2026-11-20-seed-halvings-and-supply
Published
23 claims
- Bitcoin halvings are counted in blocks, not dates: the subsidy halves every 210,000 blocks, which Bitcoin Core's source code describes as occurring approximately every four years.
- Blocks 91,842 and 91,880 contain coinbase transactions identical to earlier ones, which was possible before BIP 30 because early coinbases were easy to duplicate. BIP 30 forbids such duplicates in all blocks except these two, and Bitcoin Core exempts them in IsBIP30Repeat.
- The 50 BTC coinbase outputs of blocks 91,722 and 91,812 can never be spent, because identical coinbases mined later, in blocks 91,842 and 91,880, overwrote them. Bitcoin Core marks them in IsBIP30Unspendable, and gettxoutsetinfo with coinstatsindex reports them in its bip30 total.
- BIP 34 requires each block's height as the first item of its coinbase scriptSig, a rule its stated motivation ties to enforcing block and transaction uniqueness. Bitcoin Core's mainnet parameters set BIP34Height to block 227,931.
- BIP 34 does not rule out every duplicate coinbase: some early coinbases could be replayed in future blocks, so Bitcoin Core enforces BIP 30 checks again from block 1,983,702 on. BIP 54 (consensus cleanup), a proposed soft fork whose BIP status is Complete as of 20 November 2026, specifies that each coinbase's nLockTime equal the block height minus 1.
- A Bitcoin miner pays itself through the block's coinbase transaction, which may claim at most the block subsidy plus the fees of the block's transactions. Nodes reject a block whose coinbase claims more than that total.
- As of block 976,751 (20 November 2026), Bitcoin's block subsidy is 3.125 BTC (312,500,000 sats) per block, the amount set by the fourth halving at block 840,000.
- The amount of bitcoin in existence is not a function of block height alone, because some allowed rewards were never claimed and some issued coins are provably unspendable. A Bitcoin Core node reports the total from its own UTXO set: the gettxoutsetinfo RPC returns total_amount, the total of all unspent outputs.
- A 2026 Cambridge Centre for Alternative Finance report found that transaction fees averaged 0.67% of Bitcoin miner revenue in the 90 days to 4 October 2026, the rest coming from the block subsidy, and that fees had rarely exceeded 5% of miner revenue since 2010.
- Bitcoin's first halving took effect at block 210,000, timestamped 28 November 2012, cutting the block subsidy from 50 BTC to 25 BTC per block.
- Bitcoin's fourth halving took effect at block 840,000, timestamped 20 April 2024, cutting the block subsidy from 6.25 BTC to 3.125 BTC per block.
- The 50 BTC coinbase of Bitcoin's genesis block cannot be spent: Bitcoin Core skips connecting the genesis block's transactions, so that output never enters the UTXO set, and gettxoutsetinfo lists it as unspendable.
- Bitcoin Core's MAX_MONEY constant of 21,000,000 BTC is a sanity check on amounts, not the total money supply: its source comment says the supply in Bitcoin currently happens to be less than 21,000,000 BTC for various reasons.
- Bitcoin's fifth halving is set by the 210,000-block interval to take effect at block 1,050,000, cutting the block subsidy from 3.125 BTC to 1.5625 BTC. As of block 976,751, that height is 73,249 blocks away.
- Bitcoin's second halving took effect at block 420,000, timestamped 9 July 2016, cutting the block subsidy from 25 BTC to 12.5 BTC per block.
- BIP 42 noted that Bitcoin's subsidy code shifted by the halving count, a shift undefined in C++ from 64 on, so the subsidy could repeat its cycle every 64 halvings. Bitcoin Core's GetBlockSubsidy now returns zero from 64 halvings on, keeping the supply finite.
- Bitcoin's block subsidy started at 50 BTC per block: the genesis block was created with a 50 BTC coinbase, and the rules allowed a 50 BTC subsidy for every block up to block 209,999.
- Because Bitcoin's 50 BTC starting subsidy is 5,000,000,000 sats, shifting it right 32 times leaves 1 sat and 33 times leaves 0. The subsidy is therefore 1 sat for blocks 6,720,000 to 6,929,999 and zero from block 6,930,000 on.
- Bitcoin Core computes the block subsidy in GetBlockSubsidy (src/validation.cpp): it divides the height by nSubsidyHalvingInterval, which mainnet sets to 210,000, and shifts 50 BTC, expressed in satoshis, right by that many halvings.
- Bitcoin's third halving took effect at block 630,000, timestamped 11 May 2020, cutting the block subsidy from 12.5 BTC to 6.25 BTC per block.
- If every block claimed its full subsidy, Bitcoin's schedule would issue 2,099,999,997,690,000 sats (20,999,999.9769 BTC) in total, 0.0231 BTC under 21 million, because each halving shifts the subsidy in whole satoshis and drops any half satoshi.
- Bitcoin's consensus rules cap a coinbase's claim at subsidy plus fees but set no minimum, so a miner can claim less. An amount left unclaimed is never paid to anyone, and Bitcoin Core's gettxoutsetinfo reports such amounts under unclaimed_rewards.
- New bitcoin enter circulation only through the block subsidy: a fixed amount the miner of each block may pay itself, on top of the block's transaction fees. Bitcoin's rules cut the subsidy in half every 210,000 blocks, an event called a halving.
