S15 E71: Amir Taaki on Bitcoin, DarkFi, and the Privacy Year

This was the last episode of Season 15, recorded as a year-end conversation with Amir Taaki, and it ran long enough to feel like a second shift. We had already done a Twitter Spaces at Christmas that went on for nearly eight hours. He joked that I probably wanted him for New Year's as well. He was right. I wanted a review of a year in which Bitcoin privacy became something you were supposed to be embarrassed about, and I wanted it from the person who had built Dark Wallet when CoinJoin and Tor in a wallet still sounded like a research preview.

If you only have a commute, this write-up is the compressed version. The full recording is more than seven hours, and there is a corrected transcript on the site if you want the uncut argument. I am not going to pretend I covered every side quest. I am going to stay with the spine: Layer 2s and OP_CAT, who Amir thinks has been running Bitcoin, DarkFi, and what happened to privacy tools in 2024.

This was his second time on the show. Later I sat down with him again in Season 17, including the conversation with Harry Halpin and Christopher Cialone and the one with Prof. Richard Werner. Those are later talks. This one is the New Year's tape.

Perfect Is the Enemy of the Good

I opened with the year as I saw it. The company behind Wasabi was shutting down. The Samourai developers had been arrested. ETFs had been approved. Privacy on Bitcoin looked, to me, like a taboo, just as the official story was becoming "mainstream adoption." Amir did not start with a eulogy for mixers. He started with Layer 2s.

If Bitcoin gets ZK-rollup support, he said, it becomes a programmable base that people can attach their own sidechains to, including a privacy chain. That is the optimistic reading. The pessimistic one is Ethereum's, or Solana's, depending on which failure you want to study: bridges are cumbersome, users are lazy, and they may just stay on the chain they already understand. Gavin Wood left Ethereum after losing a fight about sharding versus Plasma, later got vindicated when Ethereum moved toward that kind of scaling anyway, and still did not see the crowd follow him to Polkadot. A feature existing is not the same thing as people using it.

I asked which Bitcoin L2 he actually liked. He did not want to pick an opcode winner. He wanted to talk about Eli Ben-Sasson, who had been trying to make a Bitcoin L2 and had gotten tired of waiting for a perfect new opcode.

"OP_CAT is a very simple opcode. It just says, take two pieces of data and join them together." It used to exist in Bitcoin. Eli's argument, as Amir restated it, was not that CAT is the final design. It was that waiting for a flawless opcode is how you prevent anyone from building.

"Perfect is the enemy of good. Let's bring OP_CAT... and then if there's a better opcode, we've got as much time as we want, but that way we're not holding up people building actually L2s on Bitcoin to bring some of these other features like scalability, like privacy. So I kind of, I think his logic is sound there."

I have been saying a version of this for a long time, including when StarkWare was sponsoring the show and people treated CAT as an NFT plot because they had seen it next to Taproot Wizards. Concatenation is not a demon. It is a computer instruction. The objection I keep hearing is that nobody can enumerate every future use, and that the unknown itself is a reason to do nothing. Amir's counter is older than the 2024 mailing-list fights. In 2011, after a script bug, several opcodes were commented out as an emergency patch, with the idea that they would be studied and restored in a controlled way a year later. That restoration, in his telling, did not arrive. Policy and relay rules then made the network hostile even to things that were valid, like early multisig. He used to broadcast those transactions to Luke Dashjr's pool because other miners would not relay them.

If you want a one-sentence version of his L2 position, it is this: stop waiting for a ceremony. Enable the simple thing, let people build, and improve it later.

Who Runs Bitcoin, in His Telling

Amir does not describe Bitcoin's politics as a debate about block size. He describes it as a fight about who gets to steer the project, and about which values survive.

He talked about a class of "business guys" in 2011 and 2012 who, in his words, "started to try and do a hostile takeover of Bitcoin." In his account they wanted banks and payment networks, they treated anarchism as an embarrassment, and they talked down Silk Road. He put the Bitcoin Foundation and people around that world in that camp, including Roger Ver. Then he did something more interesting than a simple villain story. He said the other side won the block size war, and that the winners now talk as if they "saved Bitcoin from a coup, from a hostile takeover."

"That's how they see it. And it's not completely wrong."

