Block 961,632: The Soft Fork the Math Already Killed

Block 961,632 is where two versions of Bitcoin part ways, if you believe the rhetoric. Dathon Ohm, the pseudonymous author of BIP-110, says that block is the moment Bitcoin Core becomes insecure, the moment miners using Core templates produce invalid blocks on an incoherent chain. Michael Saylor reads the same block and tells BIP-110's backers to stand down, because Bitcoin continues normally while BIP-110 stalls or forks into irrelevance. Same event. Opposite verdicts. Both cannot be right. On August 6, Ohm posted an all-caps thread on X: "ATTENTION. THIS IS NOT A DRILL. Mandatory signaling for BIP-110, which fixes critical vulnerabilities, begins in 290 blocks (~48 hours from now)." He thanked the miners he said are signaling readiness. The monitor his project cites shows a modest uptick, not a breakthrough. The data suggests neither side fully controls the outcome. At block 961,421, the BIP-110 monitor logged 47 signaling blocks out of 1,806 mined. That is 2.60%. The rate slipped from 2.70% at block 961,022, where Saylor counted 38 signals. Eight new signals over 392 blocks. Not a surge. A leak. Early lock-in requires 1,109 signaling blocks within the period. With 217 blocks remaining, the highest reachable total is 263. No completed two-week stretch since December has finished above 1.29%. The block confirms what the eyes missed: this soft fork was never close. BIP-110 is a temporary soft fork that caps the amount of data a Bitcoin transaction can carry. It is a response to the blockspace spam debate — the fight over arbitrary data embedded in transactions, the same fight that produced inscriptions, ordinals, and a series of increasingly loud arguments about what Bitcoin blocks are for. The proposal ships in Bitcoin Knots, a smaller client maintained by Luke Dashjr, not in Bitcoin Core. Mandatory signaling begins when versionbit 4 must be set in every block header. After that threshold, BIP-110 nodes treat non-signaling blocks as invalid. Ohm's warning to users was explicit. "It is not recommended to run Bitcoin Core, as it will become insecure when mandatory signaling begins, and miners getting their templates from Core may produce invalid blocks on an incoherent chain that keeps being wiped out, along with any earnings." I have been reading protocol documentation and auditing smart contracts since 2017, when I caught an overflow bug in an ICO's batchMint function that would have cost the project $2.4 million. That experience taught me a rule: when a developer's public statements exceed their own technical documentation, something does not verify. Bip110.org does not call Bitcoin Core insecure. It presents BIP-110 as a curb on arbitrary data. The word "critical" appears in Ohm's Twitter feed. It does not appear on the project's site. Let me also state what Ohm is describing, technically. A chain that "keeps being wiped out" is a chain that cannot hold against a reorg. That is not a security failure. It is a hash power failure. A minority chain gets wiped out because it cannot outwork the majority, not because the majority client is insecure. "Insecure" is doing a lot of work in that sentence. The claim needs evidence. The site does not provide it. Hash the truth, verify the story. Let me treat this like an order book, because that is what it is. Signaling is not a vote in the democratic sense. It is a commitment backed by block rewards. Miners who signal are betting their revenue on BIP-110 nodes achieving sufficient deployment. Miners who do not signal are betting their revenue on miners who do. The numbers say the majority bet is on inaction. 47 blocks out of 1,806. 2.60%. To hit the 1,109 threshold, the network would need a sudden, sustained surge of signaling in the remaining 217 blocks. The required rate exceeds 490% — not merely improbable in a two-week adjustment period. Arithmetically impossible within a single period. The lock-in window closes with the period boundary. There are not enough blocks left. Dashjr's position is that the outcome is already settled. Miners who refuse to signal lose block rewards entirely, and their invalid blocks mislead nodes that have not updated. Technically true, if BIP-110 nodes constitute the chain. But that is the conditional. The soft fork does not define Bitcoin unless enough hash power enforces it. Consensus is not a flag in a header. It is a distribution of economic weight. Dashjr also told BeInCrypto there is no material opposition to the change. The signaling data contradicts that claim on its face. If there is no material opposition, where are the blocks? In Bitcoin, hash power is the only ballot that counts. Twitter is not consensus. Adam Back's characterization is blunter: BIP-110 "doesn't work, and completely fails at technical consensus," and he contrasts it with SegWit, which had both technical and ecosystem consensus before activation. That distinction matters. SegWit had months of public debate, wallet support, exchange coordination, and a clear majority of hash power before it locked in. BIP-110 has a minority of signaling miners, a competing client, and