Start Building

Lesson 91: Days 8–14 Cumulative Drift Rollup Escalation Packet - Trending Mismatch Classes Before Mid-Train Escalation in RPG Live-Ops

Direct answer: A days 8–14 cumulative drift rollup escalation packet is a single signed artifact that ingests your Lesson 90 daily drift rows for calendar days 8–14 after the Lesson 89 charter, then computes trend signals per mismatch_class—so you can escalate before a quiet, compounding governance failure becomes an executive surprise. It is the difference between “we were green all week” and “green was masking a monotonic slide in the same failure mode.”

Pixel airplane suggesting a mid-train climb where only instrument-rated evidence should clear you through weather

What this lesson solves

Daily drift checks are honest but noisy. Teams that only read day-over-day severity miss cumulative stories: repeated yellows in one class, slowly widening deltas, or “fixed” bullets that keep re-opening the same mismatch_class. This lesson forces a second-week rollup that is hash-linked to the underlying drift evidence, so mid-train escalation meetings cannot substitute narrative slides for the packet.

Prerequisites: Lessons 86–90, a complete lesson78_first_week_intervention_governance_drift_log.csv for the train through at least day 14 (or a documented waiver for missing days). Expected time: about eighty-five minutes including one tabletop “green camouflage” rehearsal.

What you will build

  1. lesson78_days_8_14_cumulative_drift_rollup_policy.md (sections below)
  2. lesson78_days_8_14_cumulative_drift_rollup_packet.csv (one row per rollup cycle—typically one per train per days 8–14 window)
  3. A mid-train escalation gate that blocks stakeholder-facing escalation until the rollup is signed and linked to source drift row ids

Step 1 - Define rollup gate classes

gate fail signal action
R1 – Incomplete window fewer than five signed Lesson 90 rows in days 8–14 unless a waiver id explains absence hold rollup; do not declare “no trend”
R2 – Trend masked by daily color monotonic worsening in numeric delta or repeated same-class yellow/red streak while mean severity looks “fine” force trend_severity = red_orange per policy ladder
R3 – Escalation without co-sign incident_command has not confirmed source_drift_row_ids against the live log export block external escalation comms
R4 – Detached narrative slide deck cites outcomes not listed in intervention_bundle_ids tied to the rollup open governance defect; rollback deck

Step 2 - Author lesson78_days_8_14_cumulative_drift_rollup_policy.md

Minimum sections:

  1. Purpose – compress second-week drift into an escalation-grade summary without losing causal traceability.
  2. Scope – days 8–14 after charter_signed_at_utc; which trains and which mismatch_class taxonomy (reuse Lesson 87 vocabulary).
  3. Trend methods – define how you compute trend: slope on % delta, streak counts, or rank change across days; pick one primary and one backup for sparse data.
  4. Severity fusion – how daily drift_severity rolls up into cumulative_drift_severity and trend_severity.
  5. Escalation triggers – explicit thresholds that mandate mid-train actions (extra ingestion window, deck rewrite, readout delay).
  6. Evidence chain – require listing source_drift_row_ids and their drift_evidence_hash values in an appendix block (can be a hashed list object in your tooling).

Step 3 - Author lesson78_days_8_14_cumulative_drift_rollup_packet.csv

column purpose
rollup_packet_id stable id for this packet
train_cycle_id matches Lesson 89
charter_row_id_ref ties to opening charter
window_start_day_index 8
window_end_day_index 14
source_drift_row_ids comma-separated Lesson 90 ids included
mismatch_class_trend_summary short text per class with directionality
cumulative_drift_severity fused rollup severity
trend_severity worst trend outcome across classes
blocks_mid_train_escalation true if gates say “no external escalation yet”
intervention_bundle_ids linked remediation packages or tickets
signer_divergence initials
signer_incident_command initials
rollup_evidence_hash sha256 over canonical concatenation of prior hashes + ids

Step 4 - Run the rollup ceremony (45 minutes)

  1. Export the Lesson 90 log slice for days 8–14; verify each row’s drift_evidence_hash parses.
  2. Compute trends per mismatch_class; flag R2 patterns even when individual days are yellow-green.
  3. Fuse severities; set blocks_mid_train_escalation honestly—politics is not a column.
  4. Bundle interventions; refuse empty intervention_bundle_ids if trend_severity is not green.
  5. Sign divergence + incident command; store immutable archive pointer consistent with Lesson 88 patterns.

Step 5 - Tabletop - green camouflage

Simulate a week where every day is green except a single mismatch_class whose delta creeps upward monotonically but stays under the daily red threshold. Outcome: rollup must still elevate trend_severity and attach a mandatory intervention bundle—no “we’re fine” conclusion.

FAQ

Is this the same as Lesson 87 reconciliation?

No. Lesson 87 reconciles deck vs ingestion for a readout. This packet asks whether second-week drift dynamics justify escalation posture before you spend executive attention.

What if we skipped a day in Lesson 90?

Log R1; either backfill with a time-boxed catch-up export (signed) or record a waiver that names the missing evidence and downgrades what you are allowed to claim in escalation.

Who owns intervention_bundle_ids?

Divergence proposes bundles tied to causal factors; incident command validates that bundles map to real work tracked in the incident system.

Lesson recap

You can earn mid-train escalation by proving trend-level governance truth—not by winning the daily color lottery.

Next lesson teaser

Next run Lesson 92: Mid-Train Escalation Dry-Run so this rollup packet is the only approved source for the war-room deck—stakeholders cannot smuggle alternate narratives past the hash chain.

Related learning

Treat the rollup as flight instruments, not a post-meeting PDF.