Embedding telescope

64d → 256d → 512d

The rectangles are ordered by size — bigger level, longer list — but not drawn to scale.The neighbourhood rectangle sits inside the city one because, under this method, its 64 numbers are the first 64 of the city's 256. The region rectangle (512) holds both. New information is added at each step (numbers 65–256, then 257–512), never lost on the way up. That is the argument that holds up: not a weighted average, but a start that stays consistent.

Cell · Economic × City

City ECI composite (256d)

The Generation-2 cell — the territory the index's earlier generations measured before the full matrix existed; what Boeing measured for Hamburg, what Utopies measured for Paris. The City ECI (Economic Complexity Index) composite has dimensionality 256. The first 64 dimensions come from Community-tier indicators (the prefix); dimensions 65–256 add the city-level industry and consumption signal (NACE / COICOP classifications, Metroverse).

Community indicators that prefix this cell (64d)

Region indicator this cell prefixes (512d)

A note on what this is and isn't

The nested rectangles are a picture of how the numbers stay consistent under Matryoshka Representation Learning, not a literal drawing of the vectors. Readers who want the exact projection function (concatenation-then-PCA? a learned projection? weighted-mean-then-pad?) are sent to Aggregation Rules, where the gap is named openly: version 0 has not yet fixed the projection function for all four pillars — an open item, under review toward v1.