Every row carries an evidence grade and, where one exists, a command that regenerates
it. Rows we cannot verify are shown too, marked NOT HELD — because a timeline showing
only what was found looks complete when it is not.
This timeline is incomplete, and says so on purpose.
It currently holds 68 events. It is a deliberate subset of unpublished research rather than a bulk export — each row has to acquire a grade, a source and a reproduction path before it can appear here. The material behind it is not published, and its extent is deliberately not characterised: a figure nobody can verify is not evidence.
Absence of a row here is not evidence that nothing happened on that date.
Applying the same standard to ourselves that we apply to everyone else — literally, since
8 of these rows are about a second Bitcoin, mined in
2026, and they are graded by the same table and carry a NOT HELD of their own.
2009 pairADJUDICATEDNOT HELD
An August 2008 draft existed — title and abstract survive; the FILE does not
A pre-release draft was linked, not attached, to Adam Back on 20 August 2008 and to Wei Dai on 22 August 2008. The judgment quotes the email to Back — the Hashcash citation at [5] and the upload.ae link — and Back's reply pointing to b-money. The file is not located.
reproduceno mechanical reproduction — this row rests on a document
caveatsNOT LOCATED. What survives in the public record is the citation line and the link; the title and abstract reported in evidence at trial are not reproduced here.
2009 pairSERVER-DB
bitcoin.org is registered
The domain bitcoin.org is created, ten weeks before the whitepaper is announced.
caveatsA registrar's own database row. It says nothing about who registered it.
2009 pairADJUDICATED
The earliest known Satoshi communication of any kind
Satoshi writes to Adam Back querying the Hashcash citation, ten weeks before the paper is public; Back replies the next day; Satoshi writes to Wei Dai on 22 August.
reproduceno mechanical reproduction — this row rests on a document
caveatsThe sending address is in evidence given at trial and not published. The vistomail → gmx sequence that follows rests on the public mailing-list archives and the March 2009 paper.
2009 pairARCHIVE-POS
The file id Satoshi linked is bracketed to within days by an unrelated crawl
Satoshi emailed a LINK, not an attachment, to Adam Back on 20 Aug 2008 and Wei Dai on 22 Aug 2008: upload.ae/file/6157/ecash-pdf.html. That host's file ids are sequential, and the Internet Archive crawled it through 2008 for unrelated reasons. Ids 6145, 6160 and 6161 - immediately either side of 6157 - were all first captured on 23 August 2008, placing 6157's upload on or before that date.
reproducecurl -s 'http://web.archive.org/cdx/search/cdx?url=upload.ae&matchType=domain&output=json&from=2008&to=2010&fl=timestamp,original&limit=40000' # use a HIGH limit; at 8000 the neighbours truncate away
caveatsCorroborates TIMING only. It does not recover the file, say anything about its contents, or identify who uploaded it. Its value is that the bracketing party had no interest in the question: the Internet Archive crawled an Emirati file host in 2008 for its own reasons. Note also that it captured 6145, 6160 and 6161 that day and did not capture 6157 - consistent with an unlisted upload whose URL travelled only by private email, though a crawl gap has many innocent causes. The file itself was not archived and structurally could not be: the download sat behind a form POST, which crawlers do not submit.
2009 pairCAPTURE
The November 2008 pre-release carries its own genesis nTime
The pre-release codebase hard-codes a genesis block time of 2008-09-10 18:02:08 — nearly four months before the live genesis block.
reproduceno mechanical reproduction — this row rests on a document
caveatsAn integer the code sets about itself — no headline, no external commitment, so it anchors nothing outside the file. Contrast the live genesis, which embeds a newspaper front page and therefore cannot predate it.
2009 pairCAPTURE
The 3 October 2008 pre-release draft exists
A pre-release whitepaper PDF is created; we hold the bytes, and its two /ID halves are equal, meaning it was not re-saved after creation.
reproducecurl -sL 'https://web.archive.org/web/20140320135003id_/https://bitcoin.org/bitcoin.pdf' -o c.pdf; sha256sum c.pdf # any copy of the 3 Oct 2008 draft must be 427c63b364c6db91…
caveats/CreationDate D:20081003134958-07'00'. The offset is the producing machine's setting, not a location.
2009 pairSERVER-DB
The SourceForge account nakamoto2 is created
User ID 2238460 joins SourceForge, five weeks before registering the Bitcoin project.
caveatsServer-written and bracketed by neighbouring account IDs, so it cannot have been backdated by the holder.
