S15 E28: Amaury Séchet (Deadalnix) on Forking Bitcoin, Big Blocks & Scaling

Season 15 Episode 28 is the April 2024 deposition of the man most Bitcoiners only meet as a cartoon villain. Amaury Séchet, better known as Deadalnix, is the developer who shipped the client that forked Bitcoin into Bitcoin Cash in August 2017, later left that chain after a funding fight, and by the time we recorded was building eCash with a second consensus layer on top of Nakamoto proof of work. The site blurb framed the questions as provocations: why did he not destroy Bitcoin when he learned about the 2018 inflation bug, why do large blockers keep forking, and how did Craig Wright get exposed inside big-block circles. Amaury answered like an engineer who still thinks game theory beats white-paper theology.

I later sat with him again for a much longer archaeology session in S17 E19. This article sticks to the first tape: roughly two hours and forty-three minutes recorded around 5 April 2024, while Bitcoin Cash had just had a messy post-halving day and BTC was about to halve later that month.

Engineer First, Fork Second

Amaury says he was interested in Bitcoin long before he coded for it. As a developer he arrived late, around 2016, because that is when the scaling debate heated up into something matching his expertise. Scaling, in his framing, is never one silver bullet. You remove a bottleneck, then you meet the next one. The first obvious bottleneck is block size: relatively easy technically, socially radioactive. He also rejects the cartoon version of his own camp. Growing the limit forever and hoping for the best is not a plan. With the tech available then, he said, things work fine up to about 20 MB. Beyond that you need real engineering, not just a bigger integer in a config file.

His favorite analogy is a road trip from France to Romania. Raise the highway speed limit and you arrive sooner, until the car and the pavement cannot survive the speed. Keep raising the number and the car cannot even go that fast. Block size works the same way. Raise it a bit and you buy capacity. Raise it blindly and you discover that propagation, validation, and incentives were the real walls.

Before Bitcoin Cash, he was interested in extension blocks (Christopher Jeffrey’s soft-fork-shaped capacity proposal) because hard forks looked politically doomed. Then the New York Agreement and SegWit2x vacuumed the oxygen out of that path. Amaury’s read on SegWit2x was blunt even before it collapsed: activate SegWit first, promise a later capacity increase, and you have built a bait and switch. When that became the big-block mainline strategy, he concluded the only remaining move was a minority fork. That is where Bitcoin ABC comes from. The original purpose, he told me, was to implement extension blocks. When that died, he scraped the extension-block work and turned the client into the Bitcoin Cash activation vehicle.

Minority Fork, Not “The Real Bitcoin”

Why ABC and not Bitcoin XT or Bitcoin Unlimited? Because those projects still wanted to be Bitcoin. A minority fork is an admission that Plan A failed. Amaury was fine with that admission. ViaBTC mined the first BCH block. Bitmain was supportive early. Payment-oriented businesses that were already choking on multi-dollar BTC fees showed up quickly. Many of those services, he noted dryly in 2024, had since disappeared.

I asked whether Bitcoin Cash killed SegWit2x by offering big blocks first. He said no. SegWit2x was dead on arrival. Big blockers, in his autopsy, misunderstood Bitcoin’s game theory while small blockers understood it brutally well. His exhibit A is UASF: nodes announcing they would activate SegWit and refuse a capacity-increasing client. He compares it to the movie chicken game where you rip out the steering wheel and throw it out the window so the other driver knows you cannot swerve. A few people on the big-block side tried to present the same game theory. Most of that camp, Amaury says, kept quoting the white paper as if paper could repeal incentives.

On branding he is almost allergic to the phrase “the real Bitcoin.” Different forks can have equal legitimacy as flavors of the same project pursuing different values. The moment you declare exclusive ownership of the word, you antagonize everyone else and burn energy that could have gone into building. For a wider fossil record of that naming war, my older piece Bitcoin Cash: The Big Block Breakaway still maps the split from the outside.

White Paper as Aspiration, Not Scripture

The white paper’s opening goal matters to him: peer-to-peer electronic cash. That goal gives legitimacy. Treating every technical sentence as inviolable does not. Nodes do not follow the “longest” chain in the naive sense; they follow the heaviest proof-of-work chain. He cites Adam Back’s courtroom correction of Craig Wright on how proof of work actually works as a comic example of people defending the document against the software. The first version of anything is never perfect. Pretending otherwise, he argues, creates religious tension for no engineering reason.

