Artifact: Rewriting a Postmortem into Blameless Language

ex-43 · exercises co-23 · the same incident, named-individual language rewritten systemic.

Draft (names an individual):

Sam ran the stale ops-toggle script without checking its current state first, which re-disabled
gift-card redemption for 39 minutes. Sam should have verified the flag's state before running the
script.

Rewritten, blameless:

The ops-toggle script re-disabled gift-card redemption for 39 minutes when it was run without a
pre-run state check. The script had no built-in guard against re-applying a change that was already
in effect, and no confirmation step surfaced the flag's current state before running -- both gaps
that any operator running this script, on any day, could equally have hit.
 
Fix: the script now prints the current flag state and requires an explicit confirmation before
applying a change (shipped same day). Longer term: audit other ops scripts for the same missing
guard.

Verify: the rewritten version contains zero sentences attributing the incident to a specific person -- every sentence's subject is the script, the gap, or the fix, satisfying co-23's actor- neutral rule.

Key takeaway: "Sam should have checked" blames a person for a gap the SYSTEM should have closed; "the script had no guard" names the same gap as a fixable property of the tooling, which is also the only version of the sentence that actually prevents a recurrence.

Why It Matters: blameless language is not about politeness for its own sake -- it changes what the postmortem's fix section can even contain. "Sam should be more careful" produces no code change and will happen again to whoever runs the script next; "the script had no confirmation guard" produces the confirmation-prompt fix shown above, which protects EVERY future operator, including Sam.