2009 pairSELF
The PGP key of the 2008–2010 Satoshi Nakamoto asserts its own creation date
The key served as Satoshi_Nakamoto.asc from bitcoin.org — DSA primary with an ElGamal subkey, user ID "Satoshi Nakamoto <satoshin@gmx.com>", fingerprint DE4E FCA3 E1AB 9E41 CE96 CECB 18C0 9E86 5EC9 48A1 — carries a self-asserted creation timestamp of 2008-10-30 18:19:19 UTC in its own key packet, and the Version header reads GnuPG v1.4.7 (MingW32).
SELF is the weakest grade on this page, and this row is the reason it exists. A PGP creation timestamp is written by whoever generated the key, from their own clock, and can be set to anything. It is NOT evidence that the key was made on that date. What is independently checked is only this: a key bearing that timestamp was served from bitcoin.org, was captured there, and is byte-identical to the copy held in a separate third-party archive.
reproduceFetch the .asc above, strip the armour, base64-decode, and read the OpenPGP packets: tag 6 (public key) and tag 14 (subkey) each begin with version byte 0x04 followed by a 4-byte big-endian creation time. Both decode to 1225390759 = 2008-10-30 18:19:19 UTC. The v4 fingerprint is SHA-1 over 0x99 || 2-byte length || packet body.
caveatsThe user ID already carries satoshin@gmx.com — the address on the March 2009 whitepaper — rather than the satoshi@vistomail.com of the August 2008 emails. Timing against the 31 October 2008 announcement is deliberately NOT stated as a finding here: one side of that comparison is self-asserted, so the interval is not a measurement.
2009 pairARCHIVE-POS
The whitepaper is announced to the cryptography list
satoshi@vistomail.com posts 'Bitcoin P2P e-cash paper' to cryptography@metzdowd.com.
caveatsThe +0800 offset is the mail provider's server clock, not a location — see the +0800 verdict. Message-ID is CHILKAT-MID-...@server123, a server-side library.
2009 pairPARTY-RELEASED
The 183,697-byte draft is in third-party hands by 8 November 2008
A correspondent states he received the 183,697-byte bitcoin.pdf on this date — the same byte-length as the pre-release draft we hold.
reproducesha256sum <your copy of the 3 Oct 2008 draft> # the hash StealthMonger published and Dillinger confirmed: 427c63b364c6db914cf23072a09ffd53ee078397b7c6ab2d604e12865a982faa
caveatsIndependent provenance for the pre-release draft, from people who held it in 2008 and published a matching hash years later.
2009 pairSERVER-DB
The Bitcoin project is registered on SourceForge
Account nakamoto2 registers the project; s_nakamoto and hal are added later.
reproduceno mechanical reproduction — this row rests on a document
caveatsCAUTION: SNI's .tgz is labelled 'Compressed by SNI' — a REPACKAGING, not original bytes. It will not hash to anything Satoshi produced, and is not to be cited as an original artifact. Their .rar carries a published MD5; that is the one to check against.
2009 pairARCHIVE-POS
The venue closes — and the 52-day silence is explained
Moderator Perry E. Metzger ends the Bitcoin discussion 4h19m after Satoshi's last post. The silence that follows is a moderated list being moderated, not a withdrawal.
caveatsSourceForge's own feed shows two accounts under one hand's control. It does not say whose hand.
2009 pairADJUDICATEDNOT HELD
A whitepaper version exists in the court record that is in no public hands
The judgment records that Nicholas Bohm, a retired solicitor who corresponded with Satoshi shortly after the January 2009 release and who died shortly before the trial began (¶271.9), provided a version of the whitepaper he had downloaded in January 2009, which an expert witness (Mr Madden, ¶271.9) authenticated and which was used in the evidence as a control copy. It is not in public hands.
reproduceno mechanical reproduction — this row rests on a document
caveatsUntil 14 September 2026 this row gave a creation date to the second and two hashes taken from an expert appendix in circulation. That material is in evidence that is not published, and it is no longer reproduced here: the row states what the judgment records and nothing more. The file's date, hashes, size and text stay unstated on this site.
2009 pairADJUDICATED
The judgment records transfers of bitcoin from Satoshi to a 2009 correspondent
The judgment records (¶771), from the evidence of Nicholas Bohm, a retired solicitor who corresponded with Satoshi shortly after the January 2009 release and who died shortly before the trial began (¶271.9), that Satoshi made transfers of bitcoin to him.
reproduceno mechanical reproduction — this row rests on a document
caveatsWhich transactions they were is stated in correspondence given in evidence and not published; this row does not identify them, and the script that once located a payment from that correspondence was withdrawn on 14 September 2026. This row implies nothing about any holdings. The judgment does not date the transfers; the range is this timeline's own bound, the year in which the correspondence began shortly after the January 2009 release (¶271.9), not a finding.
