Developers propose
They write BIPs and implementations. Publishing or merging code does not activate it on nodes already running in the world.
Bitcoin consensus, without the mythology
Who decides Bitcoin's rules?
No one votes once for everyone. Every full node decides, block by block, what it will accept.
CONSENSUSis what we call it when independent validators converge on the same valid history.
The validation gate
Protocol
Miner
Candidate block
Transactions + proof of work
Your full node
Check every rule
Software you chose, validation you control
Accept
Valid candidate
Eligible for the most-work valid chain
Proof of work orders blocks only after validation.
Four parts. No king.
Each role has real power. None can rewrite everyone else's rules by decree.
Operators choose
They select software and configuration. Their nodes independently accept or reject every candidate block under those rules.
Miners produce
They assemble blocks and spend proof of work ordering valid history. Their coordination shapes security, speed, and fork risk.
Markets value
Users, wallets, exchanges, merchants, and builders decide which valid chain and asset they will accept, price, and build upon.
Signaling is the clock, not the crown.
The 55% threshold is the fast lane. Mandatory activation is the final path.
The path encoded today
Voluntary miner signaling can lock in early. If it does not, BIP-110 nodes reject non-signaling blocks during the mandatory window and lock in no later than block 963,648.
bit 4header signal
55%1,109 / 2,016 blocks
52,416active blocks, then expiry
StartedVoluntaryminers may signal bit 4
961,632Mandatorynon-signaling blocks rejected by enforcing nodes
963,648Locked inno later than this height
965,664Activetemporary rules enforced
+52,416Expiredtemporary rules end
Activation belongs to a history
A node derives deployment state from the branch and ancestors it is evaluating. Height alone is not a global election result.
Same height
Two nodes can report different states when they accept different histories.
Invalid ancestor
Signaling later on a legacy-only history cannot heal the block BIP-110 nodes already rejected.
Compatible return
A miner can return to a BIP-110-valid tip. Legacy nodes may follow that branch because it remains valid to them.
The fork is an ancestor, not a switch.
Three moments explain what signaling alone cannot.
One block can satisfy everyone.
A signaling block that follows the narrower rules is valid to BIP-110 and legacy nodes. During mandatory signaling, setting bit 4 preserves one history.
Lowest-friction miner path: make a reward both economies recognize.
Now the histories diverge.
A non-signaling mandatory-window block, or a rule-breaking active block, can remain legacy-valid while BIP-110 nodes reject it and every descendant behind it.
Later signaling does not repair an ancestor the enforcing node considers invalid.
Work must return to a valid tip.
Miners can extend the BIP-110-valid history. If it becomes the most-work history legacy nodes consider valid, those nodes can reorganize to it.
More work can pull legacy nodes toward the compatible branch; it cannot pull BIP-110 nodes across an invalid block.
Causal forecast
Why this house expects convergence
The case does not depend on Luke Dashjr, GrassFedBitcoin, or any other personality. It is a chain of mechanism and incentive:
- 01The deadline is deterministic. On a BIP-110-valid branch, voluntary signaling does not provide a permanent veto.
- 02Compatibility favors one history before a split. The miner can make a block accepted by both rule sets instead of creating risk for no larger market.
- 03The asymmetry survives a split. The BIP-110-valid history remains valid to legacy nodes; the legacy-only history remains invalid to BIP-110 nodes.
- 04Miners need purchasing-power revenue. After a split, price, liquidity, customers, and reorganization risk determine which reward can keep buying hashpower.
- 05Resistance must organize its own rule. Without a counter-rule rejecting BIP-110-compatible blocks, preserving one compatible history has the lowest coordination cost.
What could defeat the forecast
- False signalbit 4 without rule enforcement can postpone the conflict until an active-rule violation
- Low hashthe enforcing branch can advance slowly or stall before difficulty adjusts
- Two marketsboth histories can survive if each keeps demand, infrastructure, and miners
- Peer splitenforcing nodes need reliable paths to find and extend their valid branch
- Transactionsreplay, reorganization, and double-spend exposure can differ by branch
- Counter-ruleopposition can remove the compatibility advantage by rejecting BIP-110 blocks
- Code errorfunctional tests show intended behavior, not independent review or freedom from bugs
- Expiryending temporary rules cannot erase histories or market choices already made
Inspect the mechanism and tests
- 01Bitcoin Developer Guide: validation and chain selection
- 02Bitcoin Core lifecycle: who enforces consensus changes
- 03BIP 8: branch state, mandatory signaling, and peer risk
- 04BIP 110: current rules and deployment heights
- 05Reference test: mandatory lock-in and activation
- 06Reference test: split, compatible reorg, and expiry
- 07Delving Bitcoin: markets, mining cost, and economic majority
Test the model against both sides
- 01Knut Svanholm: miners answer to present and future customers
- 02US Debt Clock Snap: the first incompatible ancestor is the split
- 03Lyn Alden via TFTC: a minority fork becomes a market test
- 04Peter Todd: false signaling can postpone the real conflict
- 05Luke Dashjr: pool accounting can isolate non-enforcing risk
The standard: sources substantiate the model; they do not substitute for it. Read the full arguments in the Library.