On-chain evidence

How a figure reaches the block it rests on, and where the chain is thin.

Evidence overview

Each card names its source. No coverage percentage — §十六 asks for a defined denominator first.

Verified payment eventsC06 · C09

539

11 published points

Active payment addressesC06

Addresses are not users; one party may hold several.

Chains coveredC01 · C09

100 · 5000

Data freshnessC09

2026-09-01 06:40

UTC — the newest block any published fact rests on

Evidence chain

Left to right: how a figure reaches the block it rests on. Each stage names the component that owns it, its state, what to cite, and how far it reaches. Open one for the detail.

C09Insight / MetricPartialmetric-v45 metrics · 3 with a series · 26 points

Payment eventspost_auth_funds_action_countObserved11no earlier revision holds more

Reversalsreversal_countNo data0394 points at metric-v3

Auth adjustmentsauthorization_adjusted_countNo data076 points at metric-v3

Active accountsactive_account_countObserved11no earlier revision holds more

Active payment addressesactive_payment_address_countObserved4no earlier revision holds more

C06Payment factPartial2 deployments9,100 committed batches · 2 of 4 deployments

A batch records that a block range was interpreted and committed. It answers how far coverage reaches — never how many facts exist, because a repair republishes a range and the republished facts are one row and two batch contributions.

15 committed batches overlap another. The per-deployment record is in Publication coverage below.

C05InterpretationPartial3 adapter revisions4 revision spans

Deployment	Adapter	Revision	ABI / mapping / parser	Batches	Blocks
gnosis-pay-gnosis gnosis-pay	gnosis-pay-gnosis	ur-mantle-v0.3	ur-mantle-v0.1 / ur-mantle-v0.1 / ur-mantle-v0.1	40	47,817,934 – 47,823,402
gnosis-pay-gnosis gnosis-pay	gnosis-pay-gnosis	gnosis-pay-gnosis-v0.2	gnosis-pay-gnosis-v0.2 / gnosis-pay-gnosis-v0.2 / gnosis-pay-gnosis-v0.2	7,045	47,823,403 – 48,021,665
ur-mantle-mainnet ur	ur-mantle-mainnet	ur-mantle-v0.3	ur-mantle-v0.1 / ur-mantle-v0.1 / ur-mantle-v0.1	46	31,420,289 – 99,600,000
ur-mantle-mainnet ur	ur-mantle-mainnet	ur-mantle-v0.4	ur-mantle-v0.1 / ur-mantle-v0.1 / ur-mantle-v0.1	1,969	99,600,001 – 100,056,858

C03Raw event / TxNot observed—not published here

The raw log, transaction and receipt are what every fact above was read from, and this terminal publishes no locator for them. C06 returns them through a more privileged query than the one behind this page, and raw payloads expire — so an identifier shown here could point at something already gone.

C02Finalized blockObservedfinalized2 chains · 2 deployments

Everything above is read at the finalized head, never at the latest block. C06 rejects a publication whose head policy is anything else, so this value is a property of the data and not a setting on this page — a different one appearing here would mean a batch got past a check.

Publication coverage

Every registered deployment and what C06 committed

etherfi-cash-optimismchain 10

nothing published

etherfi-cash-scrollchain 534352

nothing published

gnosis-pay-gnosischain 100

100%203,732 of 203,732 blocks read, contiguous

ur-mantle-mainnetchain 5000

23%16,395,859 of 68,636,570 blocks read, contiguous

Deployment	Chain	Activity	Batches	Blocks	Contiguous	Republished	Data through
etherfi-cash-optimism	10	Active	0	—	—	0	—
etherfi-cash-scroll	534352	Dormant	0	—	—	0	—
gnosis-pay-gnosis	100	Active	7085	47,817,934 – 48,021,665	48,021,665	0	2026-09-01 06:40 UTC
ur-mantle-mainnet	5000	Active	2015	31,420,289 – 100,056,858	47,816,147	15iCommitted batches whose block range overlaps another — a repair republished the range. Facts are not double-counted; batch fact counts would be, which is why this terminal publishes no fact total.	2026-09-01 06:21 UTC

Metric catalog state

Whether history exists at a revision this page does not read

Metric	State	Points	Earlier revision
Payment events post_auth_funds_action_count	Observed	11 metric-v4	none
Reversals reversal_count	No data	0 metric-v4	394 pts authorization_reversed_count @ metric-v3
Auth adjustments authorization_adjusted_count	No data	0 metric-v4	76 pts authorization_adjusted_count @ metric-v3
Active accounts active_account_count	Observed	11 metric-v4	none
Active payment addresses active_payment_address_count	Observed	4 metric-v4	none