His complaint about the present is not that those business people were secretly right about 8-megabyte blocks. It is that, in his view, they at least had a vision for Bitcoin's future, and that the people in charge now do not. "They're not fit to lead Bitcoin, to take Bitcoin forward. You know, to take it anywhere, really, just kind of squat it." He thinks a new influx of people from the Ethereum world is arriving with ideas about an ecosystem, and that a certain kind of Bitcoiner feels that as competition and calls it a shitcoin reflexively.

He also called Bitcoin maximalism, in its "there shall be one economic system" form, authoritarian, even communist in flavor. I am leaving that as his provocation, not as a slogan I need to underwrite. The part I am willing to carry in my own voice is narrower. If a community's immune system is only good at saying no, it should not be surprised when the builders go somewhere else, or when "adoption" arrives as ETFs and custodial piles instead of as tools.

The Peter Todd section of this interview is the one that will make people angry, so I want to be precise. Amir was not, on this tape, claiming he knows what Peter does now. He was talking about the old CoinJoin bounty on Bitcoin Talk, about John Dillon, and about emails that later got a much larger audience through the HBO documentary. In Amir's account, he implemented the first CoinJoin, later released Dark Wallet, and still got nothing from that bounty. He says Peter, who had been brought in as a kind of chief scientist "as a counterbalance to Gavin Andresen," helped stop the funds from going to Amir's crew, including by dealing with a person who presented themselves as an agent. Amir thinks any decent person would have told that person to get lost. He also thinks the lack of an apology, years later, is its own tell.

I am reporting his account of those events, not retrying them in a blog post. I also told him, on the tape, that one popular interpretation is that Peter was John Dillon. Amir said that was not the part he was on about. He wanted people to read the bounty-fund thread. If you are going to take a side, take it from the primary documents, not from my summary of a seven-hour livestream.

DarkFi, Dark IRC, and Anonymous DAOs

The product conversation is why I wanted him at New Year's, not only the history.

DarkFi, as he described it in December 2024, is a chain for anonymous smart contracts, with generalized programmability and ZK proofs, plus actual products people can use. The first product is chat. They built it, he said, after watching what happened to the Tornado Cash developers, whose Signal conversations were used against them.

"We don't want to go to jail. What do we need? We need an anonymous chat. So we made like the world's strongest fully P2P anonymous chat in DarkFi."

The chat is Dark IRC. At the time of the recording it was still a command-line thing, with a UI he said he was building for release "in this next month," including a phone app. People in the live chat were already asking for a link and comparing it to SimpleX. His answer was: go to the DarkFi website, open Community, and join. He also wanted deniability, not only encryption. If a message cannot be tied to you, a prosecutor has a harder time turning a group chat into an org chart.

He is not shy about the political use case. AssangeDAO, he said, raised $55 million "to free Assange by all means necessary," which he contrasted with a Bitcoiner boasting about a $100,000 airfare. DAOs already exist, "but they're all transparent. But on DarkFi, we have anonymous DAOs." The work they had just done was to split roles inside a DAO: who can execute a proposal, who can see the treasury, whether a running vote stays hidden until the end. The point is to make it possible to pool capital and act without turning the membership list into a target.

He kept coming back to a distinction I wish more "privacy projects" would say out loud. "When we say free speech, we mean 100% pure free speech." And: "When we talk about anonymity, we're not talking about this privacy bullshit." In his mouth, a little bit of Taproot folklore is not anonymity. He wants no compromises, or he does not want the word.

The launch status, as of that recording, was: a new testnet was out, "the final testnet before we go to mainnet," which they wanted to run for a few months; a wallet was "coming out like, you know, I don't know, in a couple of weeks"; and a desktop and Android client would use the p2p network they had been optimizing for phones. I am leaving those timelines as December 2024 statements. This is not a shipping blog.

What I took from it was the design order. They did not start with a token narrative. They started with "we do not want our Signal chats read in court," then built a chat, then put the chat next to payments, tasks, and DAOs in a wallet that can grow plugins later. You can think that is grandiose. You cannot say it is unmotivated. 2024 gave them the motivation in public.

The Year Privacy Tools Got Hunted

My frame for the year is still the simple one I used when I opened the show. "We saw how privacy on Bitcoin became something that is taboo nowadays. They shut down Wasabi Wallet, at least the company behind it. They arrested the developers behind Samourai, and then they approved ETFs." The market got a regulated product, and the people who had been doing the unromantic work of trying to make bitcoin harder to trace got a raid, a shutdown, or a lecture.

