BIP-110 Stratum signalling dataset
==================================

Coverage
--------

Observed from 2026-08-04T04:50:03.476566925Z through
2026-08-11T23:59:57.293443857Z across 25 public pool endpoints.

The dataset contains 758,399 received Stratum V1 mining.notify jobs. Of those,
122,208 carried the versionbits top pattern 001 with bit 4 set. There were 18
observed transitions between signalling and non-signalling states.

Files
-----

bip110-stratum-signalling-summary-20260804-20260811.json
  Canonical summary used by the history page. It contains per-endpoint,
  per-date counts, observation bounds, flip timestamps, totals, and the exact
  BIP-110 predicate.

bip110-stratum-signalling-jobs-20260804-20260811.csv.gz
  Sanitized gzip CSV with one row per received mining.notify job.

SHA256SUMS
  SHA-256 digests for both data files.

Job-level CSV schema
--------------------

observed_at_utc
  Timestamp at which the listener received the job.

source_id
  Public pool endpoint label in pool/mode form.

version_hex
  Eight-character block-header version supplied in mining.notify.

signals_bip110
  true when (version & 0xe0000000) == 0x20000000 and bit 4 is set;
  false when it is not; blank only when the version field is malformed.

clean_jobs
  The mining.notify clean_jobs boolean; blank if the field was absent or not
  a boolean.

Sanitization
------------

The job-level export deliberately excludes authorization frames, payout
addresses, job IDs, coinbase template material, merkle branches, connection
identifiers, collector sequence numbers, and pool network addresses. The raw
captures are not published.

Integrity and reproduction
--------------------------

Every source .jsonl.zst archive was verified against its matching SHA-256
checksum file before decompression. The release was produced by
scripts/stratum_history_dataset.py in the bitcoin-forks.org source repository.
The gzip stream is deterministic: rebuilding from the same verified inputs
produces identical bytes.

Interpretation
--------------

This records jobs advertised by the monitored endpoints. It does not measure
hashrate, prove that any pool enforced BIP-110, or cover miners that construct
their own templates, including DATUM miners. Multiple jobs may describe the
same template state, so job counts are not votes and must not be treated as a
pool-share statistic.

Licence
-------

To the extent possible under law, this dataset is dedicated to the public
domain under CC0 1.0:
https://creativecommons.org/publicdomain/zero/1.0/