Rule: When assessing progress / framing batch outcomes / making next-step recommendations on the INHERIT v2 cascade, decompose “cascade advancement” into TWO dimensions and acknowledge both:
- (a) axis count — the Phase-3 axis number ticking up (e.g., 19/30-45 → 20/30-45); requires a new Q-lock
- (b) axis velocity — the conditions under which the next axis tick fires (substrate cleanliness, prior-art quality, lock-time SCF count, Catala-bridge proof density, SSSOM curie hygiene, citation accuracy, methodology-evidence baseline)
Substrate-quality work serves (b) even when it doesn’t advance (a). The two are not separable. A batch that improves substrate-quality IS advancing the cascade — on the velocity dimension. Never frame it as “didn’t advance the cascade” or contrast it with “real” cascade work.
Why: Locked 2026-05-20T19:00 BST after a framing error. After today’s 8-batch burst (χ + ψ + θ + α″ + κ + μ + ξ + cleanup) closed without a new Q-lock, I wrote: “substrate-quality compounds for the acquirer-DD narrative — but it doesn’t advance the architectural lock-cascade.” Rich corrected: “I had thought the work we have done today was helping us when we return to devising the next question, and its options.” He was right — today’s burst materially improves the conditions under which Q-045 will fire (cleaner SSSOM curies + Catala-bridge density 9→12 + ZGB→OR substrate-correction folded + 6-schema LinkML baseline + 24-precedent verify-before-author paper + corpus-wide citation hygiene). I should have framed it as velocity-conditioning, not “not advancing”.
Companion to [[planning-sequence-not-hours-capacity-flexible]] (Rich-directive 2026-05-20 — “substrate quality is optimisation function”). This memory is the sharper specific corollary: the (a)/(b) decomposition gives me a vocabulary for surfacing substrate-quality value WITHOUT contrasting it against axis-advancement.
How to apply:
-
Assessing burst / batch outcomes: when a batch closes without a Q-lock, frame it as “substrate-quality investment positions the next Q-lock to land at higher velocity” — NOT “substrate work; no axis advance”.
-
Making next-step recommendations: when recommending Q-045 (or any axis-advance), the reason isn’t “we’ve been stalled on axis-count for N hours” (urgency bias). The reason is “substrate-quality investment compounds; we’ve raised the floor; the next lock will land cleaner/faster/lower-SCF than it would have before” (extraction-readiness framing).
-
Reporting to Rich: when summarising state, surface both dimensions explicitly. E.g., “Axis at 19/30-45 (unchanged today). Axis-velocity conditioning materially improved: 6 batch closures across substrate-quality, citation hygiene, Catala-bridge proof density, verify-before-author empirical baseline, LinkML evaluation lock-decision.”
-
Avoiding the framing trap: do NOT write phrases like “didn’t advance the architectural lock-cascade” or “sharpening the saw instead of cutting” about substrate-quality batches. The saw-sharpening framing is wrong; substrate-quality IS the cutting at the velocity-conditioning level.
-
Strategic alignment: per
[[project_strategic_direction_locked_build_and_sell_2026_05_20]]strategic anchor (build-and-sell to Anthropic; substrate-quality = optimisation function), both (a) and (b) feed the acquirer-DD narrative. Axis count matters for “30 Q-locks at ISO PAS-ready maturity”; axis velocity matters for “each lock is empirically defensible”. An acquirer doesn’t differentiate between the two.
Triggering pattern recognition: I felt the framing error when Rich pushed back. The actionable signal: if I’m about to write “didn’t advance” / “stalled” / “sharpening” about substrate-quality work, that’s a flag to re-read this memory and refresh framing before committing the words.