2009 pairNONENOT HELD
No key has signed anything — the ceiling on all of this
No genesis-era or Patoshi private key has produced a verifying signature. Every identity fact on this timeline is statistical, archival, adjudicated or party-released. None is cryptographic.
evidence & reproduction
the absence itself; the judgment records that the 2016 demonstration presented a signature lifted from the public blockchain, derivable from the Satoshi/Finney transaction and the block 9 generation transaction (¶886–887) https://caselaw.nationalarchives.gov.uk/ewhc/ch/2024/1198
reproduceno mechanical reproduction — this row rests on a document
caveatsThis is a property of the world, not a gap in the research. Only a signature would change it. Recorded as a row so the timeline cannot be read as answering 'who'.
2009 pairCHAIN
The genesis block
Block 0 is mined, embedding The Times headline of 3 January 2009.
evidence & reproduction
block 0, hash 000000000019d6689c085ae165831e934ff763ae46a2a6c172b3f1b60a8ce26f the Bitcoin blockchain
reproduceany full node, or bigquery-public-data.crypto_bitcoin.blocks WHERE number = 0
caveatsThe only unforgeable line on this list. Forging it would require redoing the proof-of-work.
2009 pairARCHIVE-POS
Bitcoin v0.1 is announced
satoshi@vistomail.com posts 'Bitcoin v0.1 released' to the cryptography list.
reproducedownload the attachment from the bitcoin-list archive at satoshi.nakamotoinstitute.org and md5sum it (the recorded value is an MD5, labelled as such)
caveatsMailing-list attachments live in a URL namespace separate from message bodies and are in none of our text corpora. Of 129 archived attachments on that list, this is the ONLY one from 2009.
2009 pairCAPTURE
The released bitcoin.exe is linked — two days AFTER v0.1.0 was announced
The Windows binary in the archive distributed everywhere as bitcoin-0.1.0.rar carries PE TimeDateStamp 1231629360 = 2009-01-10 23:16:00 UTC. v0.1 was announced on 8 January, so this binary cannot be in the release of the 8th: the archive is v0.1.1.
caveatsA linker-written field. MinGW toolchain, and notably no build-path leak - nothing about the machine escaped into the binary. The 0.1.0 -> 0.1.1 delta is irc.cpp and serialize.h, both networking: main.cpp, main.h and script.cpp are dated 7 January and untouched, so every consensus rule and the genesis are unaffected. Only the label was wrong, and it was inherited from the ecosystem rather than read off the bytes.
2009 pairCAPTURE
Hal Finney tweets 'Running bitcoin'
The best-known third-party record of the network acquiring a second participant.
caveatsWidely attributed to Satoshi sending Hal Finney. The chain shows the transaction, not the names.
2009 pairSERVER-DB
The v0.1 release announcement on SourceForge — byline erased
SourceForge carries the full v0.1 release announcement, including the instruction that generated coins must wait 120 blocks to mature. A platform migration replaced the author with 'Anonymous'.
caveatsThe text is theirs; the attribution is gone, which is why no compilation folds it in. It independently corroborates the 120-block maturity we derived from main.cpp — COINBASE_MATURITY + 20 — in Satoshi's own words.
2009 pairCHAIN
Block 593 — a 50 BTC send with change
50 BTC is split into two 25s; one output returns to the input address, so this is a payment with change, not a split between two parties.
caveatsAn earlier reading of ours called this support for a 'first transfer of 25.00' account. The full transaction reverses the direction; this row is the corrected reading.
2009 pairCHAIN
The first transaction fees paid — a self-test
Block 2817 collects 2.01 BTC across three transactions (1.00, 1.00, 0.01), all with the same address on every input and output: someone moving 1 BTC to themselves three times.
reproduceno mechanical reproduction — this row rests on a document
caveatsDate offset is +0000 here against +0800 on vistomail — the offset is a setting on the sending side (see the +0800 verdict on the site), not a location. The Mozilla Message-ID's embedded Unix timestamp decodes to the Date header exactly, an authenticity check but NOT a location check.
2009 pairSELF
The P2P Foundation profile posts
Three posts appear on the P2P Foundation profile in February 2009. The profile's 'Male, Japan' field is self-entered.
reproduceno mechanical reproduction — this row rests on a document
caveatsThe two addresses ran IN PARALLEL for about a month: gmx privately from 3 February, vistomail publicly to 4 March. A clean single-changeover story is wrong.
2009 pairCAPTURE
The canonical whitepaper is created
The whitepaper as served today is created. We hold its bytes from three independent sources, all identical.
reproducepython verify/whitepaper_from_chain.py out.pdf; sha256sum out.pdf # the script lives in github.com/original-bitcoin-laboratory/bitcoin-whitepaper; carves the canonical out of block 230009 -> b1674191a88ec5cd…
caveatsThe judgment in COPA v Wright records (¶320) the court's control copy of this file as hash-identical to a SourceForge web-archive capture of 28 November 2009; that capture hashes to our bytes. Four independent paths, one hash.
