The "spam filter" fork needs 55% of miners and has roughly 1%. With its voluntary deadline weeks away, the loudest protocol fight of the year is quietly resolving itself — and teaching a lesson about how Bitcoin changes.
Miners vote with the blocks they mine — and almost none are voting for BIP-110.
Bitcoin's most argued-about proposal of 2026 is failing the only vote that counts. BIP-110 — the "Reduced Data Temporary Softfork" published in December by a pseudonymous developer writing as Dathon Ohm — needs 55% of blocks in its signaling window to activate. Through this week, signaling has hovered around 1% of blocks, most of them from a single small pool. No major
mining pool has endorsed it, and its voluntary lock-in window closes in early August.
The proposal's goal is blunt: make Bitcoin temporarily hostile to arbitrary data. For roughly one year — 52,416 blocks — it would cap OP_RETURN outputs at 83 bytes, limit data pushes to 256 bytes, and squeeze out the inscription-style techniques that stuff images and tokens into the blockchain. Supporters call that data spam that crowds out payments, bloats the chain every
node must store, and drifts Bitcoin away from money. Opponents answer that fee-paying transactions are valid by definition, and that deciding which ones are "legitimate" is exactly the kind of judgment Bitcoin exists to avoid.
What makes BIP-110 genuinely radical isn't the filter — it's the activation math. A traditional
soft fork waits for 95% of miners to signal readiness. BIP-110 lowers the bar to 55% and, failing that, contemplates a user-activated path in which enforcing nodes simply reject non-signaling blocks. That mechanism deliberately echoes 2017's UASF standoff — except this time the economic pressure never materialized. Bitcoin Core declined to implement the rules; the proposal lives only in the alternative Bitcoin Knots client, which a minority of nodes run.
The commentary has been lopsided too. Adam Back argued the fork's low threshold risks splitting the network over a temporary policy preference. Strategy's Michael Saylor was more dismissive, quipping this month that there are "110 things more dangerous to Bitcoin than spam." Even sympathetic developers concede the practical point: with inscription activity far below its 2024 peak and fees near cycle lows, the emergency the fork was written for is hard to locate on a
mempool chart.
If the window closes without lock-in, BIP-110's fallback phases keep a mathematical path open into the fall, but a fork activating against 99% of hashpower would create a minority chain almost immediately — an outcome even most supporters say they don't want. The likelier ending is quieter: the proposal expires as a marker of how high the bar for changing Bitcoin really sits. That bar frustrated the fork's authors. It is also, veterans of 2017 would note, the product's core feature.
BITCOIN ALMANACK ANALYSIS
Signaling vs. the 55% bar
ACTUAL SIGNALING, PAST WEEK~1%
VOLUNTARY DEADLINE
Early Aug
LOCK-IN WINDOW
MAJOR POOLS FOR
0
OF THE TOP POOLS
DURATION IF LIVE
~1 year
52,416 BLOCKS
Signaling share approximate, from public block-signal trackers, week to July 18 · Chart: Bitcoin Almanack
WHY IT MATTERS
Whatever you think of data on the blockchain, BIP-110 is a live demonstration of Bitcoin's governance: no company, foundation, or developer group could impose the change, and neither could a motivated minority. For holders, the reassuring read is that the same immune system that blocks a filter you might like also blocks changes you'd hate — including ones aimed at the 21 million cap.
What to watch next
1.The early-August lock-in deadline. If the voluntary window closes near 1%, the story is effectively over — watch whether proponents concede or pivot to the mandatory-signaling phase.
2.Knots node share. A sustained rise in Bitcoin Knots nodes would signal the fight moving from miners to node operators, 2017-style. It hasn't happened yet.
3.A successor proposal. The data-on-chain argument predates BIP-110 and will outlive it; expect a rewritten version with consensus-friendlier activation.