cd ~/blog && cat voting-with-blocks.md
Voting with blocks
How the rules change in a system nobody owns
On 8 August 2026, Bitcoin split in two. For a few hours there were two chains, each with its own rules. By the next day, one of them had stopped moving, and the vast majority of users hadn’t even noticed.
How can something like that happen without anyone ordering it or stopping it? To understand it, we need to start with a simpler question: who decides Bitcoin’s rules?
Rules nobody can impose
In almost any digital service, the rules are changed by whoever owns it. A company ships an update and, from that moment on, every user runs the new version.
Bitcoin has no owner. Everyone who runs a node chooses which software to use and, with it, which blocks to accept as valid. If someone wants to change a rule, they can’t impose it: they have to convince thousands of independent people to adopt it on their own.
That’s why Bitcoin’s rules change so rarely. The last change to be activated was Taproot, in November 2021.
Soft forks and hard forks
There are two ways to change the rules, and the difference matters.
A hard fork widens what’s allowed. Nodes that don’t upgrade reject the new blocks, so everyone has to switch at the same time. If part of the network doesn’t, it splits in two for good.
A soft fork tightens the rules: it forbids something that used to be allowed. Old nodes keep accepting the new blocks, because they also follow the old rules. It’s the preferred way to change Bitcoin, because nobody is forced to upgrade to keep working.
Voting with blocks
For a soft fork to be safe, most miners need to enforce the new rules. Otherwise, blocks that break them would keep reaching the network.
That’s what signalling is for: each miner can mark in their blocks that they’re ready for the change. The usual standard requires 95% of the blocks in a period of roughly two weeks to signal. Taproot used a 90% threshold, reached it in June 2021 and activated months later, to give everyone time to upgrade.
There’s another route, the user-activated soft fork (UASF). Instead of waiting for miners, a group of nodes sets a date after which they will reject blocks that don’t signal. It works if those nodes represent a large enough share of the economy: if miners don’t follow, their blocks stop counting for the people who actually use the network.
August 2026
BIP-110 proposed limiting the size of certain data fields in transactions for one year. It was designed with two mechanisms at once: signalling with a 55% threshold and, if that wasn’t reached, a mandatory signalling period from block 961,632. During that period, nodes enforcing BIP-110 would reject any block that didn’t signal.
Signalling peaked at 2.53% of blocks. On 8 August, the mandatory period began. The first block arrived without signalling, like almost all of them, and BIP-110 nodes rejected it. From then on, they followed their own chain.
That chain produced two blocks and stopped: it had hardly any miners behind it. The main network carried on as normal, with no reorganisations and no lost funds. Today the official proposals repository lists BIP-110 as closed.
A system that defends itself
The most interesting thing about August 2026 is how little happened. No referee was needed, no official vote, no company making the call. Each participant enforced the rules they chose, and the network followed the path the vast majority was already on.
That’s the underlying guarantee: in Bitcoin, the rules only change when there’s real, broad consensus. For anyone keeping their savings on the network, that slowness is exactly what makes it reliable. As with any solid standard, the rules hold because almost nobody can change them alone.