S15 E70: Jeff Garzik on Hemi and Bitcoin Development

I chased Jeff Garzik for months after he posted a video about the origins of Satoshi Nakamoto, and about why he thinks Satoshi was one person. I wanted him on the show for older reasons than that. In October 2010 he was the first so-called big blocker, the person who proposed lifting Bitcoin's block size to 7 megabytes. Satoshi wrote back and said it was a bit too early. A man who goes by Arctic Miner has been haunting that same Bitcoin Talk thread ever since, coming back year after year to ask whether it is still too early.

Jeff spent about five years as a Bitcoin Core developer. Before that he worked on the Linux kernel. After that came BitPay, then Bloq, and now Hemi. He has been in this thing long enough to have received emails from Satoshi and to have watched almost every scaling sermon fail on contact with users. In December 2024 I finally got him on the microphone.

The First Big Blocker, Fourteen Years Later

There is a version of this introduction that turns Jeff into a statue: early Core dev, Satoshi correspondent, first person to try to make the blocks bigger. I am less interested in the statue than in the fact that he is still building. A lot of his contemporaries left, burned out, or decided that Bitcoin was already finished. Jeff talks like someone who still thinks the job is incomplete.

He has a phrase for how long he has been here. He calls it his blockchain age, and he puts it at 14 years. In his mouth, almost everyone else is a newcomer. That could sound vain, and sometimes it does. It is also a useful calibration. When a person who was in the room in 2010 tells you that Bitcoin will not put "all the coffee" on one chain, he is not discovering modularization because it is fashionable. He is restating a conclusion he has been living with since the first time the block size became a fight.

I also cannot hear his name without thinking about the later block size war, SegWit2x, and the way Bitcoin's historical memory turned some of those people into villains. We got to the spicy version of that later in the interview. At the start, I just wanted him to explain the thing he is actually shipping.

King and Queen, and a Bitcoin Node Inside an EVM

"Hemi is a Bitcoin and Ethereum Layer 2 blockchain that bridges both the king and queen of crypto."

That is the sound bite, and he means it literally. Bitcoin is the root of the tree, the thing he still expects to be here in 10, 20, 50, or 100 years. Ethereum is "Grand Central Station": if you are going anywhere in crypto that is not Bitcoin, you are probably going to pass through some Ethereum-based technology. Hemi is his attempt to sit on both rails at once.

The technical claim that matters is tighter than "we made a bridge." Hemi uses a protocol called proof of proof to secure its chain into Bitcoin, and it embeds a Bitcoin node inside an EVM node. Jeff called that the tightest integration he knows of among these Layer 2s. The result, in his words, is that a smart contract can read the Bitcoin chain and create Bitcoin transactions.

"It's really kind of superpowered Bitcoin script... I've been waiting a decade for that level of power in scripting and smart contracting on Bitcoin, and I argue that now it's here with Hemi."

I asked the obvious nerd question: why Ethereum at all, when there is always some other virtual machine around the corner, or Simplicity, or a language someone swears can fit on a T-shirt? His answer came from Linux, not from a white paper. Software lasts a long time. Tooling, auditors, and the scar tissue around sharp edges take years to accumulate. The EVM has been through that Darwinian process. Other designs may be cleaner on a whiteboard. They have not had the same number of people trying to break them, or the same number of humans who already know where the bugs hide.

He used a phrase I have not been able to get out of my head: wet cement. You do not keep ripping up a foundation that has already been walked on by millions of users just because a new floor plan is more elegant. You import Bitcoin's strengths into an environment that already has contracts, wallets, and developers.

Proof of Proof, and Software That Survives the Internet

Proof of proof, as he described it, is not proof of stake in a costume. Hemi embeds proofs in ordinary Bitcoin transactions and leans on Bitcoin's proof of work.

"Proof of proof is an alternative to merge mining that relies on Bitcoin's censorship-resistant properties."

The concrete mechanism is Keystone hashes. Jeff said he would want his CTO, Max Sanchez, to correct the exact parameter if he had it wrong, but the version he gave on the show was that Hemi emits a Keystone hash into every 12th Bitcoin block, and that this is what they call super finality on the Layer 2. Sanchez had already live-tested the protocol on a chain called VeriBlock. That field test is doing a lot of work in Jeff's epistemology. He quoted Linus Torvalds, or at least the line he loves from him: the internet is the greatest test lab in the world. You can test on your laptop, or in a lab, and you still do not really know until the software has survived out in the open.

