Corrections ledger · 32 claims
The claim, then the record.
Each entry below is a claim that is widely repeated about Bitcoin’s early record, set against what the primary sources actually say. Every entry links the record it relies on and the dossier where the working is shown. This is not a list of our corrections to ourselves; that lives on the corrections page.
overstated · as stated by Bitcoin Wiki
A full six days passed between the genesis block (block 0) and block 1.
What the record shows. The two block headers are 463,160 seconds apart: five days, eight hours, thirty-nine minutes and twenty seconds (timestamps 1231006505 and 1231469665).
One further detail is routinely garbled. Block 1 (00000000839a8e6886ab5951d76f411475428afc90947ee320161bbf18eb6048) is timestamped 1231469665 — 9 January 2009 at 02:54:25 UTC. The interval is 463,160 seconds: five days, eight hours, thirty-nine minutes and twenty seconds. The widely repeated "six days" — the Bitcoin Wiki says "a full 6 days" — overstates what the two headers actually record.
Where the claim appears ↗ · Record 1 ↗ · Record 2 ↗ · Working shown here · Permalink §
unproven
Satoshi deliberately made the genesis block reward unspendable.
What the record shows. The original implementation hard-coded block 0 and never inserted its coinbase into the transaction database, so the output was never in the UTXO set. The Bitcoin Wiki says it is not known whether this was intentional or accidental.
Its 50 BTC output has never been spendable — not by design statement, but because the original implementation never put it in the database.
imprecise
The genesis address 1A1zP1eP5QGefi2DMPTfTL5SLmv7DivfNa is unspendable, so anything sent to it is lost.
What the record shows. Only the original block 0 coinbase output is unspendable. Later outputs paid to the address enter the UTXO set normally and the key carries no protocol-level restriction.
The restriction attaches to that one output, not to the address. Outputs later sent to 1A1zP1eP5QGefi2DMPTfTL5SLmv7DivfNa enter the UTXO set normally and carry no special constraint; the key itself has no protocol-level disability.
unproven · as stated by Sergio Demian Lerner (Bitslog); Whale Alert
Satoshi mined about one million bitcoin in 2009–2010.
What the record shows. The size of the Patoshi set has never been settled: Lerner argues for close to 1.1M coins, Whale Alert for 1,125,150 BTC, and BitMEX Research for 600,000–700,000 BTC with the warning that the evidence is far less robust than many have assumed.
The size of the set has never been settled. BitMEX Research, publishing on 19 August 2018, reconstructed the ExtraNonce slopes independently and concluded the dominant 2009 miner more likely produced "600,000 to 700,000 bitcoin". Their caveat was pointed: "Although there is strong evidence of a dominant miner in 2009, we think the evidence is far less robust than many have assumed."
Where the claim appears ↗ · Record 1 ↗ · Record 2 ↗ · Record 3 ↗ · Working shown here · Permalink §
unproven
The Patoshi blocks are Satoshi Nakamoto's blocks.
What the record shows. The pattern shows a single dominant early miner; equating that miner with Satoshi is an inference, not an established fact. Lerner himself writes that the relation is based on unsigned emails and not math.
The identification of the Patoshi set with Satoshi Nakamoto is a contested attribution, not an established fact. Lerner himself writes that "Because the relation is based on unsigned emails and not math, we'll never be sure," and in 2013 said he could not assure with 100% certainty that the pattern's blocks were Satoshi's.
unproven · as stated by Whale Alert
The Patoshi miner ran at least 48 computers.
What the record shows. Lerner's re-mining of the 10,025 Patoshi blocks in the first 13.2K concluded the miner was simply multi-threading in a single high-end CPU; the hardware inference is directly contradicted between sources.
Whale Alert inferred "at least 48 computers, with one machine for coordination"; Lerner's re-mining of the 10,025 Patoshi blocks in the first 13.2K concluded "Patoshi was simply multi-threading in a high-end CPU". Both claims are still in print.
Where the claim appears ↗ · Record 1 ↗ · Working shown here · Permalink §
imprecise
The first Bitcoin transaction to Hal Finney happened on 11 January 2009.
What the record shows. Block 170 is timestamped 12 January 2009 at 03:30:25 UTC, which was the evening of 11 January in California; the date depends on the time zone used.
