S16 E41: Yonatan Sompolinsky on Bitcoin, Kaspa & Proof of Work
The audio keeps playing while you browse the site.
Read the transcript52,832 words, every line links to the minute it was said
Ask this episode
I went into this livestream convinced that Yonatan Sompolinsky was the man who wants to replace Bitcoin, and I opened accordingly, with a pet turtle called Bitcoin on camera and a question about whether he was Satoshi. Yonatan's reply on the second point was to nominate someone else: "For me, I will just add Alex Chepurnoy from Ergo. For me, he's my Satoshi. He wrote 3600 lines of code, the entire Bitcoin code. He wrote it from scratch." For his own part, "The highest piece of code that I wrote was the selfish mining optimization simulation in MATLAB." He also warned me that the turtle is a Bitcoiner who would rug-pull a memecoin project, which I took as a compliment to the turtle.
Yonatan is a mathematician and a Harvard research associate, a co-founder of Kaspa, and a co-author of GHOST, PHANTOM/GHOSTDAG and DAGKnight. My assumption was that anyone with that résumé had spent a decade trying to unseat the original chain. Over seven hours and thirty-five minutes he kept sliding out of that frame. He describes his place in Bitcoin as "passively involved" early on, with nothing contributed "other than handing out interesting thoughts and, you know, challenging the status quo." About Kaspa he says "I truly don't understand why people treat Kaspa as complex or complicated," and "I'd argue 80% are like Bitcoin." Kaspa, as he put it, is "a generalization of Nakamoto consensus."
I pushed on the launch, the trust model and the claim of 100 blocks per second, and some of what came back changed how I read the project. The rest of this is that argument in the order it unfolded, with the detours.
What GHOST Got Wrong
I started with the research, because the research is where the bodies are buried. GHOST was developed with Aviv Zohar, and Yonatan walked me through the moment it broke. The paper had made a claim the authors believed: "And we erroneously claimed that GHOST can scale to any block rate and recover the same security guarantees."
The person who found the hole was Gregory Maxwell. "He immediately showed me the attack," Yonatan said, and Maxwell "didn't even need to think about it." Yonatan says that exchange is part of why the later protocols look the way they do. The core idea he wants people to take from the DAG family is that "The innovation is not in the DAG, rather in the way you order messages or blocks in the DAG." Parallel blocks are the easy part. Deciding what order they count in, and who can be fooled about it, is the research.
Later in the night he gave me the cleanest version of what separates the three designs I had trouble keeping apart: "Bitcoin maximizes the one chain, GHOSTDAG maximizes the K chain, DAGKnight finds the minimal K for which the maximization problem of K reaches 51% or 50%." He was blunt that this is where his own thinking lives now: "That's why when I'm thinking Kaspa, I'm thinking DAGKnight." The pitch is a protocol that removes latency assumptions from the security argument, so the network does not need a hard-coded guess about how slow the internet is.
Crypto Minimalism and the Bitcoin Core Problem
I asked the question I ask every non-maximalist guest: are you a Bitcoin maximalist or not? Yonatan took the label apart. "So for me, the correct term or the more telling term is crypto minimalism rather than Bitcoin maximalism." He has a clear dislike for the other pole too, and offered this as the main statement of his position on DeFi: "The main statement was DeFi is a scam."
His view of Bitcoin Core's culture is less polite. "The extreme culture that Bitcoin Core has radiated isn't very conducive to innovation," he said, and in his telling "the atmosphere is more about celebrating incremental improvements like Taproot." I introduced him, half joking, as "the guy who breaks Bitcoin," because that is the reputation his GHOST-and-selfish-mining work gave him among people who only know him by his citations. He did not argue with the label so much as step around it. "I never made efforts to upgrade Bitcoin," he told me, and added that "It's just delusional to think that anyone would consider a different chain selection protocol."
That does not mean he has stopped thinking in Bitcoin terms. He remarked on how often the tie to Bitcoin shows up even in Kaspa's own history, "Yeah, we call it actually Bitcoin. Say hello. Actually waved. Yeah, we released it, and then he came back. So I can, I guess, the tie with Bitcoin is unavoidable." The turtle had joined the conversation by then.
On bridging, he thinks it would be a mistake to treat Bitcoin and Kaspa as one market with a pipe between them. "The reason is because friction is always the main thing that's defining good money. If there's a bridge between Bitcoin and Kaspa then that bridge incurs friction." The goal he described for Kaspa is the opposite of friction: "to marry these two value propositions and pillars of crypto into one system without any friction."