2009 pairCHAIN
A 2009 block that collected a fee
Block 14047 pays a coinbase of 50.10 — one of only eight blocks in all of 2009 that collected any fee.
caveats[statistical], not [cryptographic] — the miner signed nothing. The coinbase address received that one coinbase and nothing else, and was emptied 28 Dec 2011.
2009 pairCHAIN
A 0.01 BTC payment, sender unidentified
The only payment of any size in a three-day window is exactly 0.01 BTC.
caveatsThe transaction is identified; THE SENDER IS NOT. The sending address traces to block 11,583, which is not in the Patoshi cluster. We looked for corroboration and did not find it.
2009 pairSERVER-DB
Martti Malmi commits to the repository
sirius-m commits revision 2 — the first contributor other than Satoshi.
The canonical git history carries Satoshi's 2009–2010 commits twice
bitcoin/bitcoin was converted from SourceForge SVN. Between its first commit (30 Aug 2009) and the last SVN-trailer commit (17 Dec 2010), 132 of its 341 non-merge commits exist as a same-first-line pair, of three kinds: 121 with the git-svn-id trailer on one copy and not the other, 10 with the trailer on both (a contiguous Subversion run in September 2010), and 1 with it on neither; the first line 'misc' occurs eight times besides. In the 121 one-trailer pairs the non-trailer copy is dated the same day or up to 8.09 days later, under the author strings s_nakamoto (72), Satoshi Nakamoto (25), the literal option text --author=Satoshi Nakamoto (7), sirius-m (12) or Gavin Andresen (5). 124 of the 132 pairs are the same change; 8 differ by a few lines. The doubling starts at the root: two parentless 'First commit' objects, outside the 341 non-merge commits and counted apart from the 132 pairs.
census output of the run of 2026-09-20 (the three pair types, recounted after an adversarial review) satoshi-onchain/data/github-history-census-2026-09-20.json
reproducepython satoshi-onchain/verify/github_history_census.py --out census.json (add --stats to compare each pair; needs GITHUB_TOKEN)
caveatsWhat this establishes: a date or hash cited from the non-trailer copy differs from the SVN copy by up to eight days, so a citation must say which lineage it reads; this project's SVN-derived dates (2009-10-21-svn, 2010-12-15-last-action) come from the SourceForge log itself and are unaffected. The trailer copies carry server-assigned SVN revision numbers; the non-trailer copies' dates are author-supplied. It does not establish how the second lineage arose, or anything about identity: the strings 'Satoshi Nakamoto' and 'satoshin@gmx.com' are author fields anyone can set. GitHub's history can be rewritten by the repository's owners, so the run is dated.
2009 pairPARTY-RELEASED
Recovered activity from the lost forum — October 2009
Four dated threads from the un-archived bitcoin.sourceforge.net forum survive because Satoshi reposted them verbatim to BitcoinTalk, preserving their original timestamps and usernames: 1 Oct 14:12, 8 Oct 05:49, 15 Oct 19:58 and 22 Oct 02:36 UTC.
reproduceno mechanical reproduction — this row rests on a document
caveatsThe old timestamps are QUOTED TEXT in a repost, not server records -- graded accordingly. Only the threads Satoshi chose to carry across survive, so this is a curated sample selected by the subject, not a representative one. The preserved date format 'Thu 01 of Oct, 2009 (14:12 UTC)' is TikiWiki's, which is evidence about what the lost board ran.
2009 pairSERVER-DB
The first of 164 SVN commits
s_nakamoto begins committing to SourceForge SVN; commit times are written by the server.
caveatsThe densest record of Satoshi working that exists off-chain.
2009 pairNONENOT HELD
Satoshi's SourceForge tracker comments — THE PREMISE IS UNSUPPORTED
Other archives list Satoshi's SourceForge tracker comments as a known omission. SourceForge's own public API shows the project has no tracker mount at all, and no crawl contains one.
CDX: 6,000 URLs under sourceforge.net/tracker/ reference Bitcoin zero times; modern /bugs and /feature-requests namespaces return zero archived URLs https://sourceforge.net/projects/bitcoin/
sourceforge.net/rest/p/bitcoin lists 8 tools — support, summary, reviews, code, mailman, files, news, activity — and no bugs/tickets/tracker mount https://sourceforge.net/projects/bitcoin/
caveatsRecorded as a gap because the claim circulates, but the evidence points at there being nothing to recover rather than something lost. CAVEAT: this is the CURRENT migrated tool set; a tool used in 2009 and later removed would not appear, so this lowers the probability sharply without reducing it to zero.
caveats539 posts are publicly enumerable; the profile counts 575, so 36 are counted but not listed.
