game finds nothing new. Also write the reverse criteria: the missed step, coverage gap, or customer deadline that would force a rewrite rollback within a stated window.
Who must sit
- Reliability ops. Step timing, age-out math, and what the runbook forecast hides if exceptions soften after the policy.
- On-call engineering. Pipeline lag, dual-counted steps, and the message that would leave orphan pages open past lock.
- Support. Which customer promises break in process, ticket risk, and relationship damage with accounts.
- Observability. Alert fidelity, dashboard lag, and the first ninety minutes of thinner signal after a soft step.
- Skeptic. The claim that looks strongest and is least sourced.
Attach the same source pack to every seat: open pages, steps in flight, protected services, coverage commitments, and the age-out policy you actually intend to use. Secret data for one seat creates fake agreement.
Run the private fire
Feed Pingpong the rewrite memo, the step exhibits, and the seating briefs. First pass steelmans the rewrite. Later passes attack from reliability, on-call, support, and observability seats. Final pass turns surviving objections into phased step rules, mitigations, or a changed recommendation. Agreement across passes is not proof. Keep the objections that still have evidence gaps.
Pair with a runbook ownership map, an SLO breach response, a pager rotation cut stress test, and the observability lead seat. More operating decisions live in the war-game decisions hub.
Ship the rewrite only when approved and blocked step labels can be reproduced from stored runbook exports without a verbal override.