The block timestamp is 12 January 2009 at 03:30:25 UTC, which was 11 January, 19:30 in California — which is why some accounts date it to 11 January.
unproven
Satoshi Nakamoto sent the first Bitcoin transaction to Hal Finney (stated as a proven fact).
What the record shows. The on-chain facts are verifiable, but the identity of the sender rests on Finney's own account and the emails he showed Forbes, not on a cryptographic signature: a well-corroborated attribution rather than a proof.
That the sender was Satoshi Nakamoto rests on Finney's own account and on the contemporaneous email archive he showed Forbes, not on a cryptographic signature; it is a well-corroborated attribution rather than a proof.
incorrect
Satoshi's "kicked the hornet's nest" post of 11 December 2010 was his last public forum message.
What the record shows. The Nakamoto Institute archive shows one later post, 12 December 2010 at 18:22 UTC, a routine 0.3.19 release note about DoS limits.
Corrected from an earlier draft: the 11 December 'hornet's nest' post was NOT Satoshi's last public forum message. The Nakamoto Institute archive shows one later post, 12 December 2010 at 18:22 UTC, announcing the 0.3.19 build. Any page saying otherwise is wrong.
understated · as stated by Bitcoin Magazine
WikiLeaks received about 4,000 BTC in Bitcoin donations.
What the record shows. The 4,000 BTC figure is the public address only; it understates total giving because WikiLeaks also let donors generate one-off addresses, and on-chain receipt totals include change returning to an address.
Bitcoin Magazine reported it took delivery of its 4,000th bitcoin on 12 November 2016, across more than 25,000 incoming transactions over five years. That understates total giving, because WikiLeaks also let donors generate one-off addresses.
Where the claim appears ↗ · Record 1 ↗ · Record 2 ↗ · Working shown here · Permalink §
imprecise · as stated by Federal Reserve Board working paper (Badev and Chen); UCSD "A Fistful of Bitcoins"
SatoshiDice accounted for half (or 60%) of all Bitcoin transactions.
What the record shows. The figures vary by source, metric and window: the Fed paper says more than half within weeks and over 40% for most of its life; UCSD says about 60% of overall activity. Both count transactions, not economic value, and are not directly comparable.
The 'half of all transactions' figures vary by source, metric and window: the Fed paper says more than half within weeks and over 40% for most of its life; UCSD says about 60% of overall activity. Both count transactions, not economic value, and are not directly comparable.
Where the claim appears ↗ · Record 1 ↗ · Record 2 ↗ · Working shown here · Permalink §
imprecise · as stated by CoinDesk ($11.5m); Bitcoin Magazine ($12.4m)
SatoshiDice sold for $11.5 million (or $12.4 million) in July 2013.
What the record shows. The 126,315 BTC sale price is agreed; only the dollar equivalent differs between the two reports, and the buyer has never been identified.
The 126,315 BTC sale price is agreed, but the dollar equivalent was reported as $11.5m (CoinDesk) and $12.4m (Bitcoin Magazine), and the buyer has never been identified.
Where the claim appears ↗ · Record 1 ↗ · Record 2 ↗ · Working shown here · Permalink §
imprecise
The whitepaper transaction has 945 (or 946, or 947) outputs.
What the record shows. The raw transaction has 948 outputs: 945 1-of-3 bare multisig, one 1-of-1, and two pay-to-pubkey-hash (a one-satoshi tip and change); 946 outputs carry data.
Secondary write-ups disagree on the output arithmetic — 945, 946 and 947 all circulate — because they count the data outputs, the one-satoshi tip and the change differently.
misattributed · as stated by The uploader/downloader tools' own labels, as catalogued by Ken Shirriff
Satoshi Nakamoto embedded the whitepaper in the blockchain (the upload tools carried his name).
What the record shows. Shirriff judged the Satoshi label probably false; the label is not evidence of involvement by the whitepaper's author, and the author of the transaction has never been identified.
Who created this transaction is unknown and no credible attribution exists. The uploader and downloader tools used in the same burst were labelled with Satoshi Nakamoto's name, which Shirriff assessed as probably false; that label is not evidence of involvement by the whitepaper's author, and the recurring one-satoshi payment to the genesis address is a property of those tools rather than a signed statement by anyone.
Where the claim appears ↗ · Record 1 ↗ · Working shown here · Permalink §
incorrect
El Salvador's Bitcoin Law created the country's bitcoin reserve.
What the record shows. The 2021 law says nothing about the state accumulating bitcoin; the reserve is executive practice, not statute.
