CAPTURE

An investigation into how informal power over Bitcoin Core was assembled, exercised, and defended

by hodlonaut | June 15th, 2026

Article Three of Four — The Merge

Part I: The Original Sin

Marco Falke had been a Bitcoin Core maintainer since 2016. OKCoin had funded him annually since 2020. In January 2023, OKCoin and Paradigm announced a joint grant funding Falke, marking Paradigm's first direct funding of him. On February 16, 2023, the Wall Street Journal published a feature on Bitcoin Core's funding structure and the covert-fix practice that explicitly profiled Falke's role and identified Paradigm as a new funder. Five days later, on February 21, 2023, Falke announced he was stepping down as Bitcoin Core maintainer. In his June 2026 reply to this investigation, Falke stated that the WSJ feature did not factor into his decision. He pointed instead to the Craig Wright legal cases (the Tulip Trading appeal decision in February 2023 and the COPA "Database" case defence preparation in March 2023) as the actual drivers of his departure: "I simply didn't have the time nor energy back then to continue as a maintainer."

Paradigm is one of the largest crypto venture capital firms, co-founded by Fred Ehrsam and Matt Huang. Its portfolio of Ethereum, DeFi, and Web3 investments positions it commercially in categories that compete with or benefit from Bitcoin's data-layer policy direction. Across its history of public Bitcoin Core developer grants, Paradigm has funded exactly two developers: AJ Towns in July 2020 and Marco Falke in January 2023. Both were among the six participants on the documentation change (PR #27832) in June 2023. Falke filed it. Towns ACKed it.

The change itself was surgical. The -datacarriersize option had described its scope as "data carrier transactions," a broad category. PR #27832 narrowed it to "data-carrying raw scriptPubKey," meaning only the scriptPubKey outputs of transactions, specifically OP_RETURN outputs. Witness data, which Taproot inscriptions use, was excluded from the setting's operative description. Falke, in his June 2026 reply to this investigation, said the amendment was a documentation update to match existing code behaviour that had been in place since 2015. Whether the 2015 code already limited the option to push-only OP_RETURN payload or whether the amendment narrowed the option's effective scope, the documentation change introduced specific narrower public-facing language.Anyone relying on the documented scope to argue for policy would henceforth be arguing from a foundation that had shifted.

The participant list on the documentation change PR was small, but included key people who would play significant roles in the subsequent OP_RETURN uncap sequence. Falke filed it. Peter Todd posted a NACK on June 7, objecting that the new wording was misleading about current behaviour. Gloria Zhao added a “Good First Review” label the same day. AJ Towns gave a Concept ACK on June 6 and a full ACK on June 23. Greg Sanders ("instagibbs") ACKed on June 6 and again on August 3.

Michael Ford ("fanquake") merged the documentation change PR on August 3, 2023 with only two ACKs and over Todd's stated objection.

Starting in May 2023, the month before the documentation change PR was filed, Brink fellow Gloria Zhao and Chaincode's Mark Erhardt (“Murch”) co-authored the limited weekly series about transaction relay, mempool inclusion, and mining transaction selection called “Waiting for Confirmation”. It would become a weekly feature in the Bitcoin Optech newsletters for the next ten weeks.

In Part 7 (Newsletter #257, June 28), the argument appeared that would be used to justify the OP_RETURN limit removal two years later: data publication on-chain cannot be prevented, so at least it should not live in the UTXO set forever. The harm reduction justification Zhao would deploy in 2025 was first published by Zhao in Optech the month before the documentation change was made.

The newsletter recap podcast for Bitcoin Optech Issue #257, hosted by Mike Schmidt (Executive Director of Brink and Bitcoin Optech) on June 29, 2023, had Murch and Zhao as guests. Murch restated the argument in his own words on air: "people realize that it is impossible to prevent people from publishing data on the blockchain... And when you cannot prevent people from doing that, we can at least maybe encourage some best practices where we get people, if they need to do it, to do it in a way that it doesn't pollute the UTXO set."

Zhao stated, on the same episode, the position she would deploy in her October 2023 rejection of Luke Dashjr's patch and in her January 2024 NACK summary: "as protocol developers, I don't think we have a place in saying what is legitimate payment and what is legitimate data to stuff into the blockchain, and what is wasteful or not real payments. I think that goes against the whole philosophy of Bitcoin."

The position was contested inside the same podcast by Dave Harding, the lead writer of Bitcoin Optech: "I kind of want to disagree with Gloria on this... maybe we do have a good reason to make decisions based on use case." Zhao narrowed the claim in response: "sorry, what I meant by 'we' is as protocol developers. I don't think it would be appropriate to have consensus rules that as devs we impose on people." The general principle she had stated was, when challenged, restricted to consensus rules. Policy decisions were left explicitly open.

The narrowed version did not appear again. In her January 2024 rejection of Luke's filter patch, in her April 2025 Bitcoin Magazine interview, and in the June 2025 merge of the institutional uncap PR (PR #32406), she invoked the principle as general.

Six days after the documentation change was merged, Optech's Newsletter #263 published its recurring 'Notable code and documentation changes' section. It listed one Bitcoin Core change for the week: PR #27746, a routine refactor of block storage logic. The documentation change (PR #27832) was not listed.

The recap podcast for Newsletter #263, hosted by Brink's Mike Schmidt with Chaincode's Murch, Optech's Dave Harding, and Peter Todd on August 10, 2023, walked through the newsletter on the same day it was published. The OP_RETURN policy area was an active segment: Todd was on the podcast specifically to discuss his separately-filed PR #28130 to remove OP_RETURN limits. The documentation change was not mentioned at any point in this episode either.

The documentation change that would later become the cited basis for rejecting Luke's filter patch was absent from the Optech surfaces that covered the relevant week. It did not appear in the written newsletter section dedicated to notable code and documentation changes, nor in the recap podcast discussing that edition.

Four months later, the revised documentation became a recurring reference point in arguments against extending -datacarriersize to witness data. Whether viewed as a clarification or a redefinition, the change became the foundation on which much of the subsequent sequence rested.

Part II: The Rejection

‍On December 6, 2023, Luke Dashjr posted a public service announcement on X. It got 4.6 million impressions. “Inscriptions are exploiting a vulnerability in Bitcoin Core to spam the blockchain,” he wrote. “Bitcoin Core has, since 2013, allowed users to set a limit on the size of extra data in transactions they relay or mine (-datacarriersize). By obfuscating their data as program code, Inscriptions bypass this limit.”

He named the fix. The bug had been corrected in Bitcoin Knots version 25.1, released three weeks earlier. Bitcoin Core’s upcoming v26 release was still vulnerable.

Three days later, on December 9, 2023, the bypass was assigned the identifier CVE-2023-50428 and published at the National Vulnerability Database. NIST scored it 5.3, medium severity, network-attackable, low complexity. The classification was formal and on the record.

Luke had not started here. He had spent the year before trying to fix the bug in code.

On February 1, 2023, seven weeks after the first Ordinals inscription appeared at block 767430, Luke published a quick patch as a GitHub Gist to filter the inscription bypass. The patch was distributed via community channels and later incorporated into Bitcoin Knots, but was never filed as a PR to Bitcoin Core.

In September 2023, Luke filed his filter patch PR #28408, a more substantial, merge-eligible version, to Bitcoin Core itself.

The configuration setting it modified (-datacarriersize) had existed since February 2015 as the node operator’s lever for limiting how much arbitrary data their node would relay, and had originally been a single dial covering all forms of data embedding.

Two soft forks since 2015 had added new transaction data fields the original -datacarriersize setting did not cover. SegWit, activated in 2017, introduced witness data as a separate field. Taproot, activated in 2021, expanded what could be embedded there via the Tapscript mechanism. The Ordinals inscription protocol, launched January 2023, used Taproot's script-path data to embed arbitrary content outside the -datacarriersize limit's reach. Inscriptions did not exist when the limit was designed. Inscriptions embedded their data in fields the setting had never covered.

Luke’s filter patch proposed to restore its original scope so that it would once again do what node operators had always understood it to do: limit the data carried in any relayed transaction, including the witness and script-path data that inscriptions exploited. Not a new policy. A return of the setting to its original scope, applied to include the new fields inscriptions were using.

Luke was not aware that the documentation had been amended just a month earlier. In direct messages to this investigation, he said: “No, I have no recollection of being told about the doc change. The doc change was done sneakily, buried in a PR described as doing something else, and it seems nobody even noticed it happened until after the rejection of 28408.”

Luke’s filter patch sat open while Luke continued to argue the case. Nobody told him the documentation had changed.

‍ ‍

By November 15, 2023, Luke had shipped the fix in Bitcoin Knots. The December 6 PSA was the public announcement that Knots had solved the problem and that Bitcoin Core had not. Luke confirmed the CVE assignment publicly on December 10.

The CVE-2023-50428 record stood unchallenged at the National Vulnerability Database for twenty-six days.

Casey Rodarmor, who launched the Ordinals protocol in January 2023 and later launched Runes in April 2024, commented on Luke’s filter patch on November 30, 2023. In his post, he quantified miner revenue from inscription transactions at $121,386,535 over the prior ten months and argued that the proposed filter would be “long-term incentive-incompatible for miners.”

He characterized the patch’s impact on inscription commit transaction outputs as “a mild form of confiscation,” warning that it would cause miners running the updated software to lose substantial fee revenue and become uncompetitive.

At the time, Rodarmor leaned heavily on the narrative that inscription activity had become a major and sustainable new source of miner revenue, one that miners would (and should) be unwilling to forgo. However, this revenue proved to be a short-lived hype-cycle phenomenon.

By May 2024, daily inscription fees had already imploded dramatically, and it currently sits at only 0–0.1 BTC daily (under $10,000 USD), representing a marginal << 0.025% of total daily miner revenue. In the 18 months between November 11, 2024 and May 11, 2026, inscription revenue for miners totalled only 184 BTC. In the same period, the block subsidy totalled 232,200. This meant that inscription revenue had accounted for less than 0.08% of miner revenue. Rodarmor’s narrative about the long-term importance and sustainability of inscription revenue has been conclusively proven false.

Peter Todd characterised Luke's filter patch as censorship. On September 26, 2023, he posted twice on the PR in the same day, using the word “censoring” in both comments. Todd operates Libre Relay, a direct-to-miner bypass service for the exact transactions he was characterising as targets of censorship. He presented the loophole as the baseline, and the patch that would close it as the threat.

Like Rodarmor, Todd pushed the narrative that inscription-related fee revenue had become a major and essential long-term income source for miners that should not be restricted, writing on the PR: “Concept NACK. The transactions targeted by this pull-req are a very significant source of fee revenue for miners.”

‍ ‍

On January 4, 2024, an external contributor named Léo Haf, known on GitHub as Retropex, filed PR #29173 to revert the documentation change.

