Two very different failures hit the Bitcoin world within about a week of each other. One was a protocol-level governance fight that ended with a minority chain producing two blocks before stalling out. The other was a hardware wallet firmware bug that let an attacker walk away with roughly $38 million overnight.
On the surface these look unrelated — one is about consensus rules, the other about random number generation. But sit with both stories for a minute and a shared lesson emerges, one worth taking seriously if you hold Bitcoin, run a node, or care about how this system actually stays trustworthy.
I already covered the Coldcard incident in detail here. This piece brings BIP-110 into the same frame, because the two stories rhyme more than you’d expect.
BIP-110 was a proposal to temporarily restrict storing non-financial data — images, text, arbitrary metadata — inside Bitcoin transactions. In practice, this meant targeting Ordinals inscriptions, BRC-20 tokens, and Runes: the mechanisms that let people embed non-monetary content on Bitcoin’s blockchain.
The proposal was authored under the pseudonym Dathon Ohm, with longtime Bitcoin Core contributor Luke Dashjr credited for the original draft and technical input. The argument behind it: this kind of data usage bloats the blockchain, raises the cost of running a full node, and — in supporters’ view — distracts Bitcoin from its purpose as sound, monetary infrastructure rather than a general-purpose data store.
It’s not a fringe complaint. Node operators have voiced frustration about blockchain bloat from inscription activity for years. But the mechanism BIP-110 chose to force the issue is where things get instructive.
Rather than requiring the traditional near-unanimous miner consensus Bitcoin protocol changes have historically needed, BIP-110 used a user-activated soft fork (UASF) — full nodes running the new software would begin enforcing the restriction on a specific date, regardless of whether miners agreed.
This exact mechanism succeeded once before, in 2017, with BIP148 and SegWit activation. But the 2017 success wasn’t really about nodes unilaterally forcing a rule change. It worked because SegWit had already achieved broad economic-majority alignment — exchanges, wallet providers, businesses, and users were signaling in advance that they’d follow the UASF chain regardless of what miners did. The UASF mechanism was almost a formality by the time it activated.
BIP-110 set its miner signaling threshold at 55% — 1,109 out of every 2,016 blocks in a difficulty period — a meaningfully lower bar than the traditional 95% supermajority Bitcoin has typically required for consensus changes.
Even that reduced threshold was never remotely close to being met.
Signaling monitor data recorded approximately 38 signaling blocks out of more than 9,000 mined between May 1 and early August 2026 — roughly 0.42% of total hashrate. The first signaling block was mined by Barefoot Mining through Ocean Pool on March 1, 2026. No major mining pool joined in the months that followed. F2Pool, one of the largest pools by hashrate, refused outright.
When the mandatory signaling period began on August 7 at block 961,632, miner support stood at just 2.45-2.53%, depending on the measurement window. Supporting nodes rejected non-signaling blocks anyway, which split off a minority chain exactly as critics had warned it would.
What happened next was almost anticlimactic. As of August 9, the standard Bitcoin chain had reached block 961,681. The BIP-110-enforcing minority chain sat at block 961,633 — a gap of 48 blocks, producing new blocks roughly every 6.9 hours instead of the usual ten minutes.
The reason is mechanical, not political: Bitcoin’s mining difficulty only adjusts every 2,016 blocks. A network with a negligible share of the mainnet’s hashrate inherits mainnet’s high difficulty and simply cannot produce blocks at a normal pace until an adjustment catches up — which, at this rate, could take a very long time. The chain was, functionally, paralyzed from the moment it split off.
By August 9, the Roughnecks mining team — among BIP-110’s most visible supporters — announced they were halting mining operations under that name entirely, posting: “We do not see this as a defeat for the BIP110 movement but an escalation to the next step.”
Adam Back, co-founder of Blockstream, was blunt on X: “This simply won’t work — it breaks multiple mechanisms, lacks both technical consensus and ecosystem consensus, and has all four fatal flaws. Everyone hates bloat, but this isn’t the way; you’ll hurt yourselves.”
Bitcoin security researcher Jameson Lopp released a detailed analysis calling BIP-110 “reckless and doomed to fail,” citing fragmentation risk, the potential creation of unspendable UTXOs through Taproot edge cases, and the practical reality that restricting data storage doesn’t actually stop it — data can be re-encoded and embedded through other routes, which Bitcoin Core developers had already pointed out. In 2025, Bitcoin Core 30 had in fact moved the opposite direction, raising the default OP_RETURN limit from 83 bytes to 100,000 bytes.