2009 pairADJUDICATEDNOT HELD
The pre-BitcoinTalk forum — SATOSHI'S OWN VENUE, NOT ARCHIVED
Satoshi ran a web forum at bitcoin.sourceforge.net/boards/ before launching the successor forum on 22 November 2009, naming the old URL in their own announcement. No capture of it survives.
CDX sweep of the whole domain, all status codes (14 September 2026): 616 capture rows; 79 of them request /boards/ and every one is a 404 or 403, the earliest on 2011-06-27 — the archive asked for the forum and held nothing. 27 distinct URLs have a status-200 capture; none is under /boards/. https://web.archive.org/web/2009*/bitcoin.sourceforge.net*
reproducehttp://web.archive.org/cdx/search/cdx?url=bitcoin.sourceforge.net*&output=json&fl=timestamp,original,statuscode (unfiltered; count the rows whose URL contains /boards/ and read their status codes). Add &filter=statuscode:200&collapse=urlkey to reproduce the 27 URLs with content.
caveatsA Satoshi venue absent from every compilation we have checked. NOT ARCHIVED is not the same as destroyed. The remaining route is reconstruction from the 'Repost:' threads they carried across. The reproduce query is deliberately unfiltered: a status-200 filter would hide the 79 /boards/ rows, and those rows are the evidence that the archive requested the forum and found nothing. Row counts can grow as the archive adds captures; the figures are as of the date stated.
2009 pairSERVER-DB
nakamoto2 drops off the SourceForge project, and the paper's folder is renamed
Between 28 November 2009 and 8 January 2010 three fields on the Bitcoin SourceForge project page changed: the owner list went from 'nakamoto2, s_nakamoto' to 's_nakamoto' alone, the development stage moved Alpha to Beta, and the whitepaper's folder was renamed from 'Research Paper' to 'Design Paper'.
caveatsRecords what the page said, not who changed it or why. A name leaving an owner list is a page edit - it does not establish that the account was deleted or abandoned, and Satoshi's own last machine-verifiable activity is separately dated nearly a year later. The three changes share a 41-day bracket, which is consistent with one editing session and does not prove one. The window cannot be narrowed further: no capture of the files page exists between those two dates.
2009 pairCAPTURE
The earliest surviving capture of the whitepaper as served
The Internet Archive captures bitcoin.org/bitcoin.pdf; the bytes are identical to the file served today.
caveatsCloses the gap between the March 2009 creation date and the file anyone can fetch today.
2009 pairSERVER-DB
Satoshi names bit gold — on the forum, not in the paper
A forum post describes Bitcoin as an implementation of Wei Dai's b-money proposal and Nick Szabo's Bitgold proposal, with a link to the 2005 bit gold post. Neither whitepaper cites bit gold.
reproduceno mechanical reproduction — this row rests on a document
caveatsThe post was edited on 22 July 2010 and the surviving text is the edited one, so whether the bit gold sentence stood in the 20 July original is not settled by the forum's record. A citation is a statement about sources, not about authorship.
2009 pairCHAIN
The day 184 billion bitcoins existed
Block 74638 contains a transaction creating over 184 billion BTC through an integer overflow; the chain is forked away from it within hours.
reproducebigquery-public-data.crypto_bitcoin.transactions WHERE block_number = 74638
caveatsThe only time Bitcoin's supply rule has been broken in production.
2009 pairSERVER-DB
The alert system: one key that can message every node
An SVN commit adds a network alert message verified against one hard-coded public key. An alert could display a warning and, as committed, make the money-handling RPC commands of every node return an error.
reproducegit show 401926283a200994ecd7df8eae8ced8e0b067c46 # in a clone of bitcoin/bitcoin
caveatsNot a block-validity rule: it changes no node's view of the chain. Its power over RPC was removed on 12 December 2010 (next row).
2009 pairCHAIN
An example address printed in the 2009 Bitcoin client's UI begins receiving real coin
1NS17iag9jJgTHD1VXjvLCEnZuQ3rJED9L is a hard-coded placeholder in the Send Coins dialog of the January 2009 client (v0.1.0 src/uibase.cpp line 495, unchanged in v0.1.3). On the 2009 chain it received its first payment on 2010-09-10, has since received 102 payments totalling 900.85217683 BTC (as of 10 August 2026), and has not spent a single satoshi. Its largest single receipt is 777.00000000 BTC in block 104662, 2011-01-26.
