Final week, in a second of uncharacteristically sneaky habits, a small group of builders tried to quietly change the default mempool coverage of Bitcoin Core, the world’s dominant software program for full nodes.
After re-introducing a failed 2023 proposal to boost the information storage functionality of OP_RETURN outputs, critics flagged down the 2025 copycat earlier than it merged into manufacturing. Colloquially, some are referring to the stunning incident as a form of bitcoin OP_RETURN battle.
Mononaut joked that it was akin to forking Bitcoin for quantum resistance, skipping the mailing record and BIP course of, and leaping straight into merging code into manufacturing.
Critics referred to as the Peter Todd-written pull request (PR) 32359 chaotic, insane, malicious, no consensus, shenanigans, and vandalizing. However, supporters stated the PR would standardize mempool coverage and modernize transactions that always catalogue arbitrary information elsewhere, even earlier than this OP_RETURN proposal.
In the end, considerations about censorship of opposing viewpoints, deprioritizing bitcoin’s monetary utility, and undisclosed company pursuits have halted PR 32359 from merging into mainnet.
OP_RETURN battle for hurt discount
Backfooted, the camp in favor of lifting OP_RETURN’s datacarrier restrict was fast to recast its actions as a valiant try at “harm reduction” for the long-term good thing about Bitcoin.
Devs from this camp defined their proposal with easy language. Permitting customers to stuff massive quantities of information into blocks by way of OP_RETURN was merely standardizing an already frequent follow of storing non-financial information into unconventional components of bitcoin blocks like Taproot outputs.
From their perspective, customers already retailer random textual content, pictures, laptop code, and different non-financial information in bitcoin blocks anyway. Worse, lots of them reap the benefits of the SegWit witness low cost to pay considerably discounted storage charges.
Why insist on OP_RETURN’s 83 byte limitation when plentiful choices exist elsewhere? Absolutely, it could scale back hurt by normalizing OP_RETURN with different information storage choices.
Learn extra: The company pursuits taking up Bitcoin growth
Gallantly, a few of them claimed that lifting OP_RETURN’s datacarrier restrict was tantamount to “harm reduction.” It could, of their view, enable OP_RETURN payloads of 100 bytes or above “instead of Citrea forging ahead with writing permanently to the UTXO set.”
Citrea’s 100 byte payloads – 17 bytes bigger than OP_RETURN’s restrict
Enterprise capitalist-backed bitcoin challenge Citrea, in line with senior bitcoin developer Peter Todd, must publish 100-byte information packets for sure operations. Sadly, it could publish unprunable outputs as an alternative of the extra fascinating OP_RETURN, resulting from OP_RETURN’s 83-byte measurement restrict.
Writing completely to the set of unspent transaction outputs (UTXO), Citrea proliferated the variety of UTXOs – requiring full nodes to obtain and use invaluable computation to validate an ever-increasing amount of Citrea-created UTXOs.
This was the “harm” that PR 32359 would have “reduced,” within the view of Mark “Murch” Erhardt on an in depth StackerNews thread.
Learn extra: Moderators censor Bitcoin devs as OP_RETURN battle rages on
The Chaincode Labs bitcoin developer defined that lifting OP_RETURN’s datacarrier restrict would enable company entities like Citrea to conduct its operations extra effectively – with out requiring full node operators to obtain and validate extreme portions of unprunable UTXOs.
Peter Todd, the creator of PR 32359 on the request of Chaincode Labs’ Antoine Poinsot, affirmed, “ I was asked to open [PR 32359] by an active Core dev because entities like Citrea are using unprunable outputs instead of OP_Return, due to the size limits.”
Jameson Lopp says there isn’t a battle of curiosity
Jameson Lopp, a Citrea investor and advocate for PR 32359, denies that his funding in Citrea created a battle of curiosity concerning OP_RETURN’s datacarrier restrict.
In accordance with his cited rationalization, “The Citrea protocol does not benefit from this change at all! We’re just asking them nicely to please use OP_RETURN and not bloat the UTXO set.”
In any case, the request to vary this vital default worth of Bitcoin Core software program – to not point out proscribing the flexibility of customers to self-configure that worth as self-sovereign Bitcoin Core node operators – has resulted in a quick form of OP_RETURN battle amongst bitcoin maintainers this month.
Arguments continued over the weekend and can probably proceed this week as a frenzy of social media throughout GitHub, the Bitcoin-Dev Google Group, StackerNews, Reddit, BitcoinTalk, X, and elsewhere.
Obtained a tip? Ship us an electronic mail securely by way of Protos Leaks. For extra knowledgeable information, comply with us on X, Bluesky, and Google Information, or subscribe to our YouTube channel.