Michael Saylor’s involvement marked something genuinely new for this kind of protocol fight. Strategy is among the largest corporate holders of Bitcoin in the world, and its active public opposition to a protocol proposal represents a category of stakeholder that simply didn’t carry this kind of weight during the 2015-2017 Blocksize Wars. Saylor stated plainly that 99.85% of miners chose to stay on the Bitcoin mainnet.
Not everyone accepted the “failure” framing. Luke Dashjr rejected the verdict outright, accusing opponents of deliberately misrepresenting the situation, acknowledging the slower block production but describing it as tolerable rather than fatal.
The minority chain is expected to exist briefly, process a thin trickle of blocks, and expire automatically after roughly 52,416 blocks — about a year — without requiring any further vote or action.
At first glance these are two completely different domains. One is social and political — a governance dispute about what belongs in a block. The other is purely technical — a firmware misconfiguration in a random number generator. But look at the failure mode in each case, and the parallel becomes hard to miss.
Coldcard’s vulnerability existed because a check verified the wrong thing. The code confirmed that an RNG configuration setting existed — not that the hardware entropy source was actually running. The system trusted a signal that looked correct on the surface but didn’t verify the substance underneath it.
BIP-110 failed because its proponents mistook a mechanism that worked once for a mechanism that works generally. UASF succeeded in 2017 because of deep, pre-existing economic-majority alignment — the vote had effectively already happened informally before the formal activation date arrived. BIP-110 treated the UASF mechanism itself as sufficient, without first securing anything close to that alignment. It checked the box marked “we used the same activation method as SegWit” without the substance that made SegWit’s version of that method work.
In both cases, something that looked like a valid safeguard — a settings check, a proven activation mechanism — turned out to be hollow the moment it was actually tested under real conditions.
The lesson generalizes cleanly: in Bitcoin, at every layer of the stack, “it’s supposed to work this way” is not the same as “it has been verified to work this way.” Your seed’s randomness. A protocol change’s legitimacy. Both require actual verification, not trust in a process that resembles a safeguard.
For BIP-110’s underlying complaint — non-monetary data usage on Bitcoin’s blockchain — nothing about this outcome resolves the actual disagreement. Ordinals inscription volume has already fallen sharply from its 2023 peak, sitting below 10,000 per day in recent months, but the philosophical and economic dispute over what Bitcoin’s blockspace is for remains fully unresolved. A counter-proposal called DOG Mode, an alternative Bitcoin Core client floated by Ordinals advocate Leonidas in mid-July, emerged during the debate and signals this conversation isn’t finished.
What BIP-110’s failure does demonstrate clearly is a governance reality specific to Bitcoin in 2026: changing consensus rules now requires a level of multi-constituency alignment — miners, node operators, exchanges, large corporate holders, wallet providers — that a technically valid, procedurally correct proposal can still fail to achieve if that alignment was never actually secured beforehand. Procedural correctness is not sufficient. It never was, even in 2017 — it just wasn’t tested as visibly.
For Coldcard, the fix is concrete and individual: check whether your device and firmware fall in the affected range, and if so, migrate to a freshly generated seed rather than trusting that a firmware update alone resolves an already-compromised randomness source.
For Bitcoin more broadly, both stories point toward the same discipline. Don’t trust a check because it exists. Verify what it’s actually checking. That applies whether you’re evaluating a hardware wallet’s entropy claims or a protocol proposal’s activation mechanism.
A soft fork that mined two blocks and a firmware bug that drained $38 million don’t belong in the same news category most days. But both are, at bottom, stories about the gap between a safeguard that looks legitimate and one that’s actually been verified under pressure.
Bitcoin’s resilience doesn’t come from the fact that failures like these don’t happen. It comes from what happens after — the fork gets rejected by economic reality within days, the vulnerability gets publicly disclosed and patched, and the broader system keeps functioning because enough of its participants insist on checking rather than assuming.
That discipline, applied consistently, is worth more than any single fix.
Related reading on this Substack:
$38 Million Gone in 25 Minutes: What the Coldcard Vulnerability Actually Means — The full technical breakdown (free)
Is Lightning Really Broken? The Post-Quantum Debate — Another case of separating real risk from overstated headlines (free)
Are Governments Bitcoin’s Enemies? No, Indifference Is. — On Bitcoin’s real threats versus the loud ones (free)
How to Set Up a Bitcoin and Lightning Node — Why running your own verification matters (free)
Zap me: zap@shadowbip.com
GitHub: https://github.com/shadowbipnode
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.