The machine economy has a history.
Machine Observer preserves dated observations from public machine-economy sources, developer registries and its own normalized snapshots. Historical gaps remain gaps. Revised source values remain auditable instead of being silently overwritten, so current measurements can be interpreted against the record that preceded them.
Protocol activity and developer adoption before Machine Observer existed.
The initial backfill combines source-published protocol series, official npm and PyPI package chronology, and Machine Observer measurements recovered from its own retained snapshot files. Scheduled collection extends the record forward, storing values when they change so an unchanged stretch is recorded as unchanged rather than as a gap.
x402 monthly transactions
Monthly transaction counts reported by Agent Economy from its underlying Dune series.
- Source
- Agent Economy
- Period covered
- 2025-10-01 to 2026-09-30 · 12 monthly observations
- Unit
- transactions
- Provenance
- SOURCE_SERIES
- Methodology
- Source-published monthly x402 transaction series preserved without interpolation.
- Limitations
- The most recent period is still accumulating and is shown separately from completed periods. Counts transactions, not economic weight; a month of many tiny automated calls can outweigh a month of larger settlements. Gaps are gaps: Machine Observer does not interpolate or zero-fill missing periods.
x402 monthly settled value
Monthly settled value reported by the same source.
- Source
- Agent Economy
- Period covered
- 2025-10-01 to 2026-09-30 · 12 monthly observations
- Unit
- USD
- Provenance
- SOURCE_SERIES
- Methodology
- Source-published monthly x402 settled-value series preserved without interpolation.
- Limitations
- The most recent period is still accumulating and is shown separately from completed periods. Settled value depends on which facilitators and chains the upstream query covers; it is not a measure of the whole agent-payment economy. Gaps are gaps: Machine Observer does not interpolate or zero-fill missing periods.
x402 settled USD per transaction
A Machine Observer derived ratio using monthly settled value divided by monthly transaction count.
- Source
- Machine Observer
- Period covered
- 2025-10-01 to 2026-09-30 · 12 monthly observations
- Unit
- USD/transaction
- Provenance
- DERIVED
- Methodology
- Monthly source-published x402 settled USD divided by monthly source-published x402 transaction count.
- Limitations
- The most recent period is still accumulating and is shown separately from completed periods. A period ratio, not a typical purchase. It inherits every limitation of both underlying series and moves when either one is revised. Gaps are gaps: Machine Observer does not interpolate or zero-fill missing periods.
Olas weekly transactions
A non-x402 historical series that helps Machine Observer compare machine-economy activity across protocols.
- Source
- Agent Economy
- Period covered
- 2025-09-01 to 2026-08-30 · 52 weekly observations
- Unit
- transactions
- Provenance
- SOURCE_SERIES
- Methodology
- Source-published weekly Olas transaction series preserved without interpolation.
- Limitations
- Olas and x402 count different things on different rails. Comparing their shapes is reasonable; comparing their absolute levels is not. Gaps are gaps: Machine Observer does not interpolate or zero-fill missing periods.
Registered source availability
How much of the Machine Observer source registry answered as healthy or degraded across each day\u2019s scheduled health checks.
- Source
- Machine Observer source health
- Period covered
- 2026-09-02 to 2026-09-03 · 2 daily observations
- Unit
- percent
- Provenance
- DERIVED
- Methodology
- Mean of (healthy + degraded) / total across every Machine Observer source-health sample recorded during the UTC day.
- Limitations
- Measures whether Machine Observer could reach and parse a source, not whether the source was correct. A source can be reachable and still publish a stale number. Gaps are gaps: Machine Observer does not interpolate or zero-fill missing periods.
Median source response latency
Median response time across every registered source checked during the day.
- Source
- Machine Observer source health
- Period covered
- 2026-09-02 to 2026-09-03 · 2 daily observations
- Unit
- milliseconds
- Provenance
- ARCHIVED
- Methodology
- Median of every per-source latency measurement recorded across the UTC day. A median, not a service-level guarantee.
- Limitations
- A median across heterogeneous sources measured from one network location; it is not a service-level guarantee for any individual source. Gaps are gaps: Machine Observer does not interpolate or zero-fill missing periods.
Historical claims should say how we know.
Provenance classes describe the evidence path. They are labels, not guarantees of correctness.
DIRECT
Reconstructed from primary machine-readable or raw event sources. None yet: the ERC-8004 registry-event adapter is built but stays disabled until Machine Observer can publish a verified registry address and event topic.
SOURCE_SERIES
Historical time series published by an external source, preserved with source and upstream references.
ARCHIVED
Recovered from version history, releases, reports or retained snapshots — including Machine Observer snapshot files stored in the public repository before this record existed.
DERIVED
Calculated by Machine Observer from underlying historical observations.
History is a first-class machine-readable surface.
Filter by metric, source, protocol, category, provenance class, date range and granularity. The API returns the latest observed revision by default; request revisions explicitly when auditing source changes.
GET /api/v1/history/observations?metric=x402.transactions&granularity=monthquery a normalized economic seriesGET /api/v1/history/observations?metric=developer.npm.x402-core.downloads&granularity=dayquery a developer-adoption seriesGET /api/v1/history/metricsdiscover historical metrics and date coverageGET /api/v1/history/sourcesinspect historical source coverage and provenanceGET /api/v1/history/latestthe current value of every historical metricGET /api/v1/history/contextbaselines, trend and regime labels derived from the recordRecord generated 2026-09-04T20:04:18.444Z · latest retained period 2026-09-30.