This would mark the start of strong and seemingly coordinated action from the cohort: the small, highly interconnected network of contributors, maintainers, funders, employers and moderators that this article documents repeatedly occupying decisive positions in the OP_RETURN sequence.

It began within hours. AJ Towns NACKed the documentation revert: “Docs should describe what the code does, not what we might like it to do.” His second move on the documentation, following ACKs he had given seven months earlier.

Marco Falke himself posted to defend his own change: “I changed the documentation to match the code in June. Otherwise, there would be a mismatch between the code and the documentation.” This was Falke's contemporaneous January 2024 position, consistent with the position he restated to this investigation in his June 2026 RoR reply: that the amendment documented existing code behaviour rather than altering the option's effective scope.

Marcel Hernandez (a Bitcoin developer who maintained public Bitcoin Core configuration references and who posted on X under the handle @oomahq) commented under the Github handle 1ma: “The change that is being reverted was a covert attempt at shrugging off the responsibility of addressing CVE-2023-50428 in Bitcoin Core. Exploits must be addressed by changing the C++ code, not the project’s documentation.”

Gloria Zhao NACKed the revert: “NACK, the current documentation is more accurate to what the code actually does. If you want to change what the code does, that's already its own PR #28408.” The PR she pointed Retropex to as the place to change what the code does was Luke’s filter patch, the same PR she would formally close the following day. She then immediately closed the documentation revert PR. It was then locked “as too heated and limited conversation to collaborators”, still on January 4, the day of filing.

On the same day, the CVE-2023-50428 record at the National Vulnerability Database was updated. Twenty-six days of unchallenged formal classification ended on the same calendar day Retropex’s revert was killed. The cohort’s response had operated across two channels: the repository and the formal vulnerability-tracking system. The CVE description was administratively amended to include the framing: “although this is a vulnerability from the perspective of the Bitcoin Knots project, some others consider it ‘not a bug.’” The cohort’s position is embedded in the canonical NVD record.

The next day, on January 5, 2024, Gloria Zhao posted her structured NACK summary on Luke’s filter patch. The summary was the closing document of the rejection. In it, she stated the principle she had articulated on the Optech podcast in June 2023, now deployed in her capacity as the maintainer who had rejected the patch: “People are calling this "spam" because they don't like ordinals. While I also personally think Bitcoin is best used for financial transactions, I disagree with this framing of "spam." In transaction validation and relay logic, we should define spam on the basis of network resources consumed and whether those costs are well bounded and/or paid for, not on the basis of use case. We have no way of detecting whether a bunch of bytes are real/legitimate/useful and, even if we could, it is not protocol development's place to decide which use cases of Bitcoin are legitimate and which ones aren't.”

The narrowing she had offered Harding on the June 2023 podcast, that the principle applied only to consensus rules and not to policy decisions, did not appear.

Immediately after Zhao had posted her structured summary, an external commenter named jangorecki replied on the same thread: "You missed the most important, censorship." Zhao did not respond. The word Peter Todd had used twice on the PR three months earlier was flagged as missing from her structured summary by a commenter on the day of closure. It remained absent from her documented reasoning. The framing would reappear seventeen months later, in the 31-developer letter published three days before she merged the institutional uncap (PR #32406).

Two hours after Zhao’s NACK summary, Mark Erhardt (“Murch”), Chaincode Labs employee and Zhao's co-author on the June 2023 Optech "Waiting for Confirmation" series, posted his Concept NACK. His opening definitional sentence stated: "datacarriersize refers to the payload of OP_RETURN outputs, the output type that was introduced as a data carrier to provide an alternative for more harmful ways of embedding data." This is the post-amendment definition. Murch built his rejection on the narrowed scope without acknowledging that the scope had been narrowed five months earlier.

An hour later, AJ Towns, who had ACKed the documentation change PR seven months earlier, gave an Approach NACK on Luke's patch. His stated reason was that the patch "couples the policy on OP_RETURN output size and the new limits on data embedded in redundant script code, so that it's impossible to configure your node in a way that matches the default policy for 26.0 and earlier releases."

Towns was citing the v26.0 default policy as the baseline Luke's patch violated. The v26.0 default policy was what Falke's documentation amendment had established. Towns had ACKed this change himself. He did not disclose it in the comment.

Luke’s patch had been designed against the description that preceded it. The rejections did not acknowledge that the description had changed.

15 minutes after Towns’ NACK, Ava Chow (“achow101”) formally closed Luke's filter patch (PR #28408). At the moment of closure, the PR stood at twelve Concept NACKs against nine Concept ACKs among named technical reviewers.

Still on January 5, 2024, Luke filed GitHub Issue #29187, the formal Bitcoin Core repository record of the vulnerability, CVE-2023-50428.

Gloria Zhao opposed the issue the same day. Her stated grounds: the scope of -datacarriersize was limited to scriptPubKey outputs, and could not be used in the way Luke’s patch proposed. She cited the documentation. The documentation she cited was five months old. It had been introduced in August when fanquake merged PR #27832.

The vulnerability Issue Luke had filed remained open until it was closed October 15, 2024, by Ava Chow as “completed”. The administrative status had changed. The underlying vulnerability in Bitcoin Core's code was untouched.

Luke had not been able to fix the bug in Bitcoin Core's code. He had fixed it in Bitcoin Knots. He had obtained the CVE designation at the National Vulnerability Database. He had filed his filter patch and seen it closed. The Issue #29187 filing is a formal record, kept inside the project's own infrastructure, of what the project had refused to fix.

Part III: The Burial

Antoine Poinsot joined Chaincode Labs on October 14, 2024. The same day he joined, Poinsot posted on the GitHub Issue thread for Luke's CVE-2023-50428 designation: “I personally believe the CVE system is being abused in this case. I believe this issue should be closed.” Also that day, Pieter Wuille announced in #bitcoin-core-dev that the following day's CoreDev gathering would hold its annual PR kill/shill/merge session.

The next day, October 15, 2024, Ava Chow executed fourteen PR closures in a single afternoon. Three of them were inscription and data-carrier related.

At 15:01 UTC, she closed PR #29942 (“Remove redundant -datacarrier option,” filed by vostrnad) with the stated reason “Closing due to lack of interest.”

At 15:17 UTC, she closed PR #29520 (“add -limitdummyscriptdatasize option,” filed by Léo Haf and based explicitly on Luke Dashjr’s PR #28408 work) with the stated reason “not had any activity in a while.”

At 16:42 UTC, she closed PR #28217 (“set DEFAULT_PERMIT_BAREMULTISIG to false,” also filed by Léo Haf) with the stated reason “lacks conceptual support and is still obviously controversial.” The PR had fifteen Concept ACKs and fourteen Concept NACKs.

The same afternoon, Chow also closed GitHub Issue #29187, Luke Dashjr’s CVE-2023-50428 designation. The formal security classification was extinguished as part of a coordinated administrative sweep of policy-restrictive PRs at a closed-door developer gathering. The #bitcoin-core-dev IRC channel that processed Bitcoin Core development was silent that day. No human spoke in channel. The bot logged the closures.

The administrative chronology runs across two calendar days. On October 14, Poinsot joined Chaincode Labs and that same day posted his belief on the CVE vulnerability Issue thread that the Issue should be closed. Wuille announced the kill/shill session for the following day in the developer channel. The person whose comment had recommended the closure had joined the institution at the centre of this investigation twenty-four hours earlier. The security designation issued by the National Vulnerability Database was extinguished as one item among fourteen closures in an afternoon. The cohort's response to the formal security classification had been administrative, not technical.

The same afternoon, Chaincode Labs engineer Matthew Zipkin, one of two Bitcoin Core moderators, banned an external commenter named GregTonoski for twenty-four hours after a one-word criticism of the closure of PR #29520. The PR was then locked.

On October 31, 2024, Adam Jonas published "Evolution of Bitcoin Core Priority Projects." The CEO of Chaincode Labs, architect of the developer pipeline this investigation documents across Article 1 and Article 2, codified governance principles the cohort had already been applying in practice.

The relevant passage, on Working Groups best practices: "No requirement to seek consensus from those not actively contributing to the work. Thoughtful review and criticisms should be taken into account no matter the source, but those with “skin in the game” should ultimately have the loudest voices."

This is Jonas formally codifying the principle that had been deployed rhetorically since 2020: critics who are not active contributors do not count. Uttarwar road-tested it on the What Bitcoin Did podcast in October 2020 ("extremely extractive," "small group of loud, angry voices"). Jonas embedded it in his 2023 contributor survey as the top frustration item. It was now governance text, published seven months before it was used to dismiss a 4:1 Github community opposition ratio.

Jonas’ governance document also endorsed Signal groups as the preferred coordination mechanism: "Signal groups have proven effective and remain accessible to interested participants." Signal is end-to-end encrypted and unarchived. The private coordination channel of the network had been given formal institutional sanction.

Part IV: The Distancing

Six months after Jonas published the governance document, the institutional structure he had spent six years building was named in public discourse outside the cohort.

A 2023 Wall Street Journal investigation of Bitcoin Core's developer structure documented concerns the cohort would face again two years later. Samuel Dobson, a former Bitcoin Core maintainer, told the WSJ on the record that the funding-control question came up regularly: "It comes up quite a bit — the source of the funding and whether there's a kind of unseen control." Coinbase, the publicly-traded crypto exchange, had separately disclosed to the Securities and Exchange Commission that the way Bitcoin Core developers were organised could limit bitcoin's growth, and that changes they made to the code could affect the network's "speed, security, usability or value." The structural concerns this investigation documents had been named, by a former maintainer and a publicly-traded company's regulatory filing, in a major newspaper two years before the OP_RETURN merge.

On stage at the MIT Bitcoin Expo in April 2025, Jonas delivered a twenty-four-minute address titled "Has Bitcoin Core Lost Focus?" The talk was livestreamed by Bitcoin Magazine, six weeks before the OP_RETURN uncap merge on June 9, 2025. He then moderated a panel of three Bitcoin Core developers immediately afterward: Gloria Zhao, Antoine Poinsot, and Cory Fields (long-term core developer at MIT-DCI and Brink grant selection committee member). Across the talk and the panel he moderated, the cohort's institutional operator established a set of public positions.

Jonas described Bitcoin Core as "a diverse group of people with varying motivations and goals. Today's Bitcoin Core contributors are from all over the world. They span 6 continents. They are PhDs and others with no more than a high school education. They are big tech veterans and those that have known no other job."

