Aggregation timing — surface 06 / 06

The matryoshka cannot resolve a Region score faster than its slowest input. Why your Region score is from last year.

Why each ring of the nested index — the matryoshka — refreshes at a different speed. Community sensors update hourly. City portals update quarterly. EUROSTAT regional MFA (material-flow accounting) arrives with an 18-month lag. The timing surface stacks these cadences — how often each source publishes — so the reviewer can see which scale they constrain. The slowest input governs how often the matryoshka's outermost ring can refresh — and it is the EUROSTAT lag at Region scale.

11 indicator classes · 5 cadences Slowest input · 540-day lag (EUROSTAT MFA) Fastest input · ≤ 1-hour (PM2.5 sensors)

Community-scale indicators (hours → months) City-scale indicators (hours → annual) Region-scale indicators (annual cadence · up to 18-month lag)

The matryoshka's slowest link

The Region ring cannot complete a refresh faster than its slowest input. That input is currently: 540 days

EUROSTAT regional Material Flow Accounting publishes annually with an 18-month lag. Until that lag tightens — or until a faster regional proxy is added to the methodology — the Region score is an 18-month-rear-view image. The Region ring's mock pill — the sample-data label on its score — should always note this. This is not a flaw in the matryoshka; it is a property of the regional governance scale, and the dashboard's job is to make it visible.

Why the cadence matters for ρ

The response coefficient ρ measures observation-to-action latency. A PM2.5 spike at the neighbourhood council can be acted on within a week — that's a high ρ at Community. A Generalitat (Catalonia's regional government) indicator that lands 540 days late cannot be acted on inside its own measurement cycle. ρ at Region is mathematically capped by the cadence.

And one discipline note: ρ is only a measurement where a fitted-response protocol — an agreed way of measuring how fast action follows observation — exists at that tier. Region-tier ρ is currently mock — ρ protocol pending in every pilot (no partner operates a regional policy-adoption tracker). Where ρ cannot be fitted, the dashboard renders n/a — protocol pending; ρ is never silently defaulted to 1, because an unmeasured response is not a perfect response.

What the matryoshka does to the lag

The MRL prefix-projection — the nesting rule where each scale's record begins with the smaller scale's — means the Region embedding's first 256 dimensions are the City embedding. When City refreshes quarterly, those dimensions update inside the Region embedding even before the EUROSTAT annual release. The full Region vector is only complete after the slowest input lands, but the prefix is current. Reviewers should be told this: it is the answer to “why is the Region score updating monthly when MFA only lands annually?”

↪ Return to the aggregation atlas · or read the scale-by-scale walk from the ground up.