Notably, the law said nothing about the state itself accumulating bitcoin. The reserve that followed was a matter of executive practice, not statute.
imprecise · as stated by Wikipedia (9 June); FREOPP full-text edition (8 June)
El Salvador's Bitcoin Law was passed on 9 June 2021 (or 8 June 2021).
What the record shows. The session ran past midnight and sources split between 8 and 9 June 2021, so the site gives both.
First, the date of the Bitcoin Law vote: the session ran past midnight and sources split between 8 and 9 June 2021 (Wikipedia says 9 June, the FREOPP full-text edition says 8 June), so the timeline gives both.
Where the claim appears ↗ · Record 1 ↗ · Record 2 ↗ · Working shown here · Permalink §
incorrect · as stated by Counterparty documentation
The Counterparty burn window ran for 5,000 blocks, ending on 3 February 2014.
What the record shows. The constants in config.py give blocks 278,310 to 283,810, which is 5,500 blocks, with the closing block mined on 2 February 2014 at 19:58:28 UTC.
Counterparty's own documentation describes the window as "5,000 Bitcoin blocks" running to 3 February 2014, whereas the constants in config.py give blocks 278,310 to 283,810, which is 5,500 blocks, with the closing block mined on 2 February; the code has been preferred.
Where the claim appears ↗ · Record 1 ↗ · Record 2 ↗ · Working shown here · Permalink §
overstated · as stated by Wikipedia
Counterparty's proof-of-burn destroyed approximately 2,140 BTC.
What the record shows. The canonical ledger shipped in the client records 2,125.63000728 BTC destroyed inside the valid window; larger counts include coins that arrived at the address outside the window and bought nothing.
Published totals for BTC burned also vary, with Wikipedia giving approximately 2,140 BTC, because some counts include coins that arrived at the address outside the valid window; only the in-window figure is stated here.
Where the claim appears ↗ · Record 1 ↗ · Record 2 ↗ · Working shown here · Permalink §
incorrect
Hal Finney's 10 BTC from the first transaction have never moved.
What the record shows. The address shows total_received 1004401259 sats and total_sent 1004328551 sats across 53 transactions, so nearly everything received has been spent.
Contrary to common belief the 10 BTC did not stay put: as of this check the address shows total_received 1004401259 sats and total_sent 1004328551 sats across 53 transactions, so nearly everything received has been spent.
misattributed
The address 12cbQLTFMXRnSzktFkuoG3eHoMeFtpTu3S is Hal Finney's.
What the record shows. That address is the block 9 coinbase, the sender's side of the transaction; it belongs to the early-mining set and should not be labelled as Finney's.
The block-9 coinbase address is the sender's, not Finney's — it belongs to the early-mining set and should not be labelled as his.
incorrect · as stated by Merkle Science blog
433,468 BTC of Bitfinex hack proceeds moved in April 2021.
What the record shows. The on-chain figure is 433.4689 BTC; the source misprinted the decimal. The companion figure of 259.74 BTC matches exactly.
The source contains a typographical error - it prints '433,468 BTC' where the on-chain figure is 433.4689 BTC, confirmed this session via mempool.space; the companion figure of 259.74 BTC matches exactly.
Where the claim appears ↗ · Record 1 ↗ · Record 2 ↗ · Working shown here · Permalink §
incorrect · as stated by Arkham Intelligence, as reported by The Block
The 4 July 2025 movement of the eight 2011 wallets was a wallet security upgrade, not a sale.
What the record shows. Later transfers went onward to trading desks and Galaxy Digital, which on 25 July 2025 announced it had sold more than 80,000 BTC for the client; the initial reading was overtaken.
Arkham's initial reading of the 4 July movement as a security upgrade rather than a sale was overtaken by the later transfers.
Where the claim appears ↗ · Record 1 ↗ · Record 2 ↗ · Working shown here · Permalink §
imprecise
Block 170 contains the first Bitcoin transaction.
What the record shows. Block 170 is the first widely recognized person-to-person transfer; the genesis and mining (coinbase) transactions predate it.
The phrase ‘first Bitcoin transaction’ should be qualified: block 170 is the first widely recognized person-to-person transfer, while genesis and mining transactions predate it.
imprecise
Bitcoin's maximum supply is 21 million BTC.
What the record shows. Under Bitcoin Core's integer subsidy function the maximum sum of permitted subsidies is 20,999,999.9769 BTC; "21 million" is a rounded conventional label, and issued and spendable supply are lower still.