The diversity Jonas described deserves a closer look. The maintainer who would merge the institutional uncap PR (Gloria Zhao, recruited by Jonas via cold email in January 2020) was a Chaincode Labs employee at the moment of the merge, having joined just ten weeks before. The contributor who disputed Luke Dashjr's CVE-2023-50428 security Issue in October 2024 (Antoine Poinsot) was a Chaincode Labs employee. The author of the moderation guidelines later used against named critics on PR #32406 was Adam Jonas himself, Chaincode's CEO. The moderator who tested those guidelines on OP_RETURN-adjacent comments, volunteered to enforce them, and executed the April 23, 2025 ban of Antoine Riard six weeks before the merge (Matthew Zipkin) was a full-time engineer at Chaincode Labs. The Brink-funded engineer who merged 56 percent of all pull requests to Bitcoin Core in 2025 and led the release process for every release from v28.1 through v30.0 was Michael Ford (“fanquake”).

The contributor diversity Jonas named at MIT was real at the contributor level. The institutional pipeline that produced and funded those contributors is concentrated.

Jonas then disclaimed his role in the project. "I sometimes have the bad habit of slipping into “we” when referring to Bitcoin Core. But I don't count myself among these Core devs. And I certainly don't represent them. My role in all this is quite limited. I'm much more water boy than core developer."

On the panel that followed, Cory Fields disagreed that Jonas had a “quite limited role”: "So I'd quickly like to ask Jonas, because I've heard you give your spiel a couple times about how you're not one of us and you're the waterboy and the helper and all that, but that's bullshit."

Jonas defended his framing: "That's not bullshit. I am an outside observer… I'll never be the one that has, that runs the script. So it sort of absolves me from the ultimate load. And I think, you know, I'm, that makes me again, a muggle and a waterboy like most of the ecosystem."

Jonas also stated in his presentation: "yet we have the equivalent of a volunteer fire department working on its infrastructure."

The volunteer fire department framing does not survive contact with the funding record. At least seven institutions employed or funded the senior Core developer base in 2025.

Chaincode Labs employed six to eight Core developers at an estimated $1.5 to $2.5 million in Core-specific funding. Spiral, the Bitcoin development arm of Block under Steve Lee, operated on an approximately $10 million annual budget across Bitcoin infrastructure with Matt Corallo as first hire since 2019; the Core-specific portion is estimated at $2.5 to $3 million. Brink paid the salaries of eight full-time Bitcoin Core engineers including Michael Ford (fanquake), who merged 56 percent of all 2025 PRs, at $1.5 to $2 million in Core funding. Localhost Research, founded by Mark Erhardt (Murch), Ava Chow, and Jay Berg, employs five to six Core developers with $2 to $2.5 million in Core funding. MIT's Digital Currency Initiative funds Cory Fields, Marco Falke, AJ Towns, Wladimir van der Laan, and Sebastian Kung at $1 to $2 million. OpenSats and the Human Rights Foundation contributed an additional $1.5 to $2.5 million combined for Core-specific work.

Aggregate Core funding in 2025 was approximately $10 to $14 million, with additional funding from Blockstream and other corporate funders beyond this estimate.

On the panel Jonas moderated, Gloria Zhao described working on Core in her own voice: "it's not written in anybody's job description as the volunteer fire department that, like, if a fire happens, like, it's your job to put it out. But, like, you know that if you don't do it, who will?"

The volunteer framing was not a single speaker's flourish. It was the cohort's shared vocabulary for institutional work that was, in fact, funded at a level comparable to a small American city's professional fire department.

Jonas's framing characterised critics as people who would rather invoke a shadowy cabal than argue their case coherently: "I would just add that blaming a shadowy cabal gatekeeping Bitcoin is not going to relieve the pain of engaging in this messy and often frustrating work of persuading your most qualified opponents of your thesis."

Antoine Poinsot, sitting next to Jonas on the panel, would deploy the same characterisation six months later on his personal blog when he described the merge's critics as a "parasitical group" and their concerns as "conspiracy theory."

The standard Jonas offered from the MIT stage was the messy work of persuading your most qualified opponents. The six weeks that followed are the public record of what the cohort actually did with its most qualified opponents.

The water boy was in fact the CEO. The "we" he kept slipping back into was a leader speaking for his project. The first water boy framing put distance between Jonas and the system whose outcomes were about to become contested. The "we" betrayed the narrative of distance.

Part V: The Commission

Poinsot described, on the public record, how the proposal to remove the OP_RETURN limit originated. On the What Bitcoin Did podcast with Danny Knowles in October 2025, he recounted running into someone at the MIT Bitcoin Expo and then at the UpNext conference in DC. "This guy," in his words, who explained a BitVM design that needed penalty transactions carrying zero-knowledge proof data: "he needs to have some sort of penalty transactions as you have in Lightning that needs to be confirmed fast. And this transaction needs to have some data attached to them because it's some sort of zero-knowledge proof." The person told him they were using fake outputs to get around the 80-byte OP_RETURN limit. Poinsot's reaction: "that's terrible." He came back and sent a post to the bitcoin-dev mailing list on April 17, 2025.

The post was brief: a single argument, three short paragraphs, a conclusion, and a footnote. The core argument: the existing restrictions encouraged harmful workarounds while failing to deter unwanted usage. The conclusion: drop them. The footnote named Citrea directly, citing section 6.1 of the Clementine whitepaper as the example of a project forced to use unspendable Taproot outputs because of the OP_RETURN limit. The corporate entity whose data requirements would trigger the seven-week campaign was in the founding document, named by the campaign's originator, on the day he sent it.

The harm-reduction justification that Gloria Zhao would deploy at the moment of the merge was already present in that founding post. The unspendable Taproot workaround was characterised as the consequence of the limit's existence, and the limit removal was framed as the remedy. Six weeks before the merge, the justification was written and attributed to the same corporate use case.

Sjors Provoost replied the following day, confirming from the Citrea whitepaper that a single OP_RETURN output of 144 bytes would suffice. Greg Sanders replied the same day, noting he had discussed the Citrea use case "in person." Luke Dashjr posted on April 26: "It should be needless to say, but this idea is utter insanity."

Pieter Wuille gave a Concept ACK on relaxing the OP_RETURN limits, including removing them entirely, stating he had been a strong proponent of the limits in the past but recognised the reality of demand. Greg Maxwell arrived on May 2, the day the institutional uncap PR was filed, with a long technical argument concluding that the existing rule was "a dead parrot policy."

Peter Todd filed the commissioned uncap PR (PR #32359) on April 27, 2025, "Remove arbitrary limits on OP_Return (datacarrier) outputs." It contained no disclosure of any kind. Todd had filed an earlier OP_RETURN removal PR in July 2023 (PR #28130) that did include one: he had been paid $1,000 by Christopher Allen for the work. The transparency standard Todd had applied to the lower-stakes PR, he abandoned for the higher-stakes one. Eight days after filing, on May 4, 2025, he confirmed on Stacker News that the PR "wasn't my idea" and that he had been "asked to open it by an active Core dev because entities like Citrea are using unprunable outputs instead of OP_Return, due to the size limits."

On April 28, Jason Hughes (wizkid057), a long-time Bitcoin developer and current VP of Development and Engineering at Ocean Mining, posted on Todd’s commissioned uncap PR: "If this PR is merged, Bitcoin as we know it changes forever in the most fundamental way imaginable: The reference implementation explicitly turning the Bitcoin network into an arbitrary data storage system, instead of evolving it as a decentralized currency." He characterised the change as needing to be "treated with no less scrutiny than a hard fork, because that's essentially what this is: A fork away from being a decentralized currency towards being decentralized nothingness." He concluded: "This PR might as well be titled, 'Quickest way to kill the Bitcoin project.'"

His comment then turned to resignation, when he named the outcome as already settled: "Anyway, since this is going to get merged anyway despite wide community objections”. The PR he commented on was closed on May 12 and replaced by the institutional uncap PR, which carried the same change to merge on June 9, 2025. His prediction held. The cohort's institutional power was sufficiently visible by April 28 that an independent developer on the PR thread itself was treating the merge as a foregone conclusion.

Two days after Todd filed commissioned uncap PR, the moderation infrastructure was deployed against the specific category of comment that would have surfaced the merge's undisclosed conflicts of interest. On April 29, Marcel Hernandez posted a comment on the commissioned uncap PR with a direct disclosure request: "@jlopp I think it would be appropriate for you to disclose that you're one of the VCs backing Citrea, a company that is currently abusing unspendable Taproot outputs to embed arbitrary data onchain and whose behavior motivated the ML discussion and this PR."

The comment was marked as ABUSE and hidden for anyone not logged in to GitHub. A subsequent comment from BitcoinMechanic defending the disclosure request as "entirely appropriate to point out" was marked as OFF-TOPIC and hidden too. The infrastructure designed to elevate signal over noise was, on the public record, suppressing requests for the disclosure of conflicts of interest in the PR where those conflicts were live.

Citrea had publicly disclosed Jameson Lopp as a participating investor in its $14 million Series A funding round, announced October 31, 2024, alongside Peter Thiel's Founders Fund, Maven11, Erik Voorhees, Balaji Srinivasan, and others.

Lopp publicly advocated for the limit removal and against BIP-110, the soft fork proposal designed to curb inscription spam and arbitrary non-monetary data on Bitcoin. A participant in the mailing list thread raised the conflict directly on May 2: "It feels weird — like some sort of attack or a bribe. One protocol is mentioned that would benefit from this change, and then an investor ACKs the PR. Someone points out a potential conflict of interest, and their comment gets hidden. That’s shady, no denying it."

Sjors Provoost replied: "The Citrea protocol does not benefit from this change at all." Two days later Lopp posted on X, with the Provoost comment screenshotted: "A false claim has been spread that my investment in Citrea created a conflict of interest regarding the max OP_RETURN size policy. I've been explaining why that's not the case to any who made the claim, but if you don't believe me then feel free to listen to Sjors."

He was citing as authority a denial issued into a public record where Citrea had already been named by Poinsot in the founding mailing list post, by Corallo three days earlier on X ("the entire reason for this debate is removing a filter that is going to result in a project (Citrea) shipping a product[….] "), and by Sanders, Poinsot, and Wuille on IRC the day before.

Lopp also defended the moderation infrastructure that suppressed the disclosure requests. On April 30, replying to Bob Burnett, a Bitcoin operations executive who had observed that "plenty of people, including me, feel reasonable objections/concerns have been raised." Lopp wrote: "Since you're not a Bitcoin Core maintainer, you don't get to make that judgement."

The doctrine Jonas had codified in October 2024, that those with "skin in the game" should have "the loudest voices", was being deployed on X by a named investor in Citrea, against figures with substantial Bitcoin standing whose only deficiency was not being on the maintainer list or part of the cohort.

