Bitcoin

BIP-110 Explained And Why Bitcoin Is Divided

10 min read
MH
Written by Mohamed Habbat

I work in the crypto self-custody space, and when BIP-110 hit the news the question I got was rarely "who is right." It was "do I need to do anything." For a holder on the main chain the answer was no, and it still is.

Anyone can write a Bitcoin rule. Almost no one can make the network enforce it. BIP-110 is the clearest recent example. It proposed a temporary consensus rule to cap arbitrary on-chain data, the layer behind Ordinals inscriptions. On 8 August 2026 its enforcing nodes rejected block 961,632, split onto their own chain, and stalled, and the BIP editors marked the proposal Closed the next day.

What follows is the mechanics of the proposal, what happened at the split, and the shape of the disagreement.

TL;DR

BIP-110, titled Reduced Data Temporary Softfork (RDTS), was a soft fork that would have tightened Bitcoin's rules on arbitrary data for 52,416 blocks, about a year, per the BIP text. It exempted every UTXO created before activation. Miner support never came close: 51 of 2,016 blocks signalled in the last retarget period before the deadline, 2.5 percent, counted on mempool.space against a 55 percent threshold. The enforcing nodes split off and stalled, and the main chain carried on. For most holders the required action was nothing.

What BIP-110 proposes

BIP-110 targets one thing: the volume of arbitrary, non-financial data that transactions can commit to the chain.

FieldDetail
TitleReduced Data Temporary Softfork (RDTS)
BIP number110, assigned 3 December 2025 (BIP text)
AuthorDathon Ohm. Luke-Jr is credited with the original draft
TypeConsensus soft fork
TargetLarge data pushes and witness items, large output scripts, OP_RETURN above 83 bytes, OP_IF in Tapscript (BIP text)
Duration52,416 blocks, about one year after activation (BIP text)
ScopeExempts UTXOs created before activation. Applies going forward
StatusClosed on 9 August 2026, "following a chain split with stalled mining" (bitcoin/bips #2245)

The specification lists seven rules. The ones that matter most: new output scripts above 34 bytes become invalid, except OP_RETURN outputs up to 83 bytes (BIP text). Data pushes and script argument witness items above 256 bytes become invalid, per the same rule list. Tapscripts that execute OP_IF or OP_NOTIF become invalid. The rest close off upgrade hooks: the Taproot annex, undefined witness versions, OP_SUCCESS opcodes, and Taproot control blocks above 257 bytes (BIP text).

Why OP_IF? Ordinals store inscription content in Taproot script-path spends, wrapped in an "envelope" of OP_FALSE OP_IF ... OP_ENDIF, as the Ordinals handbook documents. Banning OP_IF in Tapscript breaks that envelope for new outputs. The BIP itself says OP_IF "is commonly abused today to inject spam that gets skipped at execution." It does not aim at tokens: the BIP text says non-Bitcoin tokens "are best countered in policy rather than consensus."

Two design choices soften the blow. The rule was temporary, so it was framed as a circuit breaker rather than a permanent change. And it exempted existing UTXOs: the BIP says inputs spending UTXOs created before the activation height "are exempt from all of the new rules." The BIP does admit edge cases where funds could be frozen, all involving pre-signed Taproot transactions that use OP_IF or script trees deeper than 7 levels.

How activation was meant to work

BIP-110 uses a modified version of the BIP-9 version-bits mechanism on bit 4, per the BIP text. Miners set that bit in the blocks they mine to signal readiness. Once enough blocks in a 2,016-block retarget window signal, as BIP-9 defines it, the rule locks in and activates one window later.

The unusual part is the threshold. Standard BIP-9 uses 1,916 of 2,016 blocks, 95 percent, and Taproot used 1,815 blocks, 90 percent, per BIP-341. BIP-110 set 1,109 of 2,016 blocks, 55 percent, per the BIP text.