The Gamenet and the Missing Coins
Kaspa launched in 2021, and I asked about the launch the way a skeptical Bitcoiner would, with the pre-mine accusations in mind. Yonatan's account is that he tried to talk investors out of it. The network went out as a gamenet with random rewards, it halted after two weeks, and the team created a second genesis block. His own review of the thing is harsh: "Compared to Bitcoin's impeccable launch, this is horrendous."
I brought up the conspiracy theories about missing coins, and he did not wave them off. He went at them from the other direction. "So the thought that you need to pull off such a bug in order to cheat is an insult to our intelligence." If anyone had wanted a pre-mine, in his view, there were simpler ways than a halted gamenet and a visible restart.
The part that surprised me was what he said about wanting Kaspa to fail. Describing the stance he took toward early backers, he said "Do everything you can to make it fail. And then to have the people override your allergies to new tokens." The reward curve was a separate design choice with its own reasoning: "I was wary of Grin's flat curve, and there had to be some incentive for early adopters."
When I asked what he would change if he could relaunch, the answer was short and total: "I would do everything differently." He sharpened that into something specific: he would work more spontaneously and launch earlier. It also lines up with a line he gave me about his relationship to the project now: "I'm not the face of it and I'm not the promoter of it."
Ten Seconds and Compounding Transactions
I came prepared to poke at confirmation times. Kaspa currently confirms in about ten to twelve seconds, the target is 100 blocks per second, and I wanted to know what that costs. Yonatan's answer was that there is a ceiling on what extra confirmations buy: "there isn't meaningful security above the one confirmation security." The thing that does hold up is the ordering. In his description of the DAG, "every round, there's no monopoly on sequencing."
At 100 blocks per second the bottleneck moves. "Even without transactions, just the block computation becomes too heavy," he said. He also described pruning as a trade he is aware of, with Maxwell as the imagined critic: "If Gregory saw the pruning protocol, he would say this works, but this creates a more brittle system."
We drifted into wallet behavior, because that is where my own habits show. Yonatan does not use a GUI wallet: "I don't, I use CLI." And on fees and priorities he was blunt about the podcast metaphor I used: "Don't choose low for the podcast. For the podcast you choose priority." His opinion on how wallets should handle outputs is that "Compounding transactions should be the default choice of the wallet."
On the question of money design he took a detour through Yap Island, using the Yap stone money story I have heard more than once, with a version I had not: "He concludes it with saying that every other society's money is very weird to us. Like you walk, you would walk in the Yap Island and you will see their money and you wouldn't understand them." His point was that Kaspa and Bitcoin are both fine as strange forms of money, and the useful fight is over properties, not over which weirdness is more legitimate.
Not Your Keys, and the Protestant Version of Bitcoin
This is where we disagreed, and I did not try to hide it. Yonatan thinks the standard self-custody slogan is exclusionary. "The whole not your coin, not your keys slogan is very elitistic for the tech-savvy." I asked him directly: "Is there anything elitist about telling people to use their own wallet and secure their own seed?" I also tried a generational argument: "So you have to understand responsibility. And ironically, if you explain this to boomers, they're going to be much more receptive and get it as opposed to millennials or zoomers who are so used to forgetting stuff."
His counter was about the structure of trust. "It makes more sense to trust a protocol or a code than people," he said, and what he is advocating for is "a system to embody and encode whatever trust exists." In that framing, a covenant or a vault is a way of writing down who you trust and how much, instead of pretending trust has gone away. He gave me one example I had not thought of, about coercion: with the right covenant, "you can actually prove to your captor that you can't release the coins."
He called the larger shape of this approach "a Protestant version of Bitcoin," and he is not shy about the religious language when it fits. At one point in the stream he said, "we are all atheists with respect to most of the gods that humanity ever believed in. Some of us take it just one god further." We came back to Scripture later, when he talked about the genesis block quote from Ezra 7:18 and the Aramaic word "kaspa" (silver, money) showing up in Cyrus the Great's decree. He signed off one segment with a verse from Jeremiah.