The same day, the suppression mechanism extended to a contributor proposing a bounded alternative. On April 30, jesterhodl posted on Poinsot's smaller alternative PR #32381, which sought to allow more than one OP_RETURN output per transaction, instead of uncapping the OP_RETURN size. Jesterhodl had read the Clementine whitepaper and understood Citrea's exact technical setup: "1 OP_RETURN of 80 bytes and 2 taproot outputs of 32 bytes, totaling 3 data chunks of 144 bytes in sum." The numbers matched Poinsot's own published description.

Jesterhodl proposed a configurable opreturncount setting with a default of 2, which would address Citrea's need without removing the safety measure. Poinsot's response: "@jesterhodl conceptual discussions should go on the mailing list. If you do take it there please understand the topic before sharing your strong opinion, this has nothing to do with Taproot witnesses. I just blocked you, which should prevent you from commenting further on this PR and wasting everyone's time."

The contributor whose rationale context matched the campaign originator's own numbers was blocked from the relevant PR by the campaign's originator.

The case the cohort would later make for the merge was that the change was a defensive harm-reduction response to existing inscription abuse. The directional record on the PRs themselves does not support that framing.

Between April 17 and May 2, the cohort had three options on the table: a phased increase (Poinsot's mailing list proposal), a bounded change to allow multiple OP_RETURN outputs (Poinsot's own PR #32381), and full removal (Todd's commissioned uncap PR (PR #32359), replaced by Sanders' institutional uncap PR (PR #32406).

PR #32381, the bounded version, received Concept NACKs from five named technical reviewers and an Approach NACK from AJ Towns. Poinsot closed his own bounded PR and commented that it was not “consensual among regular contributors,” and that “I don't think there is any use in keeping this PR open. We should address the concerns of those contributors and do #32359 instead." Poinsot then closed the PR, but the multiple OP_RETURN outputs it proposed was not abandoned, despite Poinsot’s statement it did not have consensus.

The institutional uncap PR ended up absorbing both options: it uncapped the per-output size and also allowed multiple OP_RETURN outputs per transaction. The choice the cohort presented as bounded versus unbounded was, in the merged result, both. It was also seemingly implied that the commissioned uncap PR was less controversial. The 4:1 Github comment vote ratio against it was not mentioned.

The reason was given by Sanders, in his own words at the May 1 IRC meeting: "I'm fine enough with 2, but prefer 3 out of humility of not knowing the next zk proof size people want to use, and do not want to revisit this again."

He was arguing forward-looking accommodation for unspecified future ZK-rollup users with unknown proof sizes. (Citrea is a ZK-rollup; Alpen Labs is a ZK-rollup; BitVM is the broader cryptographic primitive that ZK-rollups on Bitcoin require). The harm-reduction framing predicted the bounded option. The preemptive accommodation framing predicted the unbounded one. The cohort selected both, packaged as one.

At the same meeting, in a thirteen-minute window, Citrea was named in turn by three separate participants. Sanders, at 16:44: "You don't have to love Citrea's design, but actively reducing harm should be the default where possible." Poinsot, at 16:54: "Also, to be clear this type of transaction in Citrea is very unlikely to end up onchain. I raised the point because it illustrates an issue with our policy, not because there is urgency." Pieter Wuille, at 16:57: "there is very good probability that citrea, or whatever individual project, just doesn't take off." The author of PR #32406, the campaign's originator, and the senior technical authority each named Citrea as the live motivating case at the meeting that set the merge direction.

Outside the cohort, what was happening raised questions. On May 5, 2025, Keysa (@SimplestBTCBook) asked Todd on X which Core dev had asked him to open the commissioned uncap PR. The post included a screenshot of the Stacker News comment where Todd wrote that the PR "wasn't my idea" and that he had been "asked to open it by an active Core dev because entities like Citrea are using unprunable outputs instead of OP_Return."

Hours later, Lysander (@UnderCoercion) named Antoine Poinsot of Chaincode Labs publicly. Poinsot replied, confirming he had "checked with Peter" before filing his own mailing list post, and that he expected his proposal to be "fairly consensual...among protocol developers."

Later that day, Samson Mow posted on X: "One dev paid another dev to submit a contentious PR (PR laundering? Is this a thing now?)."

The same post catalogued three additional patterns from the preceding two weeks: Bitcoin Mechanic’s ban from GitHub for "abuse," the locking-unlocking-relocking sequence the admins had executed on the PR, and a deleted ACK. Mow connected the moderation patterns to "the odd behavior around changing the definition of datacarriersize." He closed: "Many things have been said about many things and there have been many multi hour podcasts, but these strange behaviours seem to just be ignored and memory holed."

Mow is not a Bitcoin Core developer, not a Chaincode affiliate, and not an external critic of Bitcoin. He is the CEO of Jan3, a pro-Bitcoin infrastructure and advisory firm that helps governments and users integrate Bitcoin more deeply into their financial systems. His characterisation is the named external industry assessment of what the Poinsot-Todd arrangement looked like from the outside.

Poinsot's public response was addressed to Mow directly: he wondered whether Todd had mentioned the commission "so the most desperate attention grabbers would reveal themselves," and asked Mow whether his followers were "that stupid." Poinsot denied paying Todd. Mow clarified that the payment he had been referring to was in 2023, not 2025: a Todd PR from that year proposing to uncap OP_RETURN, that had carried a public payment disclosure on its thread.

Citrea was the most frequently cited motivating example used by the cohort. Antoine Poinsot named it explicitly in his April 17 mailing list post that launched the campaign. Greg Sanders, Poinsot, and Pieter Wuille each referenced Citrea at the May 1 IRC meeting that set the merge direction. Peter Todd stated on Stacker News that he filed the commissioned PR because "entities like Citrea are using unprunable outputs instead of OP_Return."

In his March 2026 DM to this investigation, Poinsot disputed this account, stating he never discussed Citrea with Todd, that "Citrea did not need the change," and that "the person that asked Peter did not even talk about Citrea, because that person is me." This creates a direct tension with Todd's Stacker News statement and with how often Citrea was invoked in the public record.

Poinsot's own blog post at antoinep.com, however, describes Citrea's specific technical situation in detail: the Clementine bridge watchtower challenge transaction needing 144 bytes, and Citrea's workaround of using two permanent unspendable Taproot outputs in the UTXO set.

Regardless, the harm-reduction justification offered at the time of the merge did not hold up. Citrea's 144-byte payload remained cheaper via the existing Taproot witness workaround, so the company continued using it. The promised switch to a cleaner, prunable OP_RETURN channel never occurred for the very case used to justify the policy change.

What the merge did accomplish was send a clear signal: Bitcoin Core is willing to accommodate large-scale non-monetary data “use cases.” That signal specifically benefits the broader category of inscription ecosystems and ZK-rollups, regardless of Citrea's specific trajectory.

Part VI: The Merge

On March 31, 2025, Gloria Zhao joined Chaincode Labs. The maintainer who would merge the institutional uncap PR ten weeks later was now employed by the institution at the centre of this investigation.

Her position on the relay-policy question was already on the record. In October 2024, on the Stephan Livera Podcast, she had described the original mempool design as imperfect code that "doesn't really work for the use cases that we have today."

In an April 2025 Bitcoin Magazine interview, she stated it directly: "I don't think it's like protocol development's place to be like, this kind of transaction is bad based on use case, because that's kind of like against what Bitcoin is all about."

Two on-record statements, eight months apart, on major platforms. The position was not improvised under pressure. It was established before the action.

‍ ‍

On May 1, 2025, the weekly Bitcoin Core developer IRC meeting was chaired by achow101. The agenda item was proposed by Sanders: "let's decide a direction on OP_RETURN policy."

He presented four options. Option 0: close the PRs indefinitely. Option 1: double down on filtering, described as having been "rejected as a group repeatedly with 0 volunteers." Option 2: minimal adjustment. Option 3: remove limits entirely.

The framing was not neutral. Option 1, the position of the majority of public objectors, was presented as something nobody would volunteer for. Option 3 was the live choice. The community's position had been pre-invalidated before the PR was filed.

Gloria Zhao dismissed two of Sanders's four options before substantive discussion began. "I don't think (1) makes any sense. (0) is giving in to drama." She framed opposition as: "the garbage brigading can be ignored." And: "In general, I don't think popular opinion should stop us from doing what is right. And I don't even think this is a popular opinion tbh, more of a loud one."

The 4:1 community opposition ratio on the commissioned uncap PR (PR #32359) was already on the public record. Poinsot still characterised the cohort's position as the silent majority's.

The middle path was proposed in the same window and immediately shut down by the cohort in three minutes. Vasil Dimov offered the compromise: "I guess the most liberal is to have this configurable in Bitcoin Core so users can set whatever mempool policy they wish. This way Bitcoin Core developers will not be perceived as imposing their views on the node operators."

Zhao replied that it would be "a footgun to give options for users to reject transactions that will likely be mined."

Antoine Poinsot continued, framing opposition as bullying: "Storm in a tea pot. I don't think we should give in to bullies and do what we believe is good for actual Bitcoin Core and Bitcoin network users, ie the silent majority. I think we should merge Todd's PR and call it a day."

Ava Chow followed, building on Zhao’s footgun claim: "configurability here is basically just a placebo."

Sjors Provoost added that it would be "arguably dishonest to ship an option knowing it doesn't work, i.e. a placebo."

Wuille closed the exchange: "you should only provide configuration knobs when you can give advice on when someone should use it."

Four senior figures, in three minutes, foreclosed the mechanism the protocol was designed around. Node operator choice over relay policy was framed as footgun, placebo, and dishonest.

Murch offered the communications infrastructure to deliver the established narrative: "The upcoming Optech newsletter will cover this debate, anyone want to come on the podcast on Tuesday to talk to me about it?" Bitcoin Optech, co-founded by John Newbery, was offered as the platform for framing the change at the moment the direction was decided.

Five days later, Executive Director of both Brink and Bitcoin Optech, Mike Schmidt, hosted the Optech Newsletter #352 Recap Podcast alongside Murch and Sjors, presenting the OP_RETURN uncap position as a sensible, long-overdue technical adjustment aligned with network reality and harm reduction. They acknowledged the “drama” and community backlash, but treated the opposition largely as emotional and misguided.

On May 2, the day after the IRC meeting, and four days before the podcast, Sanders filed the institutional uncap PR (PR #32406). The new PR was the maximal version of the change. It uncapped the per-output OP_RETURN size by setting the default to 100,000 bytes, marking the existing -datacarriersize args as deprecated, and additionally permitted multiple OP_RETURN outputs per transaction. The change Poinsot had originally proposed as a bounded alternative was absorbed into the unbounded outcome. The new PR carried no association with a commissioning request and was authored by a Core developer with longer institutional standing in the project.

Eleven seconds after the bot announced the PR in the #bitcoin-core-dev IRC channel, Sanders posted: "fanquake please lock for now ^." The institutional uncap PR was locked to “collaborators only” the same day it was filed, before any community engagement could begin.

