Gas sponsors

get_address_fanout reports sponsor_shape for an address that pays other people’s gas. This is invisible to value fan-out, since sponsoring moves none of the sponsor’s own money.

A sponsor that also sent a coin to at least half the addresses it sponsors (sponsored_and_paid_count) is an operator running those wallets, the relationship build_wallet_edges links. A public relayer pays gas for strangers it never funded.

An operator that funds from one address and sponsors from another also reads relayer here; build_wallet_edges tests for that split by the first funders of the wallets the sponsor serves.

relayer and operator are proven. private_sponsor off a truncated scan is flagged provisional, since breadth only grows with the window.

build_wallet_edges needs complete balance changes before choosing a first funder. If a transaction’s remaining rows cannot be read, the lookup is unread: no first funder is chosen or cached, and no funding edge rests on that lookup. The result reports truncated: true and explains the unread lookup in notes. This also applies to the first-funder checks used for expansion and for a sponsor/funder split. An unread lookup does not mean the wallet had no funder.

For recent seed activity, an incomplete balance-change list supplies neither co-appearance nor payment evidence. Missing parties could hide a mass action, and partial payments cannot establish reciprocal flow or a sponsor/operator relationship. notes identifies the unread transactions; a failed history page identifies the unread seed scan. Both set truncated: true. Complete transactions still contribute signals, and gas sponsorship remains usable because it is read independently of balance changes.

The same truncated flag and notes accompany every wallet-edge format. graph_json includes them beside its nodes and edges; Mermaid and CSV results include a separate JSON text item, leaving the diagram or CSV data unchanged.

The small unpriced-coin grant check uses a bounded send window. If that window or any transaction’s balance changes exceed one page, the send cannot qualify as a targeted grant. A partial recipient count is never proof of a small send.