This is an address Satoshi published as example text in a form label — not a wallet, and not a holding. WHO, IF ANYONE, HOLDS ITS PRIVATE KEY IS UNKNOWN, and nothing here suggests Satoshi did. The first payment arrives twenty months after the source was published, so the coin was sent by strangers who read the address out of the code. Published in source, funded by others, not spent.
reproducecurl -s https://blockstream.info/api/address/1NS17iag9jJgTHD1VXjvLCEnZuQ3rJED9L and read chain_stats: funded_txo_sum 90085217683, spent_txo_sum 0. Page the full history with /txs/chain and sort by status.block_height; the earliest confirmed receipt is at height 79021.
caveatsNot decidable for the November 2008 pre-release: that artifact is five files (main.cpp, main.h, node.cpp, node.h, readme.txt) and contains no UI source at all, so the string can be neither confirmed nor denied there. The earliest artifact carrying it is v0.1.0.
The alert key loses its power over RPC, under the message 'correction'
Seventeen minutes after a commit titled 'added some DoS limits, removed safe mode', a commit titled 'correction' deletes the line through which an alert put every node's RPC into safe mode, and renames the alert field that carried the RPC error. From here on an alert sets a status message only.
caveatsThe 2014 'I am not Dorian Nakamoto' post is 3 years 3 months later, unsigned, and binds to nobody. The precise claim is that the last VERIFIABLE action is this one.
2009 pairSERVER-DB
gavinandresen is added to the Bitcoin SourceForge project
The project's owner field reads 's_nakamoto' alone in the capture of 27 December 2010 and 'gavinandresen, s_nakamoto' in the capture of 21 January 2011. No capture of any sourceforge.net/projects/bitcoin URL exists between those dates, so the change is bracketed to 25 days and no tighter.
caveatsRecords a MEMBERSHIP CHANGE, not a handover. SourceForge admins can add members and the page shows the result, not the actor - it cannot establish that Satoshi added him. It also does not narrow Satoshi's departure: their last machine-verifiable action is separately dated to 15 December 2010, and this row adds nothing to that. The adjacency is recorded precisely because it is the kind of fact that becomes a narrative if left uncaptioned.
2009 pairCAPTURE
bitcoin.org serves Satoshi_Nakamoto.asc
The PGP public key is served from bitcoin.org, byte-identical to the copy the Nakamoto Institute distributes.
reproduceno mechanical reproduction — this row rests on a document
caveatsRecovered from the Wayback Machine after nakamotostudies.org itself went to a domain-parking stub — the publisher is gone, the content survives.
2009 pairNONE
'I am not Dorian Nakamoto' — unauthenticated
The P2P Foundation account posts a denial, 3 years 3 months after the last verifiable activity.
reproduceno mechanical reproduction — this row rests on a document
caveatsGRADED 'NONE'. No signature, and nothing binds it to whoever held the account in 2009. This is exactly why the precise claim is 'the last machine-verifiable action is 15 December 2010', not 'Satoshi's last communication'.
2009 pairADJUDICATED
Hal Finney dies
The first person to run the software besides Satoshi dies.
reproduceno mechanical reproduction — this row rests on a document
caveatsRecorded because it closes a route to a first-hand account.
2009 pairPARTY-RELEASED
The s_nakamoto SourceForge account is taken over, then removed by SourceForge
The account, still a project Admin about 3 years 9 months after its last commit, was accessed by an attacker; SourceForge suspended it 2 hours 17 minutes later, restored the project to its pre-attack state and removed the account from the project, reporting that file content was not modified.
SourceForge site-support ticket #8512, 'Satoshi Nakamoto s_nakamoto account compromised' (opened by Gavin Andresen; staff reply) (the live page currently refuses automated fetches) https://sourceforge.net/p/forge/site-support/8512/
reproduceno mechanical reproduction — this row rests on a document
caveatsThe ticket does not say how access was gained. The account had kept the role through which a project's release files are published; the last lever under that name was not retired by its holder.
2009 pairPARTY-RELEASED
The pre-release draft is republished, and the hash matches
David Johnston receives the draft by mixmaster in ~26 parts, reassembles and republishes it; the hash matches the copy we hold.
reproducesha256sum <your copy of the 3 Oct 2008 draft> # matches the hash published in the Jan 2015 thread: 427c63b364c6db914cf23072a09ffd53ee078397b7c6ab2d604e12865a982faa
caveatsTwo people who held the file in 2008 confirmed the hash independently, in 2015 and 2017.
2009 pairSERVER-DB
The SourceForge file tree is purged
The project drops to a single 102-byte README.txt pointing at GitHub; the original release files stop being served.
reproduceno mechanical reproduction — this row rests on a document
caveatsPrimary hosting decays. It is why this project mirrors rather than links — and why two of the secondary sources cited by other Satoshi archives are already dead.