AJ Towns added the "Needs release note" label, then joking on IRC: "haha, i can add a label despite the lock! now i just need to publish a way of inscribing comments via github label edits." Towns was not a passive observer; he was administering the PR's processing from the day of filing.

The same day, Zhao posted on X defending the locking, mutings and hiding of comments happening on Github: “So yes, it is appropriate and important to keep discussions on topic by hiding off-topic comments, penalizing users who refuse to follow basic guidelines for discourse, and asking people to take conversations elsewhere.”

She described the opposition as “brigading”, “1-2 usernames and an LLM”, and “technical slop”.

Ten days later, on May 12, Zhao would close Todd’s commissioned uncap PR, with Sanders’ institutional uncap PR carrying the change forward.

On May 5, Sanders produced the project's account of the decision they had reached. He authored a public gist titled to read as the Core rationale and described the outcome in past tense, as if the matter was already settled.

It described three options considered (keep the cap, raise the cap, delete the cap) and concluded: "Option 3 earned broad, though not perhaps unanimous, support." The gist was published thirty-five days before the merge, on the same day the PR was unlocked for community comment, and before the substantive Concept NACKs from named technical reviewers had been filed.

The gist did not mention Citrea. It did not mention the massive controversy and opposition in the commissioned uncap PR (PR #32359). By May 13, 2025, archived on the Wayback Machine, the commissioned uncap PR’s description had received 401 thumbs-down emoji reactions and 102 thumbs-up, approximately 4:1 opposition, captured within sixteen days of filing and one day after the PR was closed without merge.

The gist also did not mention the public commission of PR #32359. It did not mention Luke's rejected filter patch PR #28408. It did not mention CVE-2023-50428 or the closure of that CVE's Github Issue.

The change was instead framed as "align(ing) default policy with actual network practice." The history of how that network practice had been permitted to develop despite broad opposition was the history the gist did not contain.

After the institutional uncap PR was unlocked on May 5, Luke Dashjr posted: “Concept NACK. Already been over this on multiple PRs. Spamming it with immaterial minor differences won't change the fundamental insanity and malice of the concept.”

PR #32406 remained open to community comment for thirty-five days before the merge. Among named technical reviewers, DrahtBot’s automated summary recorded 21 Concept NACKs, 9 ACKs, and 14 Concept ACKs before the merge.

AJ Towns gave a full ACK, the same Towns who had ACKed the documentation change that made Luke’s patch rejectable, and who had given Luke’s patch an Approach NACK. Fanquake gave a full ACK. Peter Todd gave a full ACK on the PR that replaced his own.

On the PR thread, Sjors Provoost addressed the Concept NACKers directly: "There's no need to Concept N(ACK) this if all you're going to do is repeat comments from #32359. They're not votes." Provoost was treating the two PRs as one continuous proposal when it suited the cohort's framing and as separate when it did not. The critics whose substantive objections had already been registered on the commissioned uncap PR were told not to register them again on the institutional version, and that their original objections had not been votes anyway.

Sanders, the PR's author, stated: "FWIW I will not engage in meta discussion here. I will respond to any deficiencies in the code itself." The community's substantive objections were categorised as either procedural noise or off-topic. The author would only discuss the code.

‍ ‍

On May 22, 2025, at the weekly Bitcoin Core developer meeting, Pieter Wuille presented a gist, stating it was written “with help from darosior (Antoine Poinsot) and glozow (Gloria Zhao).”

Two weeks later, on June 6, 2025, three days before the merge, the document appeared on bitcoincore.org as a letter signed by thirty-one Bitcoin Core contributors. Its central argument deployed the word censorship twice, characterising the existing 83-byte OP_RETURN limit as an attack on Bitcoin's censorship resistance. It was not a post-hoc defence. It was published as the merge was being prepared, framing the change in advance.

Of the 31 signatories, the large majority have documented ties to the small set of institutions (Chaincode Labs, Brink, Bitcoin Optech, Spiral, MIT-DCI and Localhost Research) that form the core of the network documented in this series.

Jimmy Song, a Bitcoin developer, educator and author, was not asked. "it certainly looks like some sort of inner circle within Bitcoin Core. Like you said, I've contributed to Core, but I wasn't even asked whether or not I wanted to sign this letter, right? And there's a lot of people that aren't on that list that I know have contributed significantly in the past that aren't on it. So it's a, it's a, um, I think it, it raised a few eyebrows, to say the least," he said on The Bitcoin Matrix podcast in April 2026.

Luke Dashjr was also outside the drafting circle. He posted a substantive NACK on X on June 7, identifying the letter's internal contradiction in treating out-of-band relay as both negative and as "an important aspect of Bitcoin's censorship resistance," its mischaracterisation of spam as "largely harmless," and the falsity of its claim to be aligned with "Bitcoin's long-term health." The cohort did not answer it.

Gloria Zhao had signed the letter, and three days later, on June 9, she merged the institutional uncap PR (PR #32406), the PR the letter defended.

Zhao referenced Issue #29187 directly in her merge comment. She listed it alongside Luke’s filter patch as “pull requests” to ‘fix the filters.’” Issue #29187 is not a pull request though. It is the formal repository record of CVE-2023-50428, the security vulnerability designation issued by the National Vulnerability Database.

The reactions to those proposals, she wrote in the merge comment, showed that “using mempool policy to filter transactions is controversial amongst contributors.”

She did not apply the same characterisation to the proposal she was merging. The institutional uncap PR stood at 21 Concept NACKs against 14 Concept ACKs and 13 ACKs at the moment of merge, with the 4:1 community ratio on the commissioned uncap PR (PR #32359) already on the record. By her own standard, the merge was more contested than the proposals she had described as controversial. The merge comment did not say so.

The harm reduction argument was built on the inscription bypass. That bypass existed because Luke's patch had been rejected. The patch had been rejected using documentation quietly amended four months before it was filed.

Part VII: The Justification

The justification offered for the merge had three empirical components and a rhetorical one. Each can be tested against the public record.

The justification presupposed that the conditions it cited (the inscription bypass at scale, the routine miner acceptance of non-standard transactions, the demand the cohort characterised as "actual network practice") were facts the policy needed to align with. The cited conditions had documented architects.

Casey Rodarmor launched the Ordinals protocol on Bitcoin mainnet in January 2023. He then announced the Runes protocol on September 25, 2023 (with launch on April 20, 2024). Runes embeds its protocol messages in OP_RETURN outputs. In a September 26, 2023 Bitcoin Magazine article, Rodarmor was documented as having designed Runes to systematically produce transactions that violate Bitcoin Core’s 80-byte OP_RETURN standardness rules, relying on miner willingness to include non-standard transactions for fees as the workaround. The section heading was “Standardness Rules Are For B*tches.”

On April 1, 2024, Peter Todd posted publicly that OP_RETURN was “already uncapped” through his Libre Relay fork, and explicitly solicited miner participation in the bypass infrastructure: “more miners need to peer with Libre Relay nodes.” Rodarmor replied in the same thread, identifying the OP_RETURN size limit as the constraint Runes’ growth would test.

The architect of the largest OP_RETURN-using protocol and the operator of the relay infrastructure designed to bypass the limit were in direct public conversation about cultivating the demand profile the cohort would later cite as the existing reality justifying removal, fourteen months before Todd filed the commissioned uncap PR.

By February 2025, the Runes protocol’s own development repository was documenting the strategic posture explicitly. On February 11, 2025, in ord GitHub issue #4219, contributor joshdoman wrote: “A more near-term compelling argument, though, is to ensure users can get transactions relayed and included in blocks, even if the runestone exceeds 80 bytes. If core devs saw that users were doing this, there would be a stronger case for changing relay policies and lifting the OP_RETURN limit.”

This is the case-construction strategy stated in the Runes protocol’s own repository, two months before the cohort’s campaign to uncap OP_RETURN began. Rodarmor confirmed it the next day, February 12, 2025: “Regarding the 80 byte OP_RETURN standardness limit, I actually do not think that this is something we should work around. Many miners already accept nonstandard transactions, and Peter Todd’s Libre Relay fork of Bitcoin Core can be used to submit oversize OP_RETURNs without needing to send them to a specific miner. It is in the rational self-interest of miners to accept transactions with oversize OP_RETURNS, and I think eventually, Bitcoin Core will lift the standardness limit, so we just shouldn’t worry about it.”

The Runes protocol architect predicted, four months in advance, the policy outcome the cohort would deliver. The same GitHub thread documents Todd offering paid annex support to the Runes development team in March 2025: “this is something I’d be happy to add to Libre Relay if someone wants to pay for my time... Rough guess this is ~10 hours of work all in. So let’s say $1500.” Rodarmor accepted and the conversation moved to private email five weeks before Todd filed the commissioned uncap PR.

‍ ‍

The uncap argument rested on a second empirical claim: that the existing 80-byte limit was an ineffective filter. That claim too can be tested against the public record. Marcel Hernandez counted non-standard OP_RETURNs across the period the cohort was making its case. Between January and April 2025, in approximately 7 million total transactions, 30 non-standard OP_RETURNs reached the chain. A rate of 0.0043 percent.

Hernandez posted again in June, showing that in May 2025 alone, in the period after PR #32406 was filed and the campaign was public, but still before the merge, 13,325 non-standard OP_RETURNs reached the chain. That is 443x more transactions in one month, than the total of the four months preceding it.

The deterrence effect of the standard relay rule had been real. Suppressed demand emerged within weeks of the announcement that the rule would be removed. The claim that the limit was ineffective in the long term was advanced even as short-term data showed it had been suppressing large non-standard OP_RETURN usage.

The uncap argument’s third empirical claim: removing the OP_RETURN limit would redirect inscription data away from the more-harmful witness route and toward the less-harmful, prunable OP_RETURN route. That prediction can be tested against the post-merge network record.

The blockspace data analyst Renaud Cuny published an on-chain analysis in December 2025, six months after the merge: “Changing OP_RETURN default policy does not seem to help so far in reducing inscriptions volume. A new channel opened up for data (by default policy) while the old one continues.”

Cuny's data documented that large OP_RETURN activity (greater than 1 KB) was near zero before the merge and activated immediately after June 2025, while inscription witness data continued at scale. As of the analysis date, 36 percent of blockspace was being consumed by non-financial transactions, 76 GB of non-financial data had been added to every node over the preceding three and a half years, and the predicted redirection had not occurred.

The change opened a new channel; it did not close the old one. The harm reduction promise was not delivered by the post-merge network. The justification was empirically unmet.

‍ ‍

The rhetorical use of the label “censorship” tracked the merge sequence asymmetrically in Gloria Zhao's own statements. On May 1 she had characterised preserving the existing policy as “giving in to drama” and the public opposition as “garbage brigading” that “can be ignored.” On May 2, she argued on X that the muting of named technical critics on the PR was not censorship and described their participation as “technical slop.” On June 6, she signed the 31-developer letter published on bitcoincore.org, which argued that the existing 83-byte OP_RETURN limit was an attack on Bitcoin's censorship resistance.

