Bitcoin did not fork on Saturday. A client did.
At block 961,632, nodes running BIP-110 rejected the block AntPool mined, because it did not signal for the proposal. The rest of the network kept that block. Roughnecks, mining through Ocean, produced the one those nodes followed instead. The Block had the heights that evening. By 6 p.m. ET the main chain was at 961,640. The branch was at 961,633. Two blocks of its own, seven behind.
That is the event. A soft fork missed a 55% bar, a minority of nodes refused the next block, and the chain they started stopped.
Two blocks, then nothing
CoinDesk’s Sunday morning count is heights. Around 6 a.m. UTC the branch was still at 961,633. Bitcoin was at 961,681. Forty-eight blocks. On one side that is most of a day. On the other it is a second block and then silence.
Support going in was 2.53% of recent blocks, against 55% to activate without a split. SegWit asked for 95%. Taproot asked for 90%.
The stall has a mechanical cause. The branch inherited Bitcoin’s difficulty, and difficulty only moves every 2,016 blocks. CoinDesk’s monitor put the branch’s next adjustment at about 350 days, against 14 for Bitcoin. A chain that cannot find a block cannot make the next one easier. That was already visible on Sunday. The two-week window where BIP-110 nodes demand a signal on every block runs to height 963,647. At one block every several hours, the branch will not get near it.
Both chains still accept the same transactions. CoinDesk spelled out what that means if you try to spend the fork coin. The signature also spends the bitcoin. Someone can take the transaction you broadcast on the branch and replay it on the chain that has the hash. I would not touch a “claim” flow on a chain that confirms once every few hours and shares a key with the real one.
@saylor posted at 09:45 UTC on Sunday: “Bitcoin worked exactly as designed. BIP-110 was free to fork, and the network was free not to follow. The result was decisive: about 99.85% of Bitcoin’s hashpower stayed with Bitcoin. The BIP-110 branch mined only two blocks and is already more than 80 blocks behind.”
His 80 and CoinDesk’s 48 are a few hours apart. The branch was stuck. Bitcoin was not.
What did not happen
Last September The Rage published messages it said were Dashjr’s, about a hard fork, a committee, and replacing data on the chain so a node would not have to hold it. On September 26 he posted that there is no hard fork. This weekend did not produce one either.
BIP-110 is a temporary rule against storing pictures, text, and other non-payment data in transactions, for about a year. That is CoinDesk’s description, and it is enough. Backers say the data clogs blocks and raises fees for people sending money. Opponents say a paid fee is the permission, and miners do not get a second vote on what the bytes are for.
That fight has been going since Ordinals. A hard fork that rewrites confirmed history, or that hands a committee a delete key, is a different fight. It would mean the ledger you already checked is not the ledger. Saturday was nodes rejecting a block that lacked a signal, then failing to extend their own tip.
Dashjr can call the main network whatever he likes. The hash did not move, and neither did the exchanges. The coin in a normal wallet is the one AntPool’s block belongs to.
Monday was a repo change
The chain story was over on Sunday. Monday was the editor list.
At 19:00 UTC on Sunday, @murchandamus posted that he had recommended removing Luke Dashjr from the BIP editors, to the mailing list. About forty minutes later Dashjr replied that the accusations are false, and that he has followed the BIP process for years. The complaints in circulation are specific. He assigned BIP-110 a number before the list had discussed it, and he merged a pull request minutes after it opened. He says that does not make a new process. That argument is about a GitHub role. It is not about which chain your coins are on.
At 19:45 UTC on Monday, @kanzure posted the result in one sentence: “Fyi, the individual known as Luke Dashjr is no longer a permissioned user of the BIP content repository.”
At 20:54 UTC Dashjr posted that he is taking a sabbatical from Ocean as chairman and CTO. “I believe this is the best use of my time and is in the best interests of Bitcoin and Bitcoin’s long-term health.”
A sabbatical is not a resignation I get to rewrite. He still maintains Knots, the client that enforced the split. A later attempt still has a codebase. As of tonight it does not have the hash, and it does not have the BIP repo.
If you hold bitcoin, your coins are on the chain that kept producing blocks. The branch that stopped at 961,633 ran out of miners. I would not treat it as a second balance, and I would not go looking for a claim page. If they later change the proof of work, or ask you to trust a committee to edit history, write that one when it happens. This weekend, the hash stayed.