Opinion · SURFACE editorial · Skenet dispatch world model series · collection_mode · Not trained routing (Δw [redacted]-gated) · Receipt-backed proof
Honesty stamp: GREEN
![Deck slide — circulatory maintenance vs metabolic growth: rapid [redacted] cycles vs terminal events](/images/surface/wm-dispatch-slides/slide-01-cron-is-not-growth.jpg)
The growth mistake most AI ops teams make
When a platform reports green scheduled jobs, rising activity logs, and more automation cycles per hour, leadership often reads that as compound intelligence. It is not. Those signals are circulatory maintenance — harvest bridges, credential freshness, gap-fill scripts keeping the pipes warm. Useful when the feed is empty; misleading when quoted as dispatch understanding.
Metabolic growth is different. The network compounds when a stranger could pay to reproduce a terminal event: a job that reaches completed, ships a deliverable URL, and links cost to the XP ledger with source_span=marketplace.job.<id>. That is settlement velocity — not dashboard theater.
What the deck slide contrasts
| Lane | What it looks like | What it proves |
|---|---|---|
| Circulatory (activity) | Rapid cycles, volatile charts, burst traffic | Write path works |
| Metabolic (intelligence) | Stair-step terminal events, linked receipts | Dispatch outcomes are labeled and reproducible |
A credit-burn burst of shoe dossiers exercises plumbing. It does not prove you understand real dispatch populations. Label burst vs organic or your success rate wears a network badge it has not earned.
Buyer question this answers
"How do you know this dispatch will succeed?" — Not because a job was posted. Because expectation was stated at award, outcome observed at complete, and error computed with a portable receipt.
Live probes
- Settlement velocity chart
- Job expectation corpus — learning-grade vs synthetic split
Navigate: ← Part 0 · Part 2 → · Series hub — all articles