We also talked through the fresh BCH halving hiccups: long gaps, hash fleeing to BTC after emission cuts. Amaury treated that as boring economics. Miners allocate by price ratio. When BCH halves, hash leaves until difficulty adjusts. When BTC halves later, the opposite happens for a while. No conspiracy required.

CVE-2018-17144: He Could Have Made It Worse

Then the spicy section. On 17 September 2018 an anonymous reporter mailed a crash bug to Pieter Wuille, Greg Maxwell, Wladimir van der Laan, Deadalnix at Bitcoin ABC, and SickPig at Bitcoin Unlimited. Within hours Matt Corallo identified that the same root cause was also an inflation vulnerability: CVE-2018-17144. Amaury was on that email thread. He is careful about credit. He did not discover the bug in the code. He learned it existed early, reviewed it, and cooperated on the fix instead of trying to print coins or force an emergency spectacle on BTC.

Why help the people you forked away from? Because destroying value was never the point, because security disclosure is a repeated game, and because once you burn the trust of people who share vulnerabilities with you, nobody shares again. “You only have one reputation,” he said. Fame for wrecking Bitcoin looked, to him, like Craig Wright’s industrial fake-document fame: loud, useless.

Developers on both sides of the ticker war, he claims, have been cooler-headed about serious bugs than Twitter marketing suggests. The bug itself was not even a consensus soft fork. It was a subtle transaction-validation optimization that multiple reviewers (Core, ABC, Unlimited) missed on first pass. Ossification fans who think “no more soft forks” equals “no more footguns” get no comfort from that story.

OP_CODESEPARATOR and the Craig Wright Stage Trick

His Craig Wright story is the one I wanted on tape. Before Bitcoin Cash, while chasing extension blocks in San Francisco with Christopher Jeffrey, prominent names were floating Craig as Satoshi.

Amaury and Christopher designed a test: ask about OP_CODESEPARATOR, a weird opcode with almost no real use cases. If the man is Satoshi, he should explain the strange design. Weeks later, at a conference where John Matonis was billed to speak, the slot bait-and-switched into a Craig Wright presentation.

Amaury asked the opcode question live, played dumb, let Craig talk, then realized the proposed use cases did not work. Christopher independently reached the same conclusion over chat. Throw jargon at the wall and it sticks with non-engineers. It does not stick with people who know the script system.

After that, Amaury was done giving Craig the benefit of the doubt, though the ecosystem kept throwing the man back into his lap: selfish mining arguments, transaction ordering fights, confident gibberish that sounds as technical as a careful rebuttal to a lay audience. He was not maximally loud in public at first, partly because he worried that a crusade would give Craig more spotlight. In private he warned people he considered valuable, with mixed success. That hesitation is one of the few places on the tape where he openly wonders if he chose wrong.

Three Factions, One Divorce, Then eCash

Bitcoin Cash, in his telling, never had one unified big-block soul. After SegWit2x failed, two unhappy groups poured in: the Craig-is-Satoshi camp, and a Bitcoin Unlimited / Roger Ver-adjacent camp that still wanted big blocks on the main BTC ticker and arrived carrying the same strategies that had already failed. Those factions never coalesced cleanly with the people who understood minority-fork game theory from day one. Amaury’s short history of BCH’s relative decline (he cites a slide from roughly 12% of BTC’s value toward 1%) blames that politics more than a single code change.

The straw that broke his camel, as he describes it, was a chained-transaction feature fight with Roger Ver. (Roger told his side of the Bitcoin Cash years about a week earlier, in S15 E26.) ABC wanted funding to maintain the complexity. Roger publicly claimed he had funded them and they refused to deliver. Amaury says that money never arrived.

Fighting Roger’s media firepower while trying to ship software was, in his words, not what he signed up for. Avalanche-style finality had been on the BCH roadmap in spirit before Avalanche itself existed; when the paper appeared, it matched what they wanted. After the exit, that work landed on eCash (XEC): still SHA-256 proof of work, still closer to BTC consensus rules than BCH is in some places, plus Avalanche for fast agreement.

Avalanche’s upside: network agreement in roughly two seconds. Downside: no historic artifact you can sync after disappearing into the woods for five years. That is why eCash keeps Nakamoto consensus underneath. Live Avalanche rounds can finalize a tip so nodes refuse fraudulent late reorgs. Pure Avalanche chains like AVAX, he argues, force a trusted bootstrap if you were offline. Hybrid is the point. For a later magazine-length autopsy of the funding split and the XEC experiment, see eCash (XEC): The Fork That Taxed Itself.