Amir's frame is wider than Bitcoin Twitter. He talked about Tornado Cash, about Roman Storm trying to raise money for a legal defense after delivering a tool the Ethereum ecosystem was happy to use, and about a broader climate in which speech itself is being treated as a crime in some places. I mentioned Virgil Griffith, and the way ecosystems tend to abandon the person once the case is ugly. I also said, on the record, that I want Roman on the show. "I do believe that writing software is not a crime." Amir's own starting point for Dark IRC was the same legal weather. "The Tornado Cash devs, they put them in jail because they got access to all their Signal chats."

This is the lunarpunk part of Amir's politics, or at least the part that does not dissolve into aesthetics. Privacy is not a branding exercise for people who already live in safe jurisdictions and want a slightly cuter receiving address. It is a condition for organizing, getting paid, and speaking, especially if your work is the kind of work that attracts prosecutors. He was scathing about privacy projects that tell you to join a Discord, and about a cryptocurrency culture that still lives on Twitter, which he called a failure for a community that claims to care about decentralization.

I have my own version of this complaint, and I have made it on the show before: we spent a year watching the official Bitcoin conversation become more comfortable with Wall Street, and less comfortable with the tools that make self-custody meaningful. Jameson Lopp can call himself a paranoid crypto anarchist and still be treated as mainstream-adjacent. The developers who shipped the actual privacy wallets got a harder category. That gap is the story of 2024, and it is why I wanted Amir at the end of the season.

Payments, Monero, and What He Thinks Is Already Solved

Later in the marathon I asked him about anonymous payments, and he was almost casual.

"I think anonymous payments is pretty much solved though now."

His current favorite, on this tape, was FCMP on Monero, which he called elegant, simple, and fast, especially next to the moving parts in zk-SNARKs, even though DarkFi itself uses zk-SNARKs. He also liked that Monero hides amounts with confidential transactions. He praised the work, and in the same stretch praised Luke and Liam Eagen in a way that made me message Liam on the spot. I have been trying to get Liam on the show for months. Amir also told me I should get Roman Storm.

I brought up Dark Wallet as a stepping stone, the thing that later wallets were able to walk on top of, including Samourai and Wasabi. He did not need me to turn it into a monument. Hal Finney, as I reminded the audience, was already talking about making Bitcoin more private days after "Running bitcoin." The struggle is older than any one wallet brand. What changed in 2024 was not the need. It was the legal weather around anyone who tries to ship the need as a product.

If payments are "pretty much solved" on other chains, that is not, for me, a reason to stop caring about Bitcoin. It is a reason to be honest that Bitcoin, as most people use it, is still a transparent ledger with optional after-the-fact rituals. I have written about those limitations before. Amir's answer is to build a place where anonymity is the default, and to stop asking a hostile relay policy to become a privacy system.

Cypherpunks, Not a Comments Section

Underneath the opcode talk and the product demo is a definition of the job. Amir got into Bitcoin because he wanted "cyber money on the internet," the Neuromancer version, "radical cyber stuff... cyber anarchy." He contrasted that with a later mood in which some of the same early developers lost interest in scripting and started talking about Bitcoin as strictly money. He also contrasted it with a culture that, in his view, moralizes, ossifies, and then wonders why the interesting work happens somewhere else.

I do not agree with every blast radius in a seven-hour stream. I do not need to. The useful pressure in this episode is the same pressure I want on myself as a host. If the name of the show is Bitcoin Takeover, then the job cannot only be to comment on price, ETFs, and whoever is occupying the timeline this week. Somebody has to keep building the tools that make a takeover possible, or at least make it possible to speak and transact without asking for permission.

Amir's version of that, at the end of 2024, was DarkFi, Dark IRC, anonymous DAOs, and a refusal to call watered-down features by a stronger name. My version, on this show, is to keep putting those conversations on the record, even when they run past midnight and refuse to become a neat listicle. If you have seven hours, listen to the whole thing. If you do not, start with OP_CAT, the chat he built so his team would not live on Signal, and the year that made that decision look less paranoid than it might have looked in 2023.