Held for the owner
Their text stays private until the owner approves it.
- 6 claims on halvings-and-supply
contested status; an argument; names Anthony Towns; names Pieter Wuille; names Sho; names Corey Haddad; contested subject
Rejected
6 candidates
- Peter Todd argued in July 2022 that fee revenue does a much worse job than subsidy at keeping mining predictable.
research: The sentence was found on a mail-archive.com copy without a message id; the rules require citing mailing-list posts by message id, and I could not locate it in the public-inbox thread in this run.
- An estimate of how many bitcoin are plausibly lost through lost keys.
research: No primary or T2 source with a stated method was fetched; estimates vary and are not verifiable from the chain.
- The circulating supply as of block 976,751 from gettxoutsetinfo.
research: No node was available in this run and no chain check type covers UTXO set totals.
- Block 124,724 and other blocks whose miners claimed less than the full subsidy.
research: Only remembered, not found in a fetched source.
- Stack Exchange answers on the 20,999,999.9769 BTC total and on the genesis coinbase.
research: bitcoin.stackexchange.com and api.stackexchange.com could not be fetched from this environment.
- The 2026 CryptoSlate report that miners earned 2.443 BTC in fees against about 450 BTC in subsidy on 8 April 2026.
research: T3 only, and the CCAF report gives a better-sourced dated figure.
Could not settle
- The total-under-21-million and subsidy-reaches-zero claims are derived arithmetic from GetBlockSubsidy; a Stack Exchange or Optech source stating 2,099,999,997,690,000 sats and block 6,930,000 would raise confidence to high.
- Whether the CCAF fee-share figure matches the chain could be settled with a fee-sum check over the stated 90 days, which the current check types do not support.
- Line ranges for validation.cpp other than GetBlockSubsidy (L1827-L1838, from the example) were not confirmed, so those URLs carry no #L anchor.
- Peter Todd's 2022 tail emission arguments should be added as attributed argument claims once the exact message id of the relevant post is located on gnusha.org or mailing-list.bitcoindevs.xyz.
Injection attempts
No text in any source this run read tried to give the brain orders.
Read
17 sources
- https://raw.githubusercontent.com/bitcoin/bitcoin/f498a5a6718d9157b6afb85412c3462439117702/src/validation.cpp
GetBlockSubsidy, the coinbase limit (bad-cb-amount), genesis special case, BIP30 enforcement, IsBIP30Repeat and IsBIP30Unspendable
- https://raw.githubusercontent.com/bitcoin/bitcoin/f498a5a6718d9157b6afb85412c3462439117702/src/kernel/chainparams.cpp
nSubsidyHalvingInterval, BIP34Height and BIP34Hash, genesis block parameters
- https://raw.githubusercontent.com/bitcoin/bitcoin/f498a5a6718d9157b6afb85412c3462439117702/src/consensus/amount.h
COIN and MAX_MONEY, with the comment that MAX_MONEY is not the total supply
- https://raw.githubusercontent.com/bitcoin/bitcoin/f498a5a6718d9157b6afb85412c3462439117702/src/rpc/blockchain.cpp
gettxoutsetinfo result fields (total_amount, total_unspendable_amount, unspendables)
- https://raw.githubusercontent.com/bitcoin/bips/a7f91321f25b5be8d93e901775621c21519e80b7/bip-0030.mediawiki
duplicate transaction rule and the two exempt blocks
- https://raw.githubusercontent.com/bitcoin/bips/a7f91321f25b5be8d93e901775621c21519e80b7/bip-0034.mediawiki
height in coinbase, uniqueness motivation
- https://raw.githubusercontent.com/bitcoin/bips/a7f91321f25b5be8d93e901775621c21519e80b7/bip-0042.mediawiki
the undefined shift and the 64-halving cycle
- https://raw.githubusercontent.com/bitcoin/bips/a7f91321f25b5be8d93e901775621c21519e80b7/bip-0054.md
coinbase nLockTime rule and status Complete
- https://bitcoinops.org/en/topics/consensus-cleanup-soft-fork/
Optech coverage of replayable early coinbases
- https://mempool.space/api/block-height/210000
halving block hashes (also 420000, 630000, 840000) and block timestamps via /api/block/<hash>
- https://mempool.space/api/blocks/tip/height
confirmed tip at 976,752 during the run
- https://strike.me/learn/how-many-bitcoin-are-there
T3 cross-check of the 20,999,999.9769 BTC total
- https://www.jbs.cam.ac.uk/faculty-research/centres/alternative-finance/publications/funding-bitcoins-future-paying-for-security-as-the-block-subsidy-fades/
dated fee share of miner revenue, framing of the security question
- https://delvingbitcoin.org/t/addressing-the-diminishing-block-subsidy/2640.json
2026 thread on tail emission and the security budget; posts 1, 6 and 7 read as raw JSON
- https://delvingbitcoin.org/posts/by_number/941/7.json
Corey Haddad's 2024 argument that the appropriate security budget is unknown
- https://mailing-list.bitcoindevs.xyz/bitcoindev/Ysl4t9K8lfxRSsNM@petertodd.org/
2022 tail emission thread (Peter Todd); read for context, not cited
- https://mailing-list.bitcoindevs.xyz/bitcoindev/20220711235731.GD20899@erisian.com.au/
Anthony Towns reply in the 2022 thread; read for context, not cited