S16 E47: Max Hillebrand on White Noise & Private Messaging on Nostr
The audio keeps playing while you browse the site.
Read the transcript36,578 words, every line links to the minute it was said
Ask this episode
Max Hillebrand has not had a bank account in six years, and this was his eighth appearance on my show, roughly sixteen hours of conversation across all of them. We recorded on October 3, 2025, and it ran past four hours. He used to work at a bank, he contributed to Wasabi Wallet through zkSNACKs, and these days he builds White Noise, a private messenger that runs on Nostr. I wanted him back because he had a new protocol to talk about, and because I had a pile of complaints about Nostr that I wanted to throw at someone who would catch them.
He caught them. He also announced Marmot for the first time on this pod, which I did not know was coming when I hit record.
Flip a Coin, Get an Identity
I started with the part of Nostr that bothers me. Lose your keys and you lose your followers and your whole identity, permanently. Max lost his first npub and nsec early on, so he has lived the thing I was worried about, and he shrugs at it: "if you lose some nsecs who cares this isn't money this is just data." He likes how small the whole design is. "it's like the first NIP is like two pages printed out it's tiny," he said, and the identity system is about as light as an identity system can get: "Flip a coin get a private key get a public key and hola that's your identity,"
That minimalism is the pitch for Nostr as a censorship tool. "information is easy to spread and hard to stifle," he said, and a protocol built from signed notes passed between relays is hard to switch off for exactly that reason.
I pushed on the key handling, because pasting an nsec into every client you try is how people get burned. Max agreed it is a bad habit: "you don't really want to give every software your private key just like that". His own setup is an old Android phone running Amber, sitting in a drawer at home as a remote signer. As he put it, "you basically have a mini server at home that has your private keys with no other applications installed". For backups he trusts the same thing Bitcoiners trust: "steel backups are awesome and both Bitcoin and Nostr have them".
While we were on the subject of identity I had told him about the afternoon I spent on npub.world and found several versions of myself, one of which was followed by Edward Snowden. Max's reply was that follower counts are fuzzier than they look, since there are "very different ways to calculate something that seems as simple as how many people are following you". Nobody has an authoritative number, which is a little comforting and a little annoying when you are the one being counted.
Why Nostr DMs Were Not Private Enough
Nostr is public by design, so encrypted group chat on top of it needs a new stack. The first attempt at encrypted direct messages had cryptography problems, and NIP-44 fixed a good part of them. Max's summary of the lesson was blunt: "Don't roll your own crypto is the summary of that,". The other problem was keys. In early Nostr messaging, "we're reusing our keys which is a big no-no in Bitcoin". Anyone who has spent time around Wasabi will recognize address reuse in a different costume.
Metadata was the bigger worry. Even with the message body encrypted, an observer on a relay could still map the social graph: "you don't even need to know private key material in order to see who is messaging whom when and how often". If you are building a messenger for people who care about privacy, that leak is the product failing.
This is where Messaging Layer Security comes in. MLS gives you group secret sharing, multi-device support and key rotation, which are all things a group chat needs and a pile of Nostr events does not provide on its own. What Max described is a division of labor in which each protocol does what it is good at. "Nostr gives us message delivery for basically free," he said, and on the authentication side, "every end user is his own authentication service". MLS normally wants some trusted third party to vouch for who is who, and Nostr keypairs remove that need.
The relay's view of all this is thin: "all that the relay sees is that some anonymous group with an unknown number of members has received a message". Anonymous group, unknown membership, and the relay carries it anyway.
Marmot, Announced Live
Max said it was "being announced for the first time live on this pod": Marmot, a middleware layer between Nostr and MLS. Before I heard anything else about the design I had a naming question, and Max had a full branding theory ready. The Bitcoin and Nostr world already has its animals: "We have the honey badger we have the ostrich oh we have a Marmot I guess food is good too right there's NUTs, there's Cashu". His test for a protocol name is a practical one: "your girlfriend will like it which is I think an important part of branding especially for protocols you know is it is it cute shouldn't be too dangerous looking I think."
The technical claim that got me to lean in was about scaling. MLS groups use a tree structure, so adding or removing members does not mean redoing work for every single member. "We get O(log n) scaling," Max said, with "only a couple milliseconds of computation". Then he caught himself mid-tangent with "this is the takeover podcast we can go down rabbit holes", which is fair, since I had asked.
I was skeptical about the MLS standard itself, which is big and has a lot of moving parts for something that is supposed to make private messaging simple. Max did not defend it as perfect. Late in the show he said, "I'm still curious if MLS is actually the right choice for this in the long run". Today's White Noise rests on it, and he is willing to watch whether it holds up.
Standards First, or Satoshi Wrote a Client
One of the threads of the night was how software like this gets adopted at all. Bitcoin's own history is the cautionary tale, and also the model, because "Satoshi just wrote a client and that is the standard now". Max wants a different order of operations for new work: "standard first driven development rather than implementation first development". Write the spec, get other people to review and implement it, and let the code follow.
The tension is that defaults decide everything. "if it's not the default it doesn't matter," he said, and for privacy that creates a hard build problem: "We need to be hardcore private by default and it's really hard to build." I raised Wasabi as the counterexample where privacy is opt-in, and he held his ground, saying "I think we do have opt-in privacy with WabiSabi". He is not claiming opt-in solves the default problem for messaging, only that CoinJoin shipped and people use it.