It also had no timeout. If miners did not lock it in, the BIP made signalling mandatory from block 961,632 to block 963,647: enforcing nodes reject any block in that window that does not set bit 4. Activation would then come at block 965,664 at the latest, which the BIP estimated as about 1 September 2026. A design like this works only if most of the hash rate follows. If it does not, the enforcing nodes leave.

What happened at the split

Miners did not signal. I counted bit 4 in every block of the last three retarget periods before the mandatory window, using the mempool.space block API. The first block of each period links to the explorer.

Retarget periodDates (UTC)Blocks signalling bit 4Share
955,584 to 957,59927 Jun to 11 Jul 202620 of 2,0161.0%
957,600 to 959,61511 Jul to 25 Jul 202626 of 2,0161.3%
959,616 to 961,63125 Jul to 8 Aug 202651 of 2,0162.5%

The threshold was 1,109 blocks (BIP text), and the best period reached 51.

Block 961,632 was mined at 19:35 UTC on 8 August 2026 with version 0x20006000, which does not set bit 4. Every node enforcing BIP-110 rejected it. Bitcoin Knots had shipped the rules as an opt-in in v29.3.knots20260508, switched on with consensusrules=rdts, so those were the nodes that left. The BIP editor's status change, bitcoin/bips #2245, records it in one line: the BIP "split to a new chain and stalled." Bitcoin Core never added the rules; the pull request that would have done so, bitcoin/bitcoin #34930, was closed without merging.

The two camps

The disagreement is about whether Bitcoin should use a consensus change to settle what one side calls spam and the other side calls valid, fee-paying use of block space.

CampCore argumentWhere it is written
AgainstTurning a spam dispute into a consensus change would invalidate transactions that are valid and fee-paying today, and that precedent is the danger.Michael Saylor on X
ForData storage competes unfairly with payments and puts a cost on node operators who never received the fee, so it should be rejected at the consensus level.The BIP-110 motivation by Dathon Ohm

On the against side, Michael Saylor posted on X on 11 July 2026: "There are 110 things more dangerous to Bitcoin than spam. BIP 110 turns a spam dispute into a consensus change that would invalidate some currently valid, fee-paying transactions. That precedent is the danger." His objection is procedural as much as technical. The worry is less about this rule and more about a precedent where consensus filters which valid transactions the network accepts.

On the for side, the BIP makes its own case. It says the fee for a data transaction "goes only to the miner," while every node operator downloads, stores and serves the data forever. It argues data storage "competes unfairly with payments, making Bitcoin payments unnecessarily costly." I attribute positions only where they are written down by the person holding them. Much of the wider debate ran on forums and X threads I could not verify, so I leave it out.

How Bitcoin changes its rules

A BIP is a document. Anyone can write one and open a pull request, and only the BIP editors assign it a number, per BIP-3. That act changes nothing about the network by itself.

A rule only becomes real when the people who run Bitcoin choose to enforce it. Those people are the economic full nodes and the miners. No committee, foundation, or company can push a consensus change onto the network, because there is no one in charge who could. This is the point people miss about why Bitcoin changes slowly: without a central authority, coordinated change is hard.

The two kinds of change matter here.

A soft fork tightens the rules. Blocks valid under the new, stricter rules are still valid under the old rules. The Bitcoin developer guide states the condition: the chain stays whole if upgraded nodes control a majority of the hash rate. SegWit, deployed under BIP-141, activated as a BIP-9 soft fork at block 481,824, per Bitcoin Core's chainparams.cpp, and Taproot activated at block 709,632, per BIP-341. BIP-110 was designed as a soft fork too.

A hard fork loosens or breaks the rules. Blocks valid under the new rules are rejected by old nodes, so unless everyone upgrades together, the network can split into two chains. For a different contested proposal, read BIP-360 and BIP-361, where the debate is about quantum-vulnerable coins rather than data.

Running your own node is where you hold power. Your node enforces whatever ruleset its software encodes. In my view, if you run your own node, the software you choose is your vote on which rules Bitcoin follows, and the sum of those choices is what activation measures.

