# Research methodology 1.0.0

This document freezes the publication rules used by Sentario Research for the
aggregate release `bitcoin-address-graph-v1.0.0`. It crosswalks to the already deployed graph
artifact `v1.0.0-draft.2` and standings enrichment
`v1.0.0-draft.2-standings.1`; it does not rename or mutate those serving
artifacts.

## Population and snapshot

- Source: `bigquery-public-data.crypto_bitcoin`.
- Snapshot: Bitcoin block 960,000.
- Population: 1,530,808,114 distinct normalized address
  records through the snapshot.
- Eligible directed paths: 1,364,936,252.
- No eligible path under the selected terminal rules: 165,871,862.

An address record is not a wallet, person, organization, or UTXO. Bitcoin does
not preserve protocol-level identity for individual satoshis after transaction
inputs are combined.

## Directed graph rule

The bipartite graph permits only
`address --spends--> transaction --pays--> address`. Address degree is half
the bipartite distance from the frozen anchor set. The graph does not traverse
backward from an output into a transaction's inputs.

## Anchor set

The frozen membership contains 28,583 coinbase-output addresses selected by the
height-gated, byte-reversed nonce predicate documented in
`methodology/methodology.json`. These outputs are described as
**Patoshi-consistent**, never as proven Satoshi-mined outputs. The deployed raw
membership hash is `sha256:e8b3b6979ef9d4777e7d352b6963b5c6067f4645eb2e42e013a5b01408d01546`; the canonical sorted unique
membership hash is `sha256:e8b3b6979ef9d4777e7d352b6963b5c6067f4645eb2e42e013a5b01408d01546`.

## Terminal hubs

An address with more than 50,000 endpoints, or a member of the frozen curated
terminal list, may be reached but not traversed. The curated terminal list is
the union of 41 addresses from active hub-labelled records and 13 retained
`hub_only_addresses`. Its canonical sorted unique hash is
`sha256:597ab88ae8a9333bc02722efd74a10f229bef3e3d1258da6c58f699ee5abb577`.

## Equal-length paths

Among shortest paths, the builder minimizes the latest transaction month along
the path, then uses stable transaction and parent-address dictionary IDs. The
full 1,530,808,114-row persisted-artifact audit found zero
degree differences. 506,285,603 connected-date
values changed; 505,294,457 became
earlier and 991,146 became later.
Every later value replaced a legacy path that violated the terminal rule.

The historical 50-path builder command is a deterministic operational sample,
not a statistically random audit. Its output bytes are not present in this
release, so no path-sample file is claimed here.

## Interpretation limits

Transaction-graph proximity does not establish ownership, identity, literal
coin ancestry, wallet membership, or contact with Satoshi Nakamoto. Batch
payments, CoinJoin, mixers, change outputs, address reuse, and the absence of
wallet clustering all limit causal interpretation.

The observed histogram extends to degree 25,349, but that maximum path has not
been independently inspected. It is therefore reported only as an artifact
boundary and is not interpreted as an economic or behavioral finding.