Start Building

Lesson 90: First-Week Intervention-Governance Drift Check - Ingestion Totals Versus Published Copy Before the Executive Readout in RPG Live-Ops

Direct answer: A first-week intervention-governance drift check is a daily signed row (plus policy contract) that compares fresh Lesson 86 ingestion totals for the new train’s windows against the published governance copy and deck version you owe under Lesson 87—before you hold the train’s first executive readout. It turns “we think interventions worked” into hash-bound drift evidence anchored to your Lesson 89 charter.

Pixel-style dog breeds lineup suggesting multiple intervention lanes that must stay distinguishable under review

What this lesson solves

New trains launch with good intentions and loud narratives. Without a drift check, teams reconcile numbers only at the readout, when politics peaks and time runs out. This lesson forces early, boring comparison so interventions are scoped, ticketed, and measurable.

Prerequisites: Lessons 86, 87, 89, and a live lesson78_next_train_opening_charter.csv row. Expected time: about eighty minutes including one tabletop “yellow readout” rehearsal.

What you will build

  1. lesson78_first_week_intervention_governance_drift_policy.md (sections below)
  2. lesson78_first_week_intervention_governance_drift_log.csv (one row per calendar day in days 1–7 after charter sign)
  3. A readout gate rule that blocks executive presentation when drift is red under your thresholds

Step 1 - Define drift gate classes

gate fail signal action
F1 – Ingestion incomplete any merge_status = pending_owner rows whose causal tags affect intervention metrics cited in published bullets hold drift sign-off
F2 – Deck mismatch governance_deck_version in slides does not match Lesson 87 reconciliation target for the train hold drift sign-off
F3 – Drift over threshold drift_severity = red per policy table block or downgrade readout
F4 – Unticketed narrative published bullet claims an intervention outcome with no intervention_ticket_id open governance defect, not a drift row

Yellow drift can proceed only with a written waiver id logged in the csv.

Step 2 - Author lesson78_first_week_intervention_governance_drift_policy.md

Minimum sections:

  1. Purpose – prove ingestion and copy agree before executives rehearse numbers.
  2. Scope – days 1–7 after charter_signed_at_utc; which causal factors are in-scope for the train.
  3. Comparison method – same rolling definitions as Lesson 87 (mismatch_class vocabulary).
  4. Threshold table – green / yellow / red bands for % delta or absolute deltas per metric class.
  5. Readout gate – who may override a red drift (incident command + divergence co-sign).
  6. Link to charter – cite charter_row_id and new_train_cycle_id from Lesson 89.

Step 3 - Author lesson78_first_week_intervention_governance_drift_log.csv

column purpose
drift_row_id monotonic id
train_cycle_id matches Lesson 89
charter_row_id_ref ties back to opening charter
check_day_index integer 1–7
check_date_utc wall-clock date of sign-off
ingestion_window_ids list or range covered that day
ingestion_snapshot_hash hash over filtered Lesson 86 rows in scope
governance_deck_version copy from published materials
published_bullet_ids ids for bullets under review
mismatch_class reuse Lesson 87 classes where applicable
drift_severity green, yellow, or red
drift_delta_summary short numeric statement
intervention_ticket_ids comma-separated when claims tie to work
blocks_readout true / false
owner_lane accountable lane
signer_divergence initials
signer_incident_command initials
drift_evidence_hash sha256 over canonical concatenation of fields above

Step 4 - Run the first-week ceremony (50 minutes)

  1. Morning: export Lesson 86 window; compute ingestion_snapshot_hash.
  2. Midday: capture deck version and bullet ids actually live or approved for rehearsal.
  3. Afternoon: classify mismatch_class; apply threshold table; set drift_severity.
  4. If red: open intervention or governance defect; do not schedule executive readout until waiver or fix.
  5. Sign: initials; archive immutable pointer consistent with Lesson 88 patterns.

Step 5 - Tabletop - yellow drift with political pressure

Simulate a yellow drift where comms wants the readout on schedule anyway. Outcome: downgraded readout with explicit waiver id, or slide rewrite before mic time—never silent rounding.

Troubleshooting

Do we compare to Lesson 88 baseline?

No for drift magnitude—that baseline closes the prior train. You do cite the charter row to prove which train you are measuring.

What if Lesson 87 reconciliation is still open?

You cannot claim governance_deck_version is authoritative. F2 fails; finish Lesson 87 mechanics first.

Who owns intervention_ticket_ids?

Divergence proposes links; incident command validates against live incident backlog.

Lesson recap

You can earn the executive readout by proving numbers and copy were aligned while interventions were firing—not only after.

Next lesson teaser

Next, complete the Lesson 91: Days 8–14 Cumulative Drift Rollup Escalation Packet so second-week trends become a signed escalation packet when daily greens hide a bad trend in the same mismatch class.

Related learning

Treat the drift log as the early warning rail, not a slide polish pass.