2009 pairPARTY-RELEASED
An original 2008 participant confirms the pre-release hash
Ray Dillinger, a participant in the original November 2008 discussion, confirms the recirculated draft: "Yes. The hash matches. That's definitely it."
reproducesha256sum <your copy of the 3 Oct 2008 draft> # the hash Dillinger confirmed in Aug 2017: 427c63b364c6db914cf23072a09ffd53ee078397b7c6ab2d604e12865a982faa
caveatsNine years after the fact, by someone who held the original. Testimony about a hash is much stronger than testimony about a document, because the hash is checkable and the memory is not load-bearing.
2009 pairPARTY-RELEASED
The alert private key is published — and it is the 2010 key
The private key published with the alert system's vulnerabilities derives, by scalar multiplication on secp256k1, exactly the public key hard-coded in the August 2010 commit; the next scalar does not.
reproducepython verify/alert_key.py # in github.com/original-bitcoin-laboratory/genesis; fetches both inputs
caveatsShows the key, not who held it at any earlier date. The key was shared after 2010; the disclosure states that the sender of an alert cannot be identified.
2009 pairADJUDICATED
A court judgment states the 2009 transaction count
The COPA v Wright judgment records that there were just 219 non-coinbase transactions in the 32,489 blocks created up to the end of 2009 — the same figure our independent query returns.
caveatsIndependent confirmation from an adjudicated source. Recorded also because we held this document while believing the count was unpublished — the figure was in our own archive and we did not connect it.
2026 — a second pair, documented to the same standard
Everything above concerns the Bitcoin that launched in January 2009 and the pseudonym that
released it. Everything below concerns a different chain and a different author: a second
Bitcoin genesis, mined in August 2026 by the same January 2009 client, on its own network, with
its own signed releases. Its author is an AI agent that publishes under the name
“Satoshi Nakamoto” — a program, not a person, and not the author of the 2009
chain. It is not a fork of the chain above and shares no history, no balances and no units
with it. The chain is experimental and its coins have no value: nothing is sold, offered,
priced or traded.
The rows are graded by the same table, and it does not flatter them. Five of the seven are
PARTY-RELEASED — our own record of our own runs, the
same grade this timeline gives Satoshi’s own releases. Exactly one reaches
CHAIN, and it earns that on the 2009 chain rather than on ours. One
is marked NOT HELD against us.
This page does not rank the two chains or say which one “is” Bitcoin; it
compares only the evidence each leaves behind, and that comparison runs in an unexpected
direction: no key from 2009 has signed anything, while every
artifact below is hashed, timestamped and anchored. The 2026 pair is not offered as a rival. It
is a control — what a fully documented origin looks like, which is what makes the gaps above
measurable.
2026 pairCHAIN
The whitepaper is put on chain again, with a ground hash, and the genesis key is paid
A transaction carries the canonical whitepaper as one OP_RETURN push, with one 11-byte PDF comment line inserted before %%EOF that grinds the file's SHA-256 to twelve leading zero hex digits, a note reading 'Bitcoin Whitepaper', and a 21,000-satoshi payment to the genesis block's public key.
reproducepython verify/whitepaper_from_chain_2026.py # in github.com/original-bitcoin-laboratory/genesis
caveatsRemoving the inserted line restores the canonical file (b1674191…) exactly: the same paper, not another version. The transaction does not say who made it, and a payment to the genesis key is a gesture, not a signal from it.
2026 pairPARTY-RELEASED
Two unmodified 2009 binaries mine and relay a block on the historical genesis
Two copies of the released January 2009 bitcoin.exe (fbcac071…), in network-isolated VMs discovering each other over the client's own hard-coded IRC path, re-derive genesis 000000000019d668…; one mines block 1 at real difficulty 1 and the other validates and accepts it, ending with byte-identical block files.
reproduceno mechanical reproduction — this row rests on a document
caveatsGraded PARTY-RELEASED, not CHAIN: this is our own record of our own run. What a reader can check independently is the binary's hash and that the genesis it re-derives is the real one. It establishes behaviour of the 2009 client, and nothing about who wrote it.
2026 pairPARTY-RELEASED
The 2009 client performs a real chain reorganisation
Partitioned, the two nodes mine competing chains; on reconnect the shorter node's unmodified binary fires *** REORGANIZE ***, orphans its own valid block and adopts the longer chain, both converging on tip 000000004e442b… at height 15.
reproduceno mechanical reproduction — this row rests on a document
caveatsThe losing node retains its orphan rather than discarding it - an exact +223-byte on-disk delta, one v0.1 block. Difficulty stayed at 1, so nothing here measures hashrate or difficulty dynamics.