The word “censorship” was available to describe a mempool filter that sought to exclude spam. It was not available to describe the muting of critics on the PR where that filter was being removed. Three days later, Zhao merged the institutional uncap PR (PR #32406).

The misuse of the “censorship label” is sharper because Zhao herself had given the correct definition six weeks earlier. At the MIT Bitcoin Expo in April 2025, in a Bitcoin Magazine interview with Shinobi, she said censorship resistance was “an emergent property of individual nodes participating in transaction relay.”

By that definition, neither Luke's patch nor the existing 83-byte limit touched censorship resistance. They were relay-policy choices by individual nodes, precisely the mechanism she had identified as the source of the property. The 31-developer relay statement letter she signed three days before merging the change, deployed the word in the opposite direction.

Zhao described the Ordinals controversy: "I was like really regretting being the mempool maintainer. People would yell at me where they're like, You need to add policies to ban inscriptions. And then there would be people who are saying you need to relax the policies that create these incentives for people to submit nonstandard transactions to miners."

In the sign-off, before any of the institutional escalations against critics documented in the rest of this article had occurred, Shinobi said: "I'd like to apologize on behalf of all of the retards in this space for stupid shit you have to put up with." Zhao replied: "Thank you. I appreciate it. Thank you for saying that."

Bitcoin Magazine's coverage of the maintainer who would merge the institutional uncap PR six weeks later, closed with the interviewer offering an apology, on behalf of the community, for the criticism Bitcoin Core was receiving. The framing the cohort would use in the weeks and months that followed, critics as garbage, trolls, parasites, brigaders, conspiracy theorists, cults, was already shared by the journalist conducting the most prominent media interview with the maintainer in the run-up to the merge.

Jameson Lopp had also participated in the same Bitcoin Magazine interview series, with the same interviewer, in the same week. Asked about the long-term risk of corporate institutions to Bitcoin governance, he framed the question and answered it: "And so then you have to ask yourself, well, if the vast majority of economic power is actually shifting into the hands of a few like large corporations and institutions, how is that going to change the governance and the future evolution of the protocol? It very well may be that those institutions don't want to see any changes because they don't really care about any of the cypherpunk aspects and the individual sovereignty of Bitcoin holders and trying to scale that sovereignty out to the entire world."

A few days later, Lopp publicly acked and advocated for Todd’s commissioned uncap PR (PR #32359), without disclosing his investment in Citrea, the corporate entity Todd had named in his Stacker News account of the commissioning behind the same PR.

The cypherpunk-aspects framing Lopp had given the Bitcoin Magazine audience a few days earlier, was now turned on its head. The change he was helping defend had been comissioned “because entities like Citrea are using unprunable outputs instead of OP_Return”, exactly the kind of corporate institution he had described as not caring about cypherpunk aspects at all.

Lopp posted publicly on X September 3, 2025, quoting a Citrea announcement: "There's a potential future in which Citrea solves the spam problem by pricing out stupid low value use of block space. A very entertaining future."

The type of corporate institution he had warned the Bitcoin Magazine audience against in April was, by September, the institution he was publicly cheering for to "price out" what he called “stupid low value use of block space.” The cypherpunk-aspects framing of April had become an entertainment value by September. He retained his Citrea investment throughout.

Part VIII: The Response

The cohort's communications operation had begun before the merge.

The direction was set at the May 1 IRC meeting. Murch had offered Optech as the communications vehicle before the meeting ended. The following week, Brink's Executive Director Mike Schmidt hosted the Optech podcast and presented the change as a neutral technical correction eliminating a mal-incentive caused by the 80-byte limit. He did not mention Citrea. He did not mention the massive community opposition ratio on the commissioned uncap PR or the large amount of Concept NACKs from named technical reviewers on the institutional uncap PR.

The framing contrast with what was said outside the network is precise. Jimmy Song described the same event on the Bitcoin Matrix podcast in April 2026:

"This change going in I think broke a lot of different precedents. Usually it required general consensus and not too much feedback. There was significant pushback... they went through with it, and this was a significant break from precedent because generally Bitcoin devs have not merged in something that had significant pushback from the community."

-- Jimmy Song, The Bitcoin Matrix podcast, April 2026

The same event. Two accounts.

The rest of the cohort response followed the template documented across Article 1 – The Network, and Article 2 – The Lever. Matt Corallo, who had established the beginnings of the rhetorical vocabulary in 2018, framed the community opposition as "stupid drama." Mark Erhardt (“Murch”) acknowledged on the Optech Newsletter #373 Recap Podcast that going straight to 100,000 bytes rather than incrementally was a strategic error: "In hindsight, if I could go back and do it again, probably setting at a lower limit first, and then if that gets circumvented, increasing it again a couple of times would have been a lot smoother networkwide." Then: "We might as well stick to it."

The process was wrong. The conclusion would be defended anyway.

‍ ‍

Poinsot, the originator of the campaign, took the response template into longer form. On his personal blog at antoinep.com, in a piece titled “Relay policy drama,” he characterised the critics as a “parasitical group” and described their concerns collectively as “conspiracy theory.” He closed the piece by recommending Adam Jonas’s MIT Bitcoin Expo presentation as evidence that Bitcoin Core is not compromised, without disclosing that Jonas was his employer at Chaincode Labs.

On X on August 6, 2025, two months after the merge, Poinsot extended the framing: “It is irresponsible for influencers with large followings to encourage people to take actions that deteriorate the network.” The actions he characterised as deteriorating the network were the actions of node operators switching from Bitcoin Core to Bitcoin Knots, the exact behaviour the protocol’s software-implementation pluralism is designed to permit.

On September 25, 2025, Adam Back replied to Chris Guida in a thread about the release notes and the 2023 documentation change. He stated on X: "If you read the original OP_RETURN release notes from 2014, it's quite clear that datacarriersize refers to OP_RETURN payload. That's the same today. If you or others want to expand the datacarriersize to cover other formats, it's ok to be honest and say that. It's however a rather bad idea to go down the heuristic direction. Arms race, and leads demands to actually censorship data after you've built the tools to do it, and demonstrated your willingness and interest."

Back’s claims are inaccurate on multiple levels. First, the -datacarriersize setting did not exist in 2014. It was only introduced in Bitcoin Core 0.10.0 in February 2015. The 2014 release notes (0.9.0) discussed OP_RETURN but made no mention of -datacarriersize at all.

Second, even if Adam had correctly referred to the 2015 notes, his statement would still be misleading. The 2015 release notes described the option in relatively broad terms as applying to “data carrier outputs”. They did not specify the narrower wording: “raw scriptPubKey”. That was introduced in the 2023 documentation change that made the description significantly narrower right as the debate over inscription data in witness scripts intensified. By pointing to the old release notes, Adam presented the 2023 wording as a simple clarification of long-standing original intent, when in reality the documentation had been tightened in a way that directly supported the rejection of attempts to close the data loophole.

Third, Back's framing of the proposal as an "expansion" of the setting reverses the documented history in another direction. The new fields where inscriptions had moved (witness data introduced by SegWit in 2017, script-path data introduced by Taproot in 2021) did not exist when -datacarriersize was designed. Luke's filter patch had proposed to restore the setting's original scope to include those new fields. Restoration was the documented action; "expansion" was Back's framing of it. The warning Adam followed with, that the path leads to "actually censorship data," echoed a rhetorical reversal the cohort had already deployed against Luke's filter patch in 2023, and in defense of the institutional uncap PR merge (by the 31-developer letter). The framing recasts filtering spam as suppressing speech, treating the network's relay defaults as a free-speech surface rather than as configurable filter policy. Back was deploying it again in September 2025, four months after the merge had succeeded.

‍ ‍

Chris Guida is a Bitcoin and Lightning Network developer, educator, and advocate who is sharply critical of Bitcoin Core's recent direction, particularly v30 policies enabling greater arbitrary data storage. He replied to Adam, identifying what Adam’s framing omitted: "Yes of course it refers to the opreturn payload, because that was the only 'officially sanctioned' method of storing arb data at the time, and it was limited to 40/80 bytes. Once inscriptions started to be exploited, a decision needed to be made about whether to 'officially sanction' such storage, or to move to disable it. Knots 'officially sanctions' inscriptions but limits them to the user's datacarriersize limit, which makes sense given the precedent. Core 'officially sanctions' them and moves the opreturn limit to the same as the inscription limit, which breaks with the precedent." Adam’s framing was contested in real time by a named developer making the same structural argument this article documents.

Almost a year after the merge, on February 17, 2026, Adam Back posted on X that the limit's specific value was technically arbitrary: "they could see it really doesn't matter if the limit is 0, 250, 520 or unlimited. i advocated for 160b and later 256b... but really i'd be ok tossing a dice." The defender's position was that the limit's value did not matter. The cohort's contemporaneous selection between bounded and unbounded options shows it mattered enough that the unbounded option was preferred over the bounded one at every fork.

‍ ‍

The community response began on the day of the merge. The reading from outside the cohort was articulated within hours. Jason Hughes, Vice President of Ocean Mining, posted on June 9: “52 days. That’s all it took for a non-Bitcoin project to co-opt Bitcoin Core’s development and direction.” The number Hughes had counted was the number of days from Poinsot’s April 17 mailing list post naming Citrea to Zhao’s June 9 merge of the institutional uncap (PR #32406).

Hughes added: “If such an insignificant shitcoin entity can shift Bitcoin Core’s development so easily with so little influence, despite massive community push back... imagine what will happen when more powerful folks learn how easy it is to derail Bitcoin Core’s development?” The contemporaneous reading of the change, on the day of the merge, by a named operational executive at a competing mining pool, was that a single corporate use case had captured Bitcoin Core’s policy direction in 52 days against community opposition. The denials issued in the months that followed were issued into a discourse where this reading was already on the public record.

The broader community responded through its nodes. Bitcoin Knots quickly surged from approximately two percent of the visible node network to around twenty percent.

On September 22, 2025, Mike Schmidt, Executive Director of Brink, filed PR #33453, reversing the deprecation of the -datacarriersize option, the specific setting Poinsot had argued should not exist. Schmidt added a disclosure to the PR description acknowledging that some reviewers were Brink-funded developers; he wrote that he had emailed them privately to confirm their review feedback would not affect their standing, funding, or employment. Ava Chow merged it on October 1, 2025.