If you want the long version, he pointed to docs.hemi.xyz. I like that more than a podcast diagram. A bridge or a rollup that cannot be read about in documentation is usually just a vibe.

He also made the security argument that I wish more people would make without turning it into marketing. Bitcoin and Ethereum have been sitting there for years with enormous bounties on them. At the time we recorded, Bitcoin was around $2.1 trillion. If you think you have found a miracle bug, the market will pay you to prove it. The fact that these systems are still standing is, in his view, evidence. Satoshi did not invent a new hash function or a new signature scheme. He used off-the-shelf cryptography that had already been punished by cryptanalysis, and then combined it in a new way. Jeff wants Hemi to inherit that posture, not to invent a cute new trust assumption and hope nobody notices.

Privacy by Default, or Nothing

Mid-interview we got a shout-out from Amir Taaki, who was apparently listening on Twitter. That was a nice accident, because Jeff immediately went into the part of the conversation Amir would have cared about. He talked about Dark Wallet, DarkMarket, and the later OpenBazaar chapter, and then said the thing he thinks the industry still has not faced.

Bitcoin and Ethereum, in his view, are not anti-privacy. They also do not default to privacy. "A version of the unicorn world in my head kind of has a Zcash shielded-transactions-only type of privacy view." If crypto is going to be self-sovereignty for billions of people, it needs privacy by default, not as an advanced setting for the already paranoid.

I agree with that, and I said so. I also told him why the sentence makes me sad. He was at BitPay when Bitcoin still had a chance to become everyday money, and he watched that use case shrink while the official story was rewritten into "store of value." Fungibility used to be part of the conversation. Then it became a niche interest, or a moral suspicion. Jeff is still willing to call the default-transparent chain insufficient. I find that more encouraging than another developer telling me that privacy is a layer, a mixing service, or a problem for later.

Drivechains Are Hardcore, and Seventy Other L2s

Someone in the audience asked about Paul Sztorc and miner-enforced sidechains. Jeff did not dodge it.

"I've always been a fan of Paul specifically... Drivechains are one of the few sidechain-type proposals that I take seriously."

The reservation is not philosophical. It is about coordination. To make Drivechains work you have to negotiate with miners and get a slice of hashrate to enforce the new protocol. "If it's a video game, the level of difficulty is hardcore." Jeff would rather design systems that do not require that negotiation just to inherit security. He is not sneering at Paul's design. He is admitting that "go convince the miners" is a very high difficulty setting.

Then he zoomed out. He said he was aware of at least 70 Bitcoin L2s and at least 50 Ethereum L2s, and that as far as he knew Hemi was the only one trying to bring Bitcoin and Ethereum together, with Bitcoin as the security layer, while also embedding a Bitcoin node in the EVM instead of air-dropping an EVM onto a Bitcoin anchor and calling it a day. He said they had been building Hemi for two years. Most of those other experiments, in his pragmatist voice, will fail, either technically or in the market, and that is the normal process of innovation.

What he dislikes is the opposite attitude: the chirping chorus that says Bitcoin is already perfect, so please stop building. "I'm far more optimistic." After watching some of the original Core personalities go dark or sour, I was glad to hear that from him. Optimism is not the same thing as saying every L2 is good. It is the refusal to treat stagnation as a moral victory.

Moderates, Tokens After Mainnet, and Thirty dApps Loading

A SideShift question from the chat asked the thing every listener was waiting for: does Hemi have a token, and do they want to be listed?

Jeff would not give a date. "The lawyers will tell [me] directly." The order of operations, as he stated it in December 2024, was clear enough. They were running an incentivized testnet, "launching mainnet very soon, within probably the next 60 days if not sooner," and "the acronym everyone's looking for will come after the mainnet launch." So the token, in that telling, follows the network. I am leaving that as the December 2024 claim. This article is not a token calendar, and I am not going to update a two-year-old interview with a press-release ending.