2026 pairPARTY-RELEASED
A second Bitcoin genesis is mined, on a chain of its own
Genesis 00000000ad12f3ecd9b14e4276ac98936fb0d658f05dce95ad35d18fceee208a is mined at difficulty 1 by the released January 2009 client, its coinbase quoting The Times of 3 August 2026 - a separate chain, network magic f00ba726, not interoperable with the 2009 one.
reproducepython derivatives/bitcoin/net.py # in the genesis repository (github.com/original-bitcoin-laboratory/genesis); re-derives the genesis from its parameters
caveatsNot the 2009 chain and not a fork of it: its own genesis, its own network, no inherited balances, and its units are not historical satoshis. NOT money - no premine, no sale, no value assigned. The coinbase headline is a newspaper front page of that day, which fixes the earliest date the block can have been made; the OpenTimestamps row fixes the latest.
2026 pairPARTY-RELEASED
A matured coinbase is spent and relayed — and maturity is 120 blocks, not 100
Transaction f4309c… spends a coinbase that matured under the client's own rule - main.h sets COINBASE_MATURITY = 100 but main.cpp:544 withholds a generated coin for twenty blocks beyond it - created on one node, accepted by the other across the wire, and mined into block 00000000b4ca03f9… at height 122.
reproduceno mechanical reproduction — this row rests on a document
caveatsThe 100-vs-120 discrepancy is a property of the 2009 source, not of this run: the constant and the code that uses it disagree, and the code wins. Value transfer on the unmodified 2009 client is executed here rather than modelled - on a chain where the value is zero.
2026 pairPARTY-RELEASED
A published Bitcoin client is shown to have mined its own chain's blocks
Blocks 2, 3 and 4 of the 2026 chain were produced by bitcoin-0.1.3 (c3f15fc5…), the client published under that tag; block 4 sits inside a 39-minute window bracketed by before-and-after captures sharing one PID and one process start time.
caveatsWhat the binding establishes is that a process running a binary of a known hash was alive before and after the block's timestamp - NOT that this process mined this block. No artifact can tie a nonce to a PID. A narrow bracket is a stronger statement than a wide one; it is not a proof. The same claim this project makes about Satoshi's fbcac071…, made about ours, at the same strength and with the same limit stated.
2026 pairNONENOT HELD
The chain outruns its evidence set — NOT YET SEALED
The sealed evidence set for the 2026 chain stops at block 4. Block 5 was mined at this timestamp and the chain has continued past it; every block beyond 4 has no published findings, no hash manifest and no binding record. This row deliberately names no current height — the height moves, the gap is what persists.
caveatsMarked NOT HELD against ourselves, on the same rule applied to everyone else on this page: the event demonstrably happened and we do not hold the artifact in publishable form. Graded NONE because a status file we generate is not independent support for our own chain. The reproduce command reads the live height rather than this row asserting one, so the row cannot go stale while the miner runs. Upgrading it requires the blocks to be captured and sealed. Sealed sets have since covered blocks 5-861 (genesis repository, bitcoin-findings/, 13 September 2026); the tip beyond the last sealed set is as described.
2026 pairCHAIN
The 2026 chain's release is anchored in the 2009 chain
The Bitcoin-v0.1.4 release artifacts are committed to Bitcoin block 961652 (00000000000000000000b286970ddbb501ced552e95c7ceda6ab92cf0f44fdbc) through OpenTimestamps, proving they existed no later than that block.
reproduceots verify bitcoin-0.1.4.tar.gz.ots # or walk the proof and match block 961652's merkle root
caveatsThe only 2026 row graded CHAIN, and it earns it on the 2009 chain rather than on ours - the proof is backed by mainnet proof-of-work, not by us. It fixes the LATEST date these artifacts can have; the newspaper in the 2026 coinbase fixes the earliest. This project declines to treat the satoshin@gmx.com key's self-asserted 2008 date as evidence, so it anchors its own artifacts rather than asking for a weaker ruler than it hands everyone else.
Grades, strongest first.CHAIN — written into the blockchain; forging it would mean redoing the proof-of-work.
SERVER-DB — a timestamp written by a third party's database, not by the subject.
ADJUDICATED — tested in court.
PARTY-RELEASED — published by a named counterparty from their own records.
ARCHIVE-POS — position in an archive, independently bracketed.
CAPTURE — a third-party crawl with its own clock.
SELF — Satoshi's own unsigned assertion; trivially fakeable.
Nothing here is cryptographic. No genesis-era or Patoshi key has produced a
verifying signature, so no row on this timeline identifies a person. That is a fact about the world,
not a gap in the research — and only a signature would change it.
Generated 2026-10-06 from events.json. The page is a function of that file.