Schmidt's PR removed the planned deprecation of node operators’ ability to limit data carrier sizes manually. It did not change the default policy. By default, Bitcoin Core v30 still relayed transactions with unlimited OP_RETURN data, exactly as the June 9 merge of the institutional uncap PR (PR #32406) had established. As Luke Dashjr pointed out the day after the original uncap was merged, the multiple-output allowance the institutional uncap PR had introduced still meant a node operator who set the value to 83, would allow relay of roughly 830 bytes per transaction, nearly nine times what Bitcoin Core v29 had permitted by default. Schmidt's PR #33453 addressed the deprecation optics. It left the new uncapped default untouched.

In September 2025, Nick Szabo, the cryptographer who coined the term "smart contract" and whose intellectual contribution to the foundations of Bitcoin is among the most cited in the field, had been silent on social media since early 2021. He returned just before the release of Bitcoin Core v30, with the clear opinion that the OP_RETURN uncap “publicly invites more non-financial data on Bitcoin.” On October 8, he posted a terse recommendation: "As a (hopefully) temporary measure, run Knots. I strongly recommend not upgrading to Core v30."

On October 10, 2025, Bitcoin Core v30 shipped. It was the first release to include the OP_RETURN uncap.

Part IX: The Aftermath

On October 30, 2025, AJ Towns quoted a Citrea hype post by Hampus (@hampus_s) about the upcoming Citrea mainnet launch. Towns asked “Have we had citrea txs on signet, and, presuming not, what additional standardness changes are needed for them?"

It established two facts simultaneously. Citrea, the corporate use case the cohort had named extensively before the merge as the live motivating example, had not by late October demonstrated its transactions on Bitcoin's Signet test network. And the OP_RETURN uncap had not been sufficient for their needs. Further standardness changes would be required to fully accommodate the architecture the merge had been justified by.

The harm reduction framing for uncapping OP_RETURN had been it would give Citrea, Alpen Labs and the general inscription ecosystem a less harmful channel for their data needs, redirecting them away from the unprunable Taproot outputs they were currently using. Four months after the institutional uncap merge, Towns posted publicly that Citrea's transactions had not appeared on signet and that further standardness changes would still be needed.

Citrea kept using the old method. The harm continued. The justification did not hold. A new door was now open regardless.

‍ ‍

On October 28, 2025, a formal institutional act against Luke Dashjr had been opened in the Bitcoin Core repository.

PR #33723 was opened by a low-volume contributor with the GitHub handle SatsAndSports, proposing to remove Luke's DNS seed on the grounds that it was returning no nodes running Core versions later than 28.1.0. Within hours, the cohort supplied the institutional case around it.

Jameson Lopp, the Citrea investor who had publicly advocated for the institutional uncap PR without disclosing his investment, posted a Concept ACK characterising Luke as having “a lengthy history of poor security practices.” Ava Chow, the maintainer who had closed CVE Issue #29187 and chaired the May 1 IRC meeting, added a separate claim: that Luke had “been in contact with the FBI regarding this hack, and has possibly given them access to machines under his control”, without supplying a source. Luke directly contested this claim in March 2026 direct messages to this investigation: “The FBI never requested nor had access to the server.”

Gloria Zhao opened Issue #33734 on October 29, formalising the case against Luke at the project level. The issue carried Ava Chow’s claim forward, asking whether Luke had “shared access to the server with investigators.”

Luke responded substantively on November 4, in the issue thread. He stated his seed was policy-compliant; that intentional version-lagging had never previously been raised as a breach; that his security practices had not been compromised in the sense the thread implied. His response received no substantive engagement from the cohort.

Fanquake merged PR #33723 on December 4, 2025, and backported it to all supported releases. Luke’s seed had been removed from the list of DNS seeds.

‍ ‍

Luke confirmed to this investigation in March 2026 that no private communication, formal or informal, preceded the seed removal. The repository's moderation system, which had been used to mute Luke and other named technical critics on the institutional uncap PR in May 2025, was not deployed against the many negative characterisations of him in the October 28 DNS seed removal PR thread or in the October 29 issue posted by Zhao.

Luke had operated DNS seed infrastructure for Bitcoin Core since the early 2010s. The removal eliminated his last remaining formal infrastructure role in the project. Six weeks earlier, on October 12, 2025, he had disclosed in a public exchange with Adam Back on X that he had also been removed earlier that year from the Bitcoin Core security disclosure list, where he had served for over a decade. He characterised that removal as someone having "abused their access to remove me this year."

The cohort's institutional response to its most qualified critic was documented across four years: BIP editor authority diluted, April 2021. Patch rejected using quietly amended documentation, October 2023 to January 2024. Security designation extinguished, October 2024. Removed from the Bitcoin Core security disclosure list, 2025. DNS seed removed, December 2025.

On February 6, 2026, Gloria Zhao resigned as Bitcoin Core maintainer, opening PR #34517 to drop her PGP key from the trusted-keys list. The resignation PR was merged the same day by Ava Chow. Zhao had joined Chaincode Labs ten weeks before the merge of the institutional uncap PR, and resigned eight months after it.

She posted a statement as an edit to the resignation PR description. Her central claim: "Over the last decade, Bitcoin Core has seen dozens of brigaded mempool policy PRs (covered to varying degrees by media). I've had the privilege of authoring, reviewing, moderating and/or merging the vast majority of them in the last 4 years and each one has strengthened the project's resistance to harassment."

She closed: "I have deep respect for the skill of capturing people's attentions/emotions. But the choice between integrity and engagement has always been an obvious one to me, and a privilege to be able to make as a maintainer of this project."

Engaging and responding to the broad concerns of the community around the OP_RETURN uncap was framed as brigading, emotional manipulation to be resisted. Not engaging was framed as integrity. The governance model Bitcoin was designed around, node operator behaviour as the structural check on developer power, appeared in her statement as the noise the project had been hardened against.

On June 6, 2026, Jameson Lopp opened Issue #46 in the bitcoin-core/meta repository, proposing that Bitcoin Core consider whether its moderation guidelines should apply to contributors who criticise the project on external channels. Matthew Zipkin, the Chaincode-funded moderator who had executed the April 2025 Riard ban and the October 2024 GregTonoski ban, replied that the institutional infrastructure already addressed the concern: "Fortunately such individuals typically get self censored anyway: no one reviews their PRs, maintainers ignore their reviews on other PRs, they don't get invited to meetings, etc." The pattern of ostracism documented in this investigation was, in Zipkin's words, a feature.

Lopp closed the issue with the framing that any contributor who publicly criticises Core while continuing to contribute is "hypocritical."

The Loop Closes

The sequence this article documents is retrievable from the public record.

It begins in June 2023, with the short documentation change PR thread, where key people who would matter in the events of the next two years were present. It runs through the rejected patch and the quiet documentation change and the extinguished CVE. It passes through the governance document that pre-authorised the dismissal of community opposition and the commissioned PR filed in response to corporate use cases. It arrives at a June 9, 2025 merge that used the loophole its participants had declined to close as the justification for not closing it.

Five years and five months separated the cold email Adam Jonas sent Gloria Zhao in January 2020 from the merge she executed in June 2025. The institution that sent the email was the institution that employed her when she did the merge. Article 1 of this investigation documented the trajectory between those two points.

What remained was a reference implementation that twenty percent of the node network had migrated away from, signalling the loss of trust in Core as the project's stewards, and a maintainer who, on her way out, characterised four years of community feedback as harassment the project had been hardened against.

A Note on What This Story Is and Is Not

This article does not argue that a secret cabal seized control of Bitcoin Core. Nor does it argue that every participant acted with the same motives.

Its narrower claim is that a small, highly interconnected network of contributors, maintainers, funders, employers and moderators repeatedly occupied the decisive positions in a sequence of events. At each stage, that network possessed disproportionate influence over which proposals advanced, which objections were amplified, which alternatives were dismissed, and which narratives became institutional memory.

This article is not a technical assessment of OP_RETURN relay policy. It is an account of how the change that was merged in June 2025 was produced and by whom, from the documentation change in June 2023 through the aftermath in 2026.

Citrea appears throughout the record, not as the sole cause of the policy change, but as the most frequently cited example used to justify it. The company was named in the originating mailing list post, referenced during IRC discussions, discussed by multiple reviewers, and repeatedly surfaced in debates over whether existing relay policy was encouraging more harmful workarounds.

Citrea did not adopt the new channel after the merge: the 144-byte payload its bridge requires sits just above the 143-byte threshold at which OP_RETURN becomes more expensive than witness data, and Citrea has so far continued using the Taproot witness technique outlined in its whitepaper. The harm-reduction framing the cohort had deployed at the merge required exactly this kind of switch. But the switch did not happen. The harm continued. Whether Citrea itself uses the new channel, or will use it in the future, is a separable question about Citrea's product trajectory and economics. The harm-reduction case at the merge was empirically built on a switch that was more expensive for the named beneficiary than the method it was supposed to replace.

Bitcoin’s world-changing mission is to separate money and state. The signal the OP_RETURN uncap sent was unmistakable: Bitcoin Core is open for business for non-monetary use cases, to the point that the project will preemptively facilitate them while refusing any attempt to curb spam.

Inscription ecosystems, ZK-rollups, and corporate non-monetary use cases take time and funding to develop. A Bitcoin Core that prioritises curbing spam does not invite the investment. A Bitcoin Core that has signalled the opposite does. The refusal to close the inscription loophole with Luke’s filter patch and the merge of the institutional uncap PR sent a clear signal that future non-monetary activity in the Bitcoin mempool and blockchain from entities like Citrea can proceed with confidence.

Defenders of the cohort and the OP_RETURN uncap would read these events differently. They would treat the documentation change as routine clarification, the rejection of Luke's patch as substantive review, the institutional concentration as the normal architecture of open-source stewardship, and the Citrea references as illustrative rather than causative. The choice is left to the reader who walks the record. The defender's reading requires each documented act to be read independently of every other. The article's documentation makes that read difficult.

The question this article raises is whether the accumulated informal power, the undisclosed commercial connections, and the institutional concentration, produced an outcome anyone outside the cohort had a meaningful voice in shaping.

The answer the network gave was to increasingly route around Bitcoin Core. That is Bitcoin working as designed.

Right of Reply

Formal right-of-reply correspondence was sent to ten named individuals whose conduct is documented in this article: Marco Falke, Greg Sanders, Ava Chow, Jameson Lopp, Adam Jonas, Matt Corallo, Mark Erhardt (Murch), Sjors Provoost, Matthew Zipkin, and Adam Back. Each was approached through documented channels; Lopp and Erhardt had previously blocked the author on X and were approached through email and contact form respectively. Marco Falke responded on the record by the deadline. The other nine did not.

Pieter Wuille was not approached. No working email address could be located, and his X account does not accept direct messages. His positions are drawn from his on-record statements on Bitcoin Core IRC, GitHub PRs, and the May 22, 2025 developer meeting where he presented the 31-developer relay statement gist.