Drivechains, Lightning, RBF, and the Dev Tax Ghost

Paul Sztorc snuck a French large-blocker job ad into the middle of the show. Amaury’s reply was warmer than tribal Twitter would predict. Drivechains are interesting. The hard problem is the reverse peg: getting coins back to the main chain when two mining systems must agree. eCash’s “subnet” idea attacks a similar problem with Avalanche so the reverse peg can be much faster. If he were not on eCash and were a new developer joining Bitcoin work in 2024, he said he would probably look at Drivechains with Paul.

Lightning gets respect and a demotion. Cool technology. Streaming money, deposits, specific use cases: great. General-purpose scaling for the planet: no. He reminds listeners that even early Lightning presentations assumed very large base-layer blocks if you wanted worldwide scale.

Liquid is “not very interesting” to him personally: more federated than a pure custodian, still one of the intermediary patterns you get when the base layer refuses to scale. RBF is neither a morality play nor a Core conspiracy.

On a congested fee market, miners make more money with it, so even if node software banned it they would reimplement it. On BCH it stays quiet because fees are low and the edge is tiny. Change the incentives or stop pretending code comments repeal economics.

On the infamous “dev tax,” Amaury’s 2024 defense is specific. He says he never personally took money from that mechanism; it paid other developers. Protocol work at Core or Ethereum scale burns millions a year; ABC was asking for far less and still could not get sustainable funding.

By the recording date, eCash had already removed the mechanism from consensus rules so unhappy users could change it without a fork, and half the proceeds (as people call them) went through a council process he pointed to at gnosis.e.cash. When a Twitter account called him a scammer, he answered with market-cap math: after ABC left, he claims BCH’s debasement cost the community on the order of $4.2 billion, dwarfing the contested funding. Read that as his claim on an April 2024 microphone, not as a settled audit.

Enough Money, Manufactured History

I asked the rude question: would he be richer if he had just held BTC and never forked? Maybe, maybe not. Alternative timelines still have events. He thinks the divorce helped Bitcoin by separating camps that were forced into a dysfunctional marriage. He still holds BTC. He sold some for BCH at the fork but not all, and he did a lot of cross-exchange arbitrage when prices were chaos. Soft answer on regrets: many tactical do-overs, little belief that the destination would have changed.

The line that stuck with me: “I have something that most people will never have. It’s enough.” Not Roger Ver rich. Enough to stop maximizing the dollar number and build the money he wants to exist. He also grants the small-block value proposition more cleanly than meme history allows. A system that will look the same if you wake from a fifteen-year coma has worth. BCH and eCash might be better cash machines in fifteen years, or someone might wreck them. Predictability is a feature people are allowed to buy.

Near the end he is bleak about written Bitcoin history. Most accounts of the block size era, he thinks, are motivated storytelling by wealthy narrators with goals. He had not read Jonathan Bier’s book cover to cover; excerpts seemed biased but trying. He has history with Steve Patterson and said he had zero trust Patterson would tell an accurate Roger-adjacent story, while admitting he cannot dismiss a book he has not read. He offered to connect me with quieter witnesses who no longer want the mud. Jeff Garzik, still on my white whale list then, got a nod as someone less political and worth chasing.

Closing the First Tape

CTOR (canonical transaction ordering) and OP_CHECKDATASIG get a calm engineering epilogue: CTOR turns a block into a set so propagation does not drown in ordering metadata; CDS unlocks signatures over arbitrary data and sprouted use cases nobody planned. Craig opposed CTOR; Amaury suspects Craig needed a fork excuse more than a technical objection. Bitcoin ABC by April 2024 was focused on eCash. XT and Classic were gone. Unlimited had wandered toward Nexa. BCHN, a fork of ABC, was the main BCH node shop, with not many hands on the node itself.

Follow him as Deadalnix on Twitter/X and GitHub. The name means nothing. He claims he has never even pronounced it in his head. Good luck with eCash, I said. Good luck with the podcast, he said. Two years later we would do the longer sequel. On this first night, the quieter story was already enough: a minority-fork engineer who reported the inflation bug instead of cashing it, grilled a fake Satoshi with an opcode trivia question, and still talks about speed limits on the road from France to Romania.

Listen to Deadalnix (Amaury Séchet) on Spotify, Apple Podcasts, YouTube, and More