Bitcoin does not allocate infinitely divisible units, though. Subsidies are integers counted in satoshis, and right shifts discard remainders once successive halvings reach odd values. Summed under the current function, the maximum is 20,999,999.9769 BTC.
imprecise
A Bitcoin halving happens every four years.
What the record shows. The rule is based on block height (every 210,000 blocks), not a calendar date; four years is an approximation from the ten-minute target and actual dates depend on block production.
The rule is based on block height, not a calendar date: mainnet divides height by a 210,000-block interval. The ten-minute target makes four years a useful approximation, but actual halving dates depend on how quickly blocks are produced.
incorrect
Bitcoin Cash increased Bitcoin's block size limit.
What the record shows. Bitcoin Cash began a separate chain under incompatible rules on 1 August 2017; nodes enforcing Bitcoin's rules did not treat its larger blocks as valid Bitcoin blocks.
No. Bitcoin Cash began a separate chain under incompatible rules on 1 August 2017. Nodes enforcing Bitcoin’s existing consensus rules did not treat the new chain’s larger blocks as valid Bitcoin mainnet blocks.
overstated · as stated by Mt. Gox's own 28 February 2014 announcement
Mt. Gox lost 850,000 BTC.
What the record shows. The 850,000 figure was a provisional company estimate; after 199,999.99 BTC was found in an old-format wallet, Mt. Gox revised the estimated disappearance to about 650,000 BTC and reported roughly 202,000 BTC held.
Mt. Gox initially reported almost all of approximately 850,000 BTC missing. After finding 199,999.99 BTC in an old-format wallet, it reported roughly 202,000 BTC held and revised the estimated missing amount to approximately 650,000 BTC; all were company-reported provisional figures.
Where the claim appears ↗ · Record 1 ↗ · Record 2 ↗ · Working shown here · Permalink §
incorrect
MicroStrategy/Strategy has never sold any bitcoin.
What the record shows. An SEC-filed 8-K reports that MacroStrategy sold 704 BTC on 22 December 2022 for approximately $11.8 million and bought 810 BTC two days later.
They also disprove an overly simple mythology: MacroStrategy sold 704 BTC on 22 December 2022 for approximately $11.8 million and acquired 810 BTC two days later. The annual balance still rose, but the filing record does not support saying that the company has literally never sold bitcoin.
incorrect
Exchange-traded Bitcoin exposure (the first Bitcoin ETF) began in January 2024.
What the record shows. GBTC registered its shares in 2019, the futures-based BITO began trading in October 2021, and Canadian and European products predate the U.S. spot cohort; 2024 introduced U.S. national-exchange-listed spot trust shares.
Earlier Canadian and European exchange-traded products also form part of global ETP history, but they should not be collapsed into the SEC decision or used to imply that exchange-traded Bitcoin exposure began in 2024.
incorrect
QuadrigaCX collapsed because its founder died holding the only keys to its cold wallets.
What the record shows. The Ontario Securities Commission's investigation concluded the collapse resulted from fraud by the co-founder, including misappropriation and fictitious trading, not merely inaccessible keys; EY found the identified cold wallets had been empty since April 2018.
QuadrigaCX was initially narrated as a lost-key problem after its co-founder’s death. The Ontario Securities Commission’s investigation instead concluded that the collapse resulted from fraud, including misappropriation and fictitious trading.
incorrect
The Bitcoin blockchain stores account balances, and coins "move between addresses".
What the record shows. No protocol-level balance exists; nodes track unspent transaction outputs, wallets sum them for display, and "movement between addresses" is an interpretation of the transaction graph.
No protocol-level account balance is stored. Nodes maintain a set of unspent transaction outputs, each with a value and spending condition. Wallet software finds relevant UTXOs and sums them to display a familiar balance.
imprecise
A Bitcoin block is produced every ten minutes.
What the record shows. Ten minutes is an average design target maintained by retargeting every 2,016 blocks, not a schedule for the next block.
Ten minutes is therefore an average design target, not a schedule for the next block.
How entries qualify
An entry needs a claim that is actually in circulation, a primary record that settles it (a block header, a transaction, source code, a filing, a contemporaneous post), and a dossier on this site where the reasoning is laid out with sources. Disagreements about interpretation do not qualify; only claims a record can check. Report an error in this ledger through the corrections process.