He also has a chess hero. "Garry Kasparov, I'm a child of the hero. The chess player. Yeah, I had the fortune to, the honor to meet him a few years ago in person." That detail explains more of how Yonatan talks about moves and counter-moves than I expected.
vProgs and the Monolithic Experience
The longest technical stretch was about where Kaspa goes next. Yonatan wants a base layer that stays thin and an application layer that does not splinter. His complaint about the usual rollup picture is incentives: "roll-up entities do not have incentives to interop." What he proposes instead, "which we recently called vProgs, or verifiable programs," is a system where "L1 does not execute transactions, L1 does not store the state of contracts." The L1 holds thin commitments, the programs are native mini ZK virtual machines, and developers pick the ZKVM they like (zkEVM, SP1 or RISC Zero). The intended result, in his words, is that "it should be a monolithic experience."
He cares about atomicity, which is the only reason he is willing to say anything positive about DeFi: "Decentralized finance or permissionless finance is just superior due to its atomicity." He has a sharper criticism of the other layers around it. "It's not enough to have just the base layer trustless," he said, which is the kind of sentence that can be read as a shot at Bitcoin and as a shot at Ethereum depending on which one you are rooting for.
He mentioned a funding idea too: letting people "keep their exposure to Kaspa while still funding projects." And in a sideline he said "I want to invent new terminology," which I am pretty sure was a warning.
On MEV he is working with another researcher, Aviv (a postdoc at Yale), and he claims that MEV auctions cannot be done cleanly in leader-based systems. His case for the DAG is that sequencing is shared: "So if you think, if you see this DAG, this DAG has the property that every round, there's no monopoly on sequencing." That is the connection he sees between the consensus work and the application layer he wants to build on top of it.
He would also like DAGKnight to feed something bigger. A phrase he used about the whole system was "World War III resilient," which is a very Yonatan way of saying he wants a chain that keeps working when the internet is bad. DAGKnight came back as a headline topic when I later recorded a second episode with him, S17 E36 on Kaspa covenants, DAGKnight and coordination markets.
ASICs, Difficulty Bombs and the Case for Proof of Work
I asked about mining, because the Kaspa mining story is one of the more interesting ones in the field. He is for ASICs: "I do think ASICs are desirable." On a recent mining hardware development he told me, "I think it's the first invention in proof of work that I've seen in recent years." And he described what he is not doing with the protocol: "We're not in the other extreme of making difficulty bombs inside the protocol every six months like Ethereum did."
He also brought up a protocol-level idea that "introduces into consensus an external clock, an external verifiable clock," which is the kind of thing I could follow in outline and not at the level of detail he gave. His position on proof of work generally is shaped by the fact that there is not much of it left in the field: when I asked early on why Bitcoiners keep drifting toward Kaspa, his answer was "I think because he or she left the scene and there aren't many other proof of work."
He has been in contact with other proof-of-work researchers, and I pressed him on whether Kaspa is isolated from the wider crypto research world. His reply was that he does not think he is not part of it, and that the community could use more exposure to it. The full thought, which he gave late in the stream, was "I think if you take the power of the Kaspa community today and combine it with more, with some more context on crypto, I think that's an unstoppable force." I do not agree that the community needs less of its current self, but I take the point that he meant it as a compliment with a note attached.
Azkaban, Mimblewimble and a Bit of Spite
Mid-stream we went into history I had not heard told by someone who was in it. Yonatan recounted the story of Azkaban, a research effort by Guy Corem and Ethan Heilman to synthesize SPECTRE and Mimblewimble. That led into his views on Mimblewimble's origin, where he voiced skepticism about the anonymity of its creator, Tom Elvis Jedusor, and said that Andrew Poelstra's backing served to kosherize the project for VCs.
He told me he was "slightly pissed by it," meaning a particular development in the field, and then added "And I decided to impress him." That is as much as I will say about it here, because the audio does it better. A related sabotage story, about a design someone tried to undermine, he described as "really a sabotage manipulation."
On the lighter side, he recalled inviting Jameson Lopp to Scaling Bitcoin in Israel, which is a good reminder that the people who argue with each other online sometimes share a room. He also told me about the AI tools he uses on the Rusty Kaspa codebase, mentioning Cognition's Devin and DeepWiki, and said he is optimistic that transparent LLMs could eventually act as autonomous maintainers. On his own role he said "I don't like the burden of being around," and when I raised the idea of him being the face of Kaspa, he said again that he is not.
One viewer question I got wrong. The premise was about the Kaspa relaunch, and I corrected it on air before answering. I also joked that children prefer proof of stake, because they get the point of earning something faster than they get the point of burning electricity.