On how his team works, he used a phrase I liked: "we deliver shippable intermediaries". Each step has to be something a person can use now, rather than a promise that pays off after the full protocol lands. He also cleared up a title that follows him around. "I was never the ceo of Wasabi because Wasabi is just code right it's a GitHub repo", he said. The company was zkSNACKs, and Wasabi is the repository.
Max's earlier visits were Wasabi Wallet's development roadmap and, before that, Bitcoin privacy advocacy and Wasabi Wallet.
Zcash, a Dance, and the Two-Way Peg
At some point I told chat that if a social media goal got hit I would perform a dance for Zcash, and the goal got hit. The "Zcash pump dance" happened on camera, around the one hour twenty-five mark. I will not describe it. Max took the moment to defend where Zcash stands now: "the original version of Zcash was shit compared to what it is now". Early cryptography is fragile for everybody, and he said so plainly: "cryptographic systems break very easily in their early days".
I asked him whether Bitcoin could get privacy on that level. His list of hurdles starts with the bridge. "a two-way peg is a really fucking hard problem," he said, and the aim would be something "cryptodynamically secure a little bit like Bitcoin". He also does not think Bitcoin on its own covers every privacy need: "I don't think that Bitcoin is a good enough answer for that like it's too narrow".
My own view runs in the same direction. I also recorded Shielded CSV with Liam Eagen, Robin Linus and Jonas Nick. On the practical side, when he looks at existing mixing tools he calls some of them "a patch on a shitty system that doesn't really solve it", and he summed up his feelings about cryptographic restrictions on developers with "guys please just let us multiply numbers".
Knots, Core v30 and the Lowest Point of Debate
The mempool filtering fight came up, as it always does now. I run Bitcoin Core v30 and said so, and I called the debate around it "the lowest point of debate". Max stayed out of the tribal part and gave a plain fact from his Wasabi days: "the zkSNACKs backend was running Bitcoin knots for a long time already before this whole drama started".
Where he gets technical is mining. He does not think transaction policy fights are the lever that matters. "that will do much more for the decentralization of Bitcoin mining than I think any transactional policy would," he said, and what he was pointing at was template selection and pool designs such as DATUM, Stratum V2 and Braidpool, where miners build their own blocks and are not just pointing hash power at someone else's template. On the argument that filtering is pointless because determined actors will route around it, he came back with a line about defense: "just because someone might have a space laser somewhere doesn't mean that you should stop defending yourself against the guy with a knife".
He also gave a view on protocol design that I keep coming back to. "if we're talking about how should the protocol be designed then we do have to mess with how people are about to use or not use the system this makes Bitcoin inherently a human affair". Fee markets, filters and defaults are all shaping what people do, and a design that pretends to be neutral about behavior is still making choices.
Cyberspace Money and a Cow
Around the one hour forty mark my cat Jerry jumped on the keyboard, locked the workstation and cranked up the volume at the same time, which is how Max and the chat found out how good my monitor speakers are. A bit earlier, at 1:11, Max had turned his camera off for a stretch.
When we got into money, Max kept returning to the point that none of this is compulsory. "you don't have to use them man like even bank accounts are totally optional both for individuals and for businesses so we have cyberspace money now guys you can use it's it's a thing". On how to react to rules you dislike, his advice was to be careful with confrontation: "if you find someone that doesn't just move on they're idiots anyway". He is also realistic about leaving as an option. "whenever there is a lot of local oppression you always have the option to leave and in many cases that's probably the right decision if you want to optimize for freedom".
He has been living with the consequences of no bank account for six years, and the stories reflect it. He described being "in the belly of the beast", meaning the fiat system, and called Bitcoin "the highest quality fiat asset" in a sense I will not try to paraphrase. His favorite example of how this works in practice is the cow: "I don't walk out with a bag of cash I walk out with a cow". On personal security he said "personal protection should be valued quite highly".
Why would anyone want money at all? Max's answer is psychological: "money is there to decrease the uneasiness that we perceive and we're uneasy because we don't know what food we want to eat tomorrow at what time". He also thinks Bitcoin is good for people who sell software: "Bitcoin is an absolute gift to I guess digital entrepreneurs who have software products." And on how much privacy a person should have, he does not set a cap: "the slider should go to infinity".
Payments tech came up as part of that. He described an approach to recurring payments in which "after setup the sender can be offline", and he mentioned Angor as a protocol in this area. He also sketched a credit-style arrangement that he called a "peer-to-peer credit line without a bank basically". Blossom, the protocol for encrypted media, got a mention as the file storage piece that makes this kind of app workable on Nostr. NoDNS, the standard for embedding DNS records in signed Nostr events, got one too.
Building Fast, Talking Slow
The tooling around all this has changed how fast people can build. Max pointed to shakespeare.diy, an AI tool for building Nostr clients, as a sign of where things are going, and he is happy to say there is no end in sight to the rabbit hole: "the rabbit hole goes deep so I haven't found the end yet".
He is also an efficient consumer of media. "Yeah I never watch live streams," he said, mid-way through a live stream, and he pushed the speed trick: "it's crazy like you actually save like days months of your life just by speeding up your content accumulation so definitely encourage people to try that out." I told him my audience of maybe a thousand weekly downloaders is mostly passive, but that "within these maybe 1000 people who