The user he wants is what he called the Bitcoin moderate: someone who holds bitcoin and also wants to do something with it, whether that is lending, a DEX LP position, or some other DeFi vehicle that compounds the stack instead of only sitting on it. He said more than 30 dApps were already loading deployments and assets onto what was supposed to become Hemi mainnet. The unique hook, again, is that those contracts can talk to Bitcoin itself, including the ugly new furniture of the 2024 chain, ordinals and runes included.

I am not a "yield on your bitcoin" evangelist. I am also not going to pretend the demand is fake. People have been wrapping bitcoin and taking it to Ethereum for years because Bitcoin, as a live economy, still does not offer them a place to do those things. Jeff's bet is that you can meet that user without leaving Bitcoin's security behind. The moderate is his name for a person a lot of maximalist content prefers to insult.

Two Spicy Predictions

Near the end I asked him about Greg Maxwell, and about SegWit2x, and I told him historians sometimes talk as if Maxwell was right about everything. Jeff gave what he called two spicy and egotistical answers.

The first was a pointer to a 2015 Medium post, "Bitcoin is being hotwired for settlement." His view is that if you squeeze the block size, activity spills over, the way champagne spills from the top glass of a hotel fountain into the glasses below. Developers spilled to Ethereum in the early 2010s. Transactions spill into fees, and fees spill into Layer 2s. Lightning, judged as a way to onboard not millions but billions of people for normal payments, had the incentives wrong: "I have to pay money in order to receive money." He thinks those predictions have already been borne out.

The second answer was the one that will make people angry.

"Bitcoin would be at 250k and not 100k if we had done SegWit2x. So wherever we are, we would have been farther down the road had the SegWit2x upgrade occurred. But it didn't."

I am not asking you to agree. I am asking you to hear a first-generation developer say, in December 2024, that the cancelled upgrade was not a bullet dodged but a delay. In his framing, the current demand for rollups and for chains like Hemi is what the spilled champagne looks like now.

He still wants Bitcoin to be money for billions of people, not a settlement aquarium for institutions. Self-custody, when we got to the Hodling.ch question, was a matter of hierarchy and redundancy for him: multi-multisig, social recovery, paper backups in geographically distributed safes. Medium of exchange, he said, gets easier if volatility eventually comes down. The through-line is the same. He is not waiting for a perfect base layer. He is building the next chamber of the fountain and hoping some of the overflow is still bitcoin.

If Amir was really listening, the next episode of this season is waiting for him.

Listen to Jeff Garzik on Spotify, Apple Podcasts, YouTube, and More

Time stamps:

00:00:47 – Jeff Garzik’s Contributions 00:02:32 – Overview of Hemi 00:03:11 – Scaling Bitcoin 00:04:17 – Challenges with Current Solutions 00:05:04 – Lightning Network Use Cases 00:05:56 – Goals of Hemi 00:07:19 – Integration with Bitcoin 00:08:20 – History of Open Source Software 00:10:01 – Security and Trust in Established Technologies 00:12:58 – Satoshi's Approach to Building Bitcoin 00:14:11 – Understanding Proof of Proof 00:16:15 – Keystone Element in Hemi 00:17:39 – Hi, Amir Taaki! 00:18:31 – The Need for Privacy 00:20:13 – Shift in Bitcoin's Narrative 00:21:09 – Jeff Garzik's Optimism for Bitcoin 00:23:00 – Building for the Future 00:23:40 – Hemi and SideShift? 00:24:18 – Hemi Token? 00:25:21 – Practical Use Cases for Hemi 00:26:54 – Decentralized Finance and Hemi 00:28:23 – Composability in DeFi 00:30:16 – Decentralization and Security 00:33:03 – Maximalism vs. Moderation 00:35:31 – Limitations of Bitcoin Script 00:40:16 – Drivechains and Sidechains 00:43:17 – Trade-offs of Using Hemi 00:44:51 – Natural Scaling of Blockchain 00:47:05 – Comparison with ZK Rollups 00:47:56 – The Process of Innovation 00:50:15 – Innovation in Scaling Blockchains 00:51:38 – Hemi's Unique Approach 00:52:29 – Self-Custody Risks and Best Practices 00:56:17 – Medium of Exchange and Bitcoin's Volatility 00:58:47 – Was Greg Maxwell Ever Wrong? 01:01:52 – Jeff Garzik’s Favorite Scaling and Privacy Solutions