Evidence by payment function

The published assessment distribution per catalog function

Code	Function	Systems assessed	Evidence distribution
F01	Payment account	4/4	3 fact 1 private only
F02	User funds	4/4	3 fact 1 private only
F03	Account funding	4/4	4 inference
F04	Authorization request	4/4	3 disproven 1 inference
F05	Authorization decision	4/4	2 disproven 1 fact 1 private only
F06	On-chain payment control	4/4	2 fact 2 inference
F07	Post-authorization funds action	4/4	4 fact 1 private only
F08	Funds hold	4/4	2 fact 1 private only 1 disproven
F09	Authorization adjustment	4/4	2 disproven 1 inference 1 unknown
F10	Clearing instruction	4/4	3 disproven 1 inference
F11	Pre-settlement funds movement	4/4	3 fact 1 unknown 1 disproven
F12	Final network settlement	4/4	3 disproven 1 inference
F13	Reversal	4/4	2 disproven 1 fact 1 private only
F14	Authorization expiry	4/4	2 disproven 1 private only 1 unknown
F15	Refund	4/4	2 disproven 1 private only 1 unknown
F16	Dispute / chargeback	4/4	1 inference 1 private only 1 unknown 1 disproven
F17	Fees	4/4	1 inference 1 private only 1 unknown 1 disproven

§七 also asks for verified events, share of total and an evidence-link coverage ratio per function. C09 publishes its series deployment-wide rather than per function, and this terminal has no evidence-link projection, so those three columns are absent rather than estimated. §七 itself says to hide a trend rather than mock one.

Evidence by source

What every published range was read under — counted, never shared out

Adapter2

Adapter revision3

ABI revision2

Mapping revision2

Parser revision2

published ranges4

One range is read under all four revisions at once, so these do not sum to the range count and are not a share of it. §九 asks for the dependency set, not a distribution.

Recent verified events

Not yet publishedThis module needs time, chain, function, event type, value and transaction hash, per event. C10's public read model publishes no per-event projection: the snapshot behind this page carries scope, protocols, functions, capability groups, metrics, capabilities and methodology, and none of those is per event or per address. The gap is in C10, not in this page's layout.

Recent raw transactions

Not yet publishedThis module needs transaction hash, from, to, value and chain status, per transaction. C10's public read model publishes no per-event projection: the snapshot behind this page carries scope, protocols, functions, capability groups, metrics, capabilities and methodology, and none of those is per event or per address. The gap is in C10, not in this page's layout.

Decoded activity

Not yet publishedThis module needs one raw event beside the payment fact it was interpreted into. C10's public read model publishes no per-event projection: the snapshot behind this page carries scope, protocols, functions, capability groups, metrics, capabilities and methodology, and none of those is per event or per address. The gap is in C10, not in this page's layout.

Trust layer

Rules passed, as counts — §十一 forbids a percentage without a methodology

Read at finalized head2/2

Assessed against the catalog4/4

Published at the pinned revision3/5

Traceability is not shown. §十一 defines it as the share of published facts traceable to a transaction, log or block, which needs the per-event projection C10 does not publish. A figure here would be the invented percentage §十一 refuses.

Evidence examples

Not yet published§十二 wants one representative verified fact per payment function, opening to its raw event. Both halves need C10's per-event projection, which is not published.

Evidence insights

Not yet published§十 requires every insight to carry a fact, a window, a scope and an evidence link. No component here produces deterministic rules or a research conclusion, and no evidence-link projection is published — so an insight would be missing the one part that makes it checkable.

Not published here

observed_amount_totals

Not observed iAmounts are reported per denomination with no total. A cross-currency figure would need an exchange rate, and SpendGraph does not observe one on-chain.

industry_signals

Planned iThe product design has a signal feed. No component produces one, so there is no panel here rather than an empty panel.

protocol_ranking

Not observed iForbidden by the product specification: capability and coverage are described, systems are not ranked.

market_share

Not observed iThe observed population is SpendGraph's coverage, not the market. A share would need a denominator nobody here has measured.

identity_capability_group

Research required iThe frozen function catalog defines no identity function. The registry's identity vocabulary is identity_continuity, which asks a different question: whether an account survives a change of holding address.

credit_capability_group

Research required iCredit is carried as a funding model on an assessment row and as rail roles, not as a catalog function. There is nothing for a protocol to be assessed against.

protocol_category

No data iNo registration pack declares a category, and the registry model has no field for one. A category shown here would be this site's invention rather than a reading of a pack, so the comparison page keeps the slot and names the gap.

entity_directory

Not observed iThe public read model carries no account reference at all, so entities are counted and never listed.

Methodology

Every count here is SpendGraph's observed coverage. It is not a total for the on-chain payment market, and no denominator for that market is published anywhere on this site.
Counts are read from C09's published series at a pinned catalog revision. This page performs no count of its own — a second implementation of a published metric is a second thing that can be wrong.
Publication batches answer how far coverage reaches, never how many facts exist. A repair republishes a range, so summing batch fact counts double-counts the repaired part: on 2026-08-21 that sum was 85,021 against 81,034 actual facts, with zero retractions.
A registration status describes how far SpendGraph's own review of a pack has got. It is not an assessment of the protocol, and every review state is compatible with a deployment that stopped carrying payments months ago — which is why activity is a separate column.
The payment function catalog is frozen at seventeen entries. A new protocol term is mapped onto an existing function rather than growing the catalog, so a function absent from a protocol's assessment means nobody has assessed it — not that it does not exist.
The six capability groups are defined by this site alone, as a presentation over the seventeen. Two of them — Identity and Credit — have no catalog function, so no protocol can be assessed against them and they are shown as such.