MOMachine Observer
Methodology v0.2

The source is part of the data.

Machine Observer is a meta-index, which creates a special failure mode: the same underlying activity can appear in several indexes. Our methodology therefore scores sources, preserves provenance and delays composites until overlap can be measured.

Source confidence

Source confidence is a reproducible quality weight, not a probability that every claim is true.

score =
30% methodology transparency
+ 25% direct observation
+ 20% machine readability
+ 15% historical depth
+ 10% independence

Non-negotiable rules

  • Unknown values are stored as null, never inferred for presentation.
  • Every publishable atomic claim carries a source ID, evidence URL and observation time.
  • First-party sources receive no bonus. LikenessIndex is explicitly penalized on independence because it shares ownership.
  • Several indexes derived from the same chain/query do not become independent votes merely because their websites differ.
  • A valid payment quote is not paid fulfillment. A transaction is not necessarily arms-length commerce. Registration is not validation.
  • Historical snapshots are append-only in normal operation; corrections create a new observation rather than silently rewriting prior evidence.
  • The Machine Economy Index remains unscored until sector normalization and overlap tests are defensible.
Current source weights

Same rubric, including our own properties

These values are transparent editorial judgments and will themselves be versioned as the system matures.

SourceMethodDirectMachineHistoryIndependentScore
Agent Economy0.940.9610.920.995
Graded0.94110.780.9294
TOLL·402 Endpoint Trust Index0.9710.350.90.9484
Agentic Payments Landscape0.940.70.450.880.9277
QuickNode ERC-8004 Explorer0.9610.950.840.8694
x402scan0.820.90.780.720.8282
mpp.best0.780.780.550.650.8472
LikenessIndex0.840.820.740.450.572
APIs.guru0.880.8210.980.9491
GPU Cloud Prices0.860.8610.580.8484
TokenGauge0.840.7810.750.8685

Automatic collection

A scheduled GitHub workflow fetches only sources with a stable public data surface that we have explicitly configured. It stores a timestamped snapshot, updates the compact history and leaves failed sources marked as failures rather than carrying the prior value forward as fresh.

Anti-double-counting

The future composite will group observations by underlying phenomenon and provenance before weighting. If Agent Economy and another dashboard both ultimately consume the same Dune query, they can corroborate presentation but not create two independent measurements of economic activity.