Soft fork or hard fork and the split risk

BIP-110 was a soft fork. That framing matters for the split question, because the two fork types carry different risks.

Soft forkHard fork
Rule directionTightens rulesLoosens or breaks rules
Old nodesStill follow the chainReject new blocks
Split riskLow if the majority of hash rate enforces itHigh unless everyone upgrades
ExamplesSegWit, TaprootBitcoin Cash, split on 1 August 2017 at block 478,558 (bitcoincash.org)

Most people fear the wrong split. A soft fork with majority hash rate leaves one chain. A soft fork with minority hash rate and a mandatory signalling rule splits off its own enforcers, and that is what BIP-110 did. With 2.5 percent of blocks signalling in the last period (mempool.space), the enforcing nodes rejected the main chain's next block and were left waiting for blocks that miners did not produce. A minority soft fork with almost no hash rate behind it does not reshape the main chain; it forks off a fringe.

What this means for you

  • Most holders needed to do nothing, and still do. The rules never activated on the main chain, and normal payments were not the target.
  • The actors who decide are node operators and miners, not holders, exchanges, or commentators. Activation is measured in signalling and node enforcement.
  • If you run Bitcoin Knots with consensusrules=rdts from v29.3.knots20260508, compare your node's block height (bitcoin-cli getblockcount) with a public explorer such as mempool.space. A node stuck near the split is not on the main chain, and its wallet balances and confirmations are not the main chain's.
  • UTXOs that existed before any activation were exempt, so a pre-existing balance was never the thing the rule reached for.

This is not investment advice. I wrote it to show how the rule worked and why the miners' answer settled it.

References

Frequently Asked Questions

What is BIP-110?
BIP-110 is a Bitcoin Improvement Proposal titled Reduced Data Temporary Softfork (RDTS), written by Dathon Ohm with Luke-Jr credited for the original draft. It proposed a temporary consensus rule that caps arbitrary data in new transactions. The BIP text now carries the status Closed, set on 9 August 2026 after a chain split with stalled mining.
What does BIP-110 actually propose to change?
It makes new outputs larger than 34 bytes invalid (OP_RETURN up to 83 bytes), caps data pushes and script argument witness items at 256 bytes, and bans OP_IF in Tapscript, the opcode the Ordinals inscription envelope uses. The BIP sets the rules to expire after 52,416 blocks, about one year. UTXOs created before activation are exempt.
Is BIP-110 a soft fork or a hard fork?
It is a soft fork. A soft fork tightens the rules, so blocks valid under the new rules stay valid under the old ones. The Bitcoin developer guide adds the condition: the chain stays whole only if upgraded nodes control a majority of the hash rate. SegWit and Taproot were deployed the same way.
Did BIP-110 split the Bitcoin chain?
It split off its own nodes, not Bitcoin. On 8 August 2026 the first block of the mandatory signalling window, block 961,632, did not signal bit 4, so nodes enforcing BIP-110 rejected it. The BIP editors recorded that the enforcing nodes split to a new chain that stalled. The main chain carried on.
Is BIP-110 dangerous to my coins?
For an ordinary holder on the main chain, no. The rules never activated there. Even if they had, the BIP exempts every UTXO created before activation and names only edge cases (pre-signed Taproot transactions using OP_IF or deep script trees) where funds could be frozen. If you ran Bitcoin Knots with BIP-110 enforcement switched on, check that your node is on the main chain.
Who supports and who opposes BIP-110?
The author, Dathon Ohm, argues in the BIP that data storage competes unfairly with payments and burdens node operators. Bitcoin Knots shipped the rules as an opt-in. On the opposing side, Michael Saylor wrote on X that BIP 110 'turns a spam dispute into a consensus change that would invalidate some currently valid, fee-paying transactions' and that 'that precedent is the danger.'
How does this compare to other protocol events like the halving?
The halving is already coded into consensus and happens automatically on schedule, with no debate required. BIP-110 was the opposite: a proposed rule change that only takes effect if node operators and miners choose to adopt it. One is automatic and settled, the other was contested and optional.