an activation threshold that was always going to be tested under adversarial conditions. Now the "critical vulnerabilities" claim. Ohm describes BIP-110 as a fix for critical vulnerabilities. The project's own website does not. That discrepancy is the kind of detail that should never survive a competent audit. Either the vulnerabilities exist — in which case they should be documented, reproducible, and disclosed — or they do not, in which case the warning about Core's security is marketing dressed as alarm. From my work auditing the 2017 ICO space, I learned that the most dangerous code is not the code that fails loudly. It is the code that asks you to trust the author's characterization of the risk. "Insecure" is a strong word. It means funds can be lost. It means nodes can be compromised. If running Bitcoin Core genuinely becomes insecure after block 961,632, that is a critical disclosure that belongs on the project's website, in the release notes, and in every channel available. It is not hidden in a tweet. Silence is the safest ledger. But the silence here is not a ledger. It is an absence where a claim should be. I also have a mechanical read on this from running arbitrage desks during the ETF launch cycle. Mechanical execution reveals intent better than any announcement. Here, the signal is unambiguous: miners have not redirected a single percentage point of hash power toward BIP-110. The rate is flat. The period rate slipped. Whatever enthusiasm exists on Twitter is not showing up in block headers. That is the most objective fact in this entire debate. Here is what most coverage gets wrong, including parts of the coverage of this story. The danger to Bitcoin holders is not a chain split. It is the operational uncertainty window around any failed soft fork attempt. The moment a credible split threat exists — even a weak one — exchanges begin pausing Bitcoin deposits and withdrawals. Custodians freeze withdrawals. Derivatives desks widen basis spreads. The market prices in the tails, however unlikely. That risk is amplified now because of the custody layer. Spot Bitcoin ETFs hold over a million BTC between them. Fund issuers are not going to hold through a disputed fork without freezing redemptions and suspending creations. That operational freeze is real even if the fork is not. The failure mode is not two chains. It is one chain running normally while the rails around it lock up. In May 2022, when Terra's stablecoin de-pegged, the narrative was panic. The mechanics were simpler: a collateralization ratio that was mathematically unsustainable. I hedged by the numbers, not by the headlines, and preserved capital while others liquidated. The same discipline applies here. Wait for block 961,632. Watch the signaling rate. Ignore the Twitter drama. The second blind spot is the precedent being set. BIP-110 does not need to activate to do damage. A single developer, a minority client, and a low signaling threshold can create enough uncertainty to freeze billions in value for days. That is not a bug in BIP-110. It is a feature — or a gap — of Bitcoin's governance. Every failed activation attempt teaches the next one how to be more efficient at creating chaos. The infrastructure that matters is not the chain. It is the speed with which exchanges, custodians, and market makers can distinguish a real fork from a failed one. Speed kills the hesitant; logic kills the greedy. The hesitant will move their coins preemptively. The greedy will try to arbitrage the split. Both will lose fees. The logic here is simple: without hash power, there is no split. Trace the anomaly, ignore the noise. The anomaly is real: a low-signaling soft fork threatening to invalidate the dominant client's blocks. The noise is everything else. Entropy claims its due in every block — and entropy here is working against BIP-110. Block 961,632 will arrive this weekend. What happens then is not a mystery. If the signaling rate stays below the threshold, BIP-110 stalls, and Bitcoin continues as if the versionbit never existed. If miners suddenly capitulate — a scenario the data does not support — BIP-110 nodes and non-BIP-110 nodes produce competing chains, and the market resolves the ambiguity within days. The reward structure makes capitulation irrational. Signaling for a soft fork that cannot lock in means burning block reward on invalid work. Miners are not ideological. They are economic. The block confirms what the eyes missed: this fork was dead on arrival, and the only thing left to lose is your composure. Do not move your coins. Do not trade the split. Watch the headers. That is where the truth already is. The open question is not block 961,632. It is what comes after. The versionbit is a template. If a low-signaling minority can generate this much uncertainty at no cost, the next attempt will be louder. The industry needs better detection mechanisms for failed governance gambits — not more opinions about whether they should succeed. The block will tell you first. You have to read it. If you cannot read it, stay out of the way — the next one will be harder to read, and the next harder still.