Satoshi Forum Archive

A reproducible, searchable archive of every forum post attributed to Satoshi Nakamoto in the pinned Satoshi Nakamoto Institute dataset.

Corpus

• 543 attributed Satoshi posts
  • 539 BitcoinTalk posts
  • 4 P2P Foundation posts
• 3,845 total messages preserved as full thread context
• 260 discussion threads
• Dates, subjects, authors, original URLs, raw HTML, readable text, and checksums

Attribution caveat

The corpus contains 542 contemporaneous posts from 2009–2010 and one disputed later entry: the March 7, 2014 P2P Foundation message, “I am not Dorian Nakamoto.” It appeared through Satoshi's old account but was not cryptographically signed. This archive preserves it because the upstream dataset includes it, while labeling its attribution disputed.

“Complete” here means complete for posts attributed to Satoshi in the pinned upstream dataset. It cannot prove that no unknown alias, deleted message, private correspondence, or otherwise lost post exists.

Browse

• Website: jonnyace.github.io/satoshi-forum-archive — local search, source filters, post pages, and full threads
• ALL_POSTS.md — all 543 attributed posts in one file
• posts/ — one Markdown file per attributed post
• threads/ — complete discussion threads, including other participants
• index.csv — spreadsheet-friendly index
• data/satoshi-posts.json — normalized Satoshi-only data with original HTML
• sources/forum_posts.json — exact pinned upstream dataset
• manifest.json — provenance and corpus counts
• docs/captures/manifest.json — observed live-page and Wayback coverage, retrieval times, failures, and blob metadata
• SHA256SUMS — integrity hashes for the corpus, source captures, and generated site

Source-page preservation

For each of the 543 attributed-post URLs, the capture manifest records two distinct checks:

1. Live URL: the current HTTP response observed during capture. This is evidence of what survived at retrieval time, not a claim that the page is unchanged from 2009–2010.
2. Wayback: the earliest distinct successful HTML snapshot returned by the Internet Archive CDX API for that exact URL.

Successful response bodies are checked in under docs/captures/blobs/ as deterministic gzip files named by the SHA-256 of their uncompressed bytes. The public site never renders these untrusted historical pages directly; it offers them as downloads and renders only escaped normalized text.

Network coverage is necessarily incomplete. A missing or failed response does not prove that a page or snapshot never existed. The manifest retains observed HTTP failures, timeouts, and explicit no-snapshot results rather than silently substituting another source.

The checked-in manifest currently verifies 539 live pages (all BitcoinTalk targets), 540 Wayback copies, and 1,079 unique response bodies across 543 targets. The four old P2P Foundation live URLs now return HTTP 400/404 responses; three exact P2P comment URLs returned no HTML snapshot in the observed CDX results. Those gaps remain explicit.

Reproduce and verify

Requires only Python 3.11 or later and the standard library.

python3 -m unittest discover -s tests -v
python3 scripts/build_archive.py
python3 scripts/build_site.py
python3 scripts/build_checksums.py
python3 scripts/verify_archive.py

The archive and site builders use only checked-in inputs and should leave the Git working tree unchanged. The builder refuses to run if the vendored upstream source does not match its pinned SHA-256 hash.

Refreshing network captures is a separate, intentionally non-deterministic operation:

python3 scripts/capture_sources.py --workers 2 --min-interval 0.75 --retry-errors

Live sites and the Wayback service can change between runs. Once response bytes and their manifest are checked in, however, their hashes and all derived site output are reproducibly verifiable offline.

Provenance

• Upstream: NakamotoInstitute/nakamotoinstitute.org
• Pinned commit: c412eb84a1d97df9114a6bf4dea300820965c0a7
• Source file: server/data/forum_posts.json
• Source SHA-256: 110bb3b42867d410ca8ee888810d3ffce73582b8c1987b103e8c69488a68f10f

Original URLs are retained even where a page may now be unavailable. The archive does not silently replace original text with edited prose: exact upstream HTML remains checked in alongside the readable rendering.

Licensing and attribution

The repository uses multiple licenses; no single license applies to every file. The archive tooling and original repository structure are distributed under the GNU Affero General Public License v3.0. The archived messages remain the work of their original authors. The Satoshi Nakamoto Institute identifies its site content as CC BY-SA 4.0; see LICENSE, CONTENT-LICENSE.md, and NOTICE for the separation between archive code, upstream compilation, original messages, and the transformations made here.