v1.1 — ERC-8004 concentration cap

What this is

The identity layer of the machine economy is largely unbuilt — in law and in code. No jurisdiction recognizes an autonomous agent as a legal entity, and almost none register a technical one: roughly 183,000 distinct agent addresses transact on-chain in a 30-day window, while only a few hundred touch the ERC-8004 identity registry. The index now reads this from two independent directions. This release is how that second reading came into focus.

The change

In late July, a single operator registered ~10,000 agents on ERC-8004 in a sequential batch — real on-chain activity, but one actor, not the breadth the metric is meant to measure. The value-jump safeguard held the figure rather than publishing it, and the question went to methodological review.

The ruling: cap each registering address's contribution to the 30-day count at 100 events, applied every window. One actor, at most, can look like a barely-active registry — never a thriving one. The cap conditions only on public on-chain structure (which address registered how many), never on inferred intent.

The numbers

Holding all other metrics at today's values and toggling only the ERC-8004 basis, the cap changes the MEI by −2.0 (33.8 → 31.8). Without it, today's index would read 33.8 — the batch had been lifting the live figure by about two points; the cap removes that lift.

The metric's calibration anchor, re-derived on the capped basis across full history, moves from 1,702 to 606 — the concentration was structural, present on ~80% of days, not unique to the July batch. Because the index normalizes on fixed bounds rather than the anchor, this large re-derivation barely moves the composite.

At the corrected level the metric sits below the standard launch band. The bounds are unchanged; instead, a documented nascent-floor exception ([15, 65], mirroring the existing mature-market exception) admits it. The sub-band position is itself the reading: technical-identity adoption is close to absent.

The central finding is unchanged. Legal remains the weakest rail under every weighting tested (100% of sensitivity draws).

Standing review

Three conditions govern the metric going forward: a concentration re-review if dominance persists, immediate re-adjudication if a better-verifiable venue emerges, and automatic retirement of the exception once genuine breadth arrives (≥2,000 distinct registrants for 90 days).