He took a long skeptical pass at criticism too. On the harsher takes about him and the project he said "I'm neither insulted nor surprised," and about the intellectual style of the design he said "a fancy name for common sense." That was his take on one of the formal terms his own field uses for itself.
Near the end, after seven hours and with the chat thinning out, I told him something I have not said on the show in quite that form: "I signed up for a kind of revolution that never happened." Yonatan answered, "All the power to you, Vlad."
Listen to Yonatan Sompolinsky on Spotify, Apple Podcasts, YouTube, and More
Time stamps:
- 00:01:22– Debunking rumors: Why some think Yonatan is Satoshi Nakamoto
- 00:02:52– Candidates for Satoshi: Charles Hoskinson, Charlie Lee, Zooko, and Alex Chepurnoy
- 00:03:41– Alex Chepurnoy as a Satoshi-like figure
- 00:04:07– Kaspa overview: DAG structure, no orphaned blocks, generalization of Bitcoin
- 00:04:55– Similarities between Kaspa and Bitcoin fundamentals
- 00:06:12– Why Kaspa couldn't be built directly on Bitcoin
- 00:08:05– Kaspa as generalization of Nakamoto consensus
- 00:11:55– Origins of GHOST protocol and early DAG concepts for Bitcoin scaling
- 00:13:16– Academic motivation for GHOST and transitioning to computer science
- 00:13:50– Turtle pet named Bitcoin
- 00:15:22– Increasing block rate in Bitcoin and GHOST protocol
- 00:16:57– Meeting Gregory Maxwell and discovering GHOST flaws
- 00:20:00– Yonatan's views on drivechains and Bitcoin maximalism
- 00:20:36– Defining Bitcoin maximalism: Capital B vs lowercase b
- 00:23:18– Satoshi's support for Namecoin and merged mining
- 00:24:12– Bitcoin culture in 2013-2018: Opposing other functionalities
- 00:26:01– Vitalik's 2014 article on Bitcoin maximalism
- 00:26:13– Andrew Poelstra's opposition to other assets on Bitcoin
- 00:26:38– Bitcoin culture: Distaste for DeFi, criticism of Ethereum as a scam
- 00:28:03– Bitcoin Cash developments: Cash tokens, cash fusion, contracts
- 00:28:39– Rejection of Ethereum in Bitcoin circles
- 00:30:18– Ethereum's successful PoS transition despite critics
- 00:35:04– Ethereum's innovation: From Plasma to ZK rollups, nurturing development
- 00:37:04– Stacks protocol and criticism from Luke Dashjr
- 00:39:02– Bitcoin culture justifying technical limitations
- 00:41:01– Declining Bitcoin adoption as money, rise of altcoins for payments
- 00:43:02– Kaspa's aspirations: Merging sound money with DeFi, beyond just payments
- 00:43:56– Possibility of tokenized Bitcoin on Kaspa
- 00:46:30– Native currency advantage and friction in bridges
- 00:48:49– WBTC on Ethereum scale vs Bitcoin L2s
- 00:53:33– Quotes: Richard Dawkins on atheism, Milton Friedman on Yap Island money
- 00:55:44– Story of Kaspa's messy fair launch in 2021
- 01:14:08– Tech demo of Kaspa wallet experience
- 01:28:45– Kaspa confirmation times & transaction fees
- 01:43:26– GHOST DAG visualizer
- 01:44:10– Mining Kaspa
- 01:55:48– Data pruning in Kaspa, DAG vs MimbleWimble
- 02:01:40– Grin & the fairest launch
- 02:12:21– Zcash scaling & ZKP OP code in Kaspa
- 02:19:50– Jameson Lopp, cold storage & self custody elitism
- 02:35:08– Social recovery
- 02:41:00– Amir Taaki, DarkFi & DAO
- 02:53:10– Nick Szabo's God Protocols
- 03:00:00– Layer twos on Kaspa for DeFi
- 03:13:09– How Kaspa's DeFi will resemble Solana
- 03:24:03– Centralized exchanges vs DeFi
- 03:32:05– The importance of community projects
- 03:37:00– DAG KNIGHT and its resilience
- 03:51:00– DAG KNIGHT tradeoffs
- 03:58:18– Blockchain vs DAG, the bottleneck for Kaspa
- 04:03:00– 100 blocks per second?
- 04:11:43– Question from Quai's Dr. K
- 04:17:03– Doesn't Kaspa require super fast internet?