Antoine Poinsot was contacted via X DM in March 2026 regarding PR #32359. He responded on the record.

Luke Dashjr's March and May 2026 direct messages to this investigation are preserved and on the record.

John Newbery, AJ Towns, and Michael Ford (fanquake) received formal right-of-reply correspondence in connection with the broader investigation across all three CAPTURE articles. None responded.

Mike Schmidt received formal right-of-reply correspondence in connection with the broader investigation and responded on the record

Gloria Zhao deleted her primary public social media account in May 2025 and resigned from Bitcoin Core via PR #34517 on February 6, 2026. No working contact channel could be established.

Responses received subsequent to publication will be incorporated into subsequent versions of this article.

Sources

All quotations from named individuals are drawn from the public record unless otherwise attributed.

GitHub pull requests and issues

Primary repository sources, all retrieved and preserved: github.com/bitcoin/bitcoin/pull/27832 (Falke documentation change, filed June 6, merged August 3, 2023). github.com/bitcoin/bitcoin/pull/28130 (Todd OP_RETURN removal, July 23, 2023). github.com/bitcoin/bitcoin/pull/28408 (Dashjr filter patch, September 2023, closed January 5, 2024). github.com/bitcoin/bitcoin/pull/29173 (Léo Haf documentation revert, January 4, 2024). github.com/bitcoin/bitcoin/issues/29187 (CVE-2023-50428 Issue, filed January 5, 2024, closed October 15, 2024). github.com/bitcoin/bitcoin/pull/29520 (Léo Haf -limitdummyscriptdatasize option, closed October 15, 2024). github.com/bitcoin/bitcoin/pull/28217 (Léo Haf DEFAULT_PERMIT_BAREMULTISIG, closed October 15, 2024). github.com/bitcoin/bitcoin/pull/29942 (vostrnad -datacarrier removal, closed October 15, 2024). github.com/bitcoin/bitcoin/pull/32359 (Todd commissioned uncap, April 27, 2025, closed May 12, 2025; includes Jason Hughes / wizkid057 substantive opposition comment, April 28, 2025). github.com/bitcoin/bitcoin/pull/32381 (Poinsot multiple-OP_RETURN bounded alternative, April 30, 2025). github.com/bitcoin/bitcoin/pull/32406 (Sanders institutional uncap, filed May 2, merged June 9, 2025 by Zhao). github.com/bitcoin/bitcoin/pull/33453 (Schmidt deprecation reversal, merged October 1, 2025 by Chow). github.com/bitcoin/bitcoin/pull/33723 (DNS seed removal, opened October 28, merged December 4, 2025 by fanquake). github.com/bitcoin/bitcoin/issues/33734 (Zhao issue formalising case against Dashjr, October 29, 2025). github.com/bitcoin/bitcoin/pull/34517 (Zhao resignation, February 6, 2026). github.com/bitcoin/bitcoin/pull/29415 (Dimov, merged January 12, 2026).

bitcoin-core/meta repository: github.com/bitcoin-core/meta/blob/main/MODERATION-GUIDELINES.md (moderation guidelines authored by Adam Jonas). github.com/bitcoin-core/meta/issues/17 (Zipkin enforcement of the April 23, 2025 Riard ban). github.com/bitcoin-core/meta/issues/46 ("Clarifying Moderation Policies for External Behavior" opened by Lopp June 6, 2026; Zipkin reply confirming ostracism as institutional practice; Atack opposition on principled engineering grounds).

bitcoincore.org repository: github.com/bitcoin-core/bitcoincore.org/pull/1139 (Wuille PR publishing the 31-developer relay statement, merged June 6, 2025 by Chow).

ord repository (Rodarmor's Runes/Ordinals project): GitHub issue #4219 (joshdoman case-construction strategy post, February 11, 2025; Rodarmor confirmation, February 12, 2025; Todd paid annex offer, March 2025).

Mailing list and IRC

bitcoin-dev mailing list: groups.google.com/g/bitcoindev/c/d6ZO7gXGYbQ (Poinsot founding post, April 17, 2025; subsequent thread through May 2025; Wuille and Maxwell emails preserved as screenshots; Sjors Provoost reply naming the Citrea whitepaper requirement; Luke Dashjr "utter insanity" post, April 26, 2025).

Bitcoin Core IRC logs, archived at erisian.com.au and gnusha.org: May 1, 2025 weekly developer meeting (four-options framing, middle-path foreclosure, Citrea references by Sanders, Poinsot, and Wuille). May 2, 2025 #bitcoin-core-dev (Sanders lock request, Towns inscription joke). May 2, 2024 #bitcoin-core-dev (Jonas moderation guidelines authorship confirmation by achow101, Zipkin volunteering to enforce). May 6, 2025 #bitcoin-core-dev. May 22, 2025 weekly developer meeting (Wuille presents the relay statement gist). October 14, 2024 #bitcoin-core-dev (Wuille announces the CoreDev kill/shill session). All retrieved and preserved.

Published statements and articles

bitcoincore.org: 31-developer relay statement letter, June 6, 2025. Bitcoin Core release notes, v30 (October 10, 2025).

Personal blog: Antoine Poinsot, "Relay policy drama," antoinep.com/posts/relay_policy_drama.

Adam Jonas, "Evolution of Bitcoin Core Priority Projects," adamjonas.com, October 31, 2024. Annual CoreDev retrospectives, adamjonas.com (2019–2020 through 2024–2025), preserved.

Sanders public gist on the merge decision, gist.github.com/instagibbs, May 5, 2025. Wuille relay statement gist, gist.github.com/sipa/2521731e65ba779e3ce9f9305c6a538c.

Podcasts and interviews

Bitcoin Optech "Waiting for Confirmation" series, newsletters #256 and #257 (Zhao and Erhardt, June 2023). Bitcoin Optech recap podcasts hosted by Mike Schmidt: June 29, 2023 (Murch and Zhao); August 10, 2023 (Murch, Harding, Todd); May 6, 2025 (Schmidt, Erhardt, Sjors Provoost on the OP_RETURN debate); September 30, 2025 (Optech #373 recap with Erhardt). Full transcripts preserved.

Gloria Zhao on the Stephan Livera Podcast, October 2024. Gloria Zhao and Jameson Lopp, Bitcoin Magazine interviews with Shinobi at MIT Bitcoin Expo, April 2025. Bitcoin Magazine article by Shinobi on Casey Rodarmor and Runes, September 26, 2023.

Amiti Uttarwar on the What Bitcoin Did podcast, October 2020 ("extremely extractive," "small group of loud, angry voices"). Antoine Poinsot on the What Bitcoin Did podcast with Danny Knowles, October 2025. Jimmy Song on The Bitcoin Matrix with Cedric Youngelman, April 2026. Adam Jonas address and panel at the MIT Bitcoin Expo, April 2025 (livestreamed by Bitcoin Magazine).

Africa Bitcoin Conference, Adam Jonas keynote, December 2024.

External press

Paul Kiernan, "Bitcoin's Future Depends on a Handful of Mysterious Coders," Wall Street Journal, February 16, 2023. Documented Samuel Dobson on the funding-control question and Coinbase's SEC disclosure on Bitcoin Core developer organisation as a regulatory risk factor.

On-chain data and analysis

Marcel Hernandez (X handle @oomahq; GitHub 1ma), non-standard OP_RETURN counts, January–May 2025. Renaud Cuny, blockspace analysis, December 2025. Block subsidy and inscription revenue figures verified independently against the blockchain record.

X (Twitter) posts, screenshots preserved

Luke Dashjr (@LukeDashjr): public service announcement December 6, 2023; CVE confirmation December 10, 2023; June 7, 2025 substantive NACK on the 31-developer letter; October 12, 2025 disclosure of security disclosure list removal.

Gloria Zhao (@glozow, account deleted May 15, 2025): May 2, 2025 X post defending moderation; June 8, 2025 post; merge-day statements (screenshots preserved before account deletion).

Jameson Lopp (@lopp): April 2025 Citrea conflict denial; April 30, 2025 reply to Bob Burnett; September 3, 2025 Citrea "entertaining future" post; October 28, 2025 Concept ACK on DNS seed removal PR.

AJ Towns (@ajtowns): October 30, 2025 Citrea/signet question quoting hampus_s.

Adam Back (@adam3us): September 25, 2025 reply to Chris Guida on release notes (with Guida same-day rebuttal preserved); February 17, 2026 post on OP_RETURN limit value.

Matt Corallo (@TheBlueMatt): April 24, 2018 diversity recruitment thread; April 2025 "stupid drama" post celebrating jesterhodl block.

Antoine Poinsot (@darosior): August 6, 2025 post characterising Knots migration; subsequent posts on the OP_RETURN debate.

Peter Todd (@peterktodd): September 26, 2023 PR #28408 comments; April 1, 2024 thread with Rodarmor on Libre Relay miner participation; May 4, 2025 Stacker News confirmation.

Samson Mow: May 5, 2025 "PR laundering" post and subsequent clarification; replies on the moderation patterns thread.

Jason Hughes (Ocean Mining, also wizkid057 on GitHub): June 9, 2025 "52 days" indictment on X; April 28, 2025 substantive opposition comment on PR #32359 quoted in this article.

Nick Szabo: October 8, 2025 recommendation to run Knots; subsequent posts including October 2 criminal liability concern.

Bitcoin Core Project (@bitcoincoreorg): June 7, 2025 distribution of the 31-developer letter.

Lysander (@UnderCoercion): May 5, 2025 naming of Antoine Poinsot following the Keysa thread. Keysa (@SimplestBTCBook): May 5, 2025 question to Todd. Marcel Hernandez (@oomahq): on-chain data posts.

Funding and institutional records

Citrea Series A funding announcement, October 31, 2024 (Lopp as investor disclosed). OKCoin and Paradigm joint grant announcement, January 2023 (Falke). Matt Huang, matthuang.com/funding_bitcoin_development, July 2020 (Towns funding). Chaincode Labs welcome posts: Antoine Poinsot, October 14, 2024; Gloria Zhao, March 31, 2025. Brink Engineering Impact Report 2025 (brink.dev). Localhost Research (5–6 Core developers at approximately $333,000 per developer per year). MIT Digital Currency Initiative (publicly documented Core developer funding includes Cory Fields, Marco Falke, AJ Towns, Wladimir van der Laan, and Sebastian Kung). Spiral (Bitcoin development arm of Block) annual budget disclosed by Steve Lee. OpenSats 2025 Year in Review (opensats.org/blog/2025-year-in-review). Human Rights Foundation Bitcoin Development Fund (hrf.org). Peter Todd, Libre Relay documentation. National Vulnerability Database, CVE-2023-50428.

Archives

Wayback Machine snapshots: PR #32359 thumbs reactions, May 13, 2025 (401 thumbs-down to 102 thumbs-up).