- 04:23:10– Are ASIC miners desirable?
- 04:33:53– Why Proof of Work matters
- 04:35:55– A short history of Bitcoin mining
- 04:44:00– DAG's sequencing
- 04:49:09– Phantom GHOST DAG
- 04:52:47– Why Kaspa had high inflation initially
- 04:55:10– Selfish mining
- 05:03:00– K Heavy Hash & other community questions
- 06:33:20– Latency settings in DAG KNIGHT for security
- 06:36:52– Aviv Zohar's involvement in Kaspa research
- 06:38:07– World priced in Kaspa after hyperinflation
- 06:39:51– Kaspa's fate intertwined with crypto
- 06:40:29– Kaspa contracts vs Solana, why better for banks
- 06:42:53– Cohesive developer experience in Kaspa like Solana
- 06:45:22– Incorporating ZK design in Kaspa smart contracts
- 06:47:22– Heroes: Garry Kasparov
- 06:48:12– Shift in attitude from academics like Hoskinson, Buterin, Back
- 06:53:07– Adam Back's criticism of Kaspa
- 06:55:57– Michael Jordan and LeBron analogy for Bitcoiners' mindset
- 06:58:02– Can Kaspa flip Bitcoin in market cap
- 07:00:34– Gold and USD market cap comparison
- 07:06:06– Collaboration with Kai team
- 07:10:37– Community improvement: More context on crypto
- 07:13:43– Theoretical maximum TPS for Kaspa
- 07:16:05– Full ZK on L1 improvements
- 07:17:45– Atomic composability and logic zones in Kaspa
- 07:23:12– Sparkle and monolithic UX feel
- 07:26:00– Wrapping up: Beating podcast length record, final thoughts on Bitcoin and Kaspa
- 07:27:31– Why Yonatan called a scammer despite explanations
- 07:32:29– Luke Dashjr's views and disconnect
- 07:33:01– Hope for Bitcoin scaling and revolution
Verify this episode
This is the SHA-256 fingerprint of the MP3 as published here, timestamped in the Bitcoin blockchain with OpenTimestamps. If one byte of the file changes, the fingerprint changes.
- SHA-256
9e502d799c47ad323d0b4f2549ab07da98497b258d34e953818f944971e183a1- Proof
- 2025-08-29_s16_e41_yonatan_sompolinsky_on_bitcoin_kaspa_proof_of_work_v2.mp3.ots · all episode hashes (proof)
- Status
- Anchored in Bitcoin block 974,312
- Timestamped
- After release. The episode came out on 29 Aug 2025, but this proof was made on 3 Nov 2026 and anchored in block 974,312, so it proves this exact file existed by 3 Nov 2026, not on the release date.
How to check it yourself
- Download the MP3 and the proof into the same folder.
- Run
sha256sum 2025-08-29_s16_e41_yonatan_sompolinsky_on_bitcoin_kaspa_proof_of_work_v2.mp3(on a Mac:shasum -a 256 2025-08-29_s16_e41_yonatan_sompolinsky_on_bitcoin_kaspa_proof_of_work_v2.mp3). The result must match the SHA-256 above. - Install the OpenTimestamps client (
pip install opentimestamps-client) and runots verify 2025-08-29_s16_e41_yonatan_sompolinsky_on_bitcoin_kaspa_proof_of_work_v2.mp3.ots. It hashes the MP3 again and checks the proof against your own Bitcoin node, then prints the block and the date the file existed by. Without a node,ots info 2025-08-29_s16_e41_yonatan_sompolinsky_on_bitcoin_kaspa_proof_of_work_v2.mp3.otsshows the block height to look up in any block explorer.
Related
S17 E36: Yonatan Sompolinsky on Kaspa Covenants, DAGKnight & Coordination Markets
Yonatan Sompolinsky came to Europe for a reason he would only describe on record as a secretive meeting of DAGKnight magicians…
S17 E38: Keagan McClelland on Lightning Metrics, Covenants, Consensus Politics & Verifying Bitcoin
There is a certain genre of Bitcoin Twitter argument that never quite resolves into a conversation. People trade metrics, memes…
S16 E10: Eric Voskuil & John Carvalho on Bitcoin
This record-breaking episode of the Bitcoin Takeover podcast, whose running time exceeds 8 hours, features John Carvalho and Eric…
Comments
Comments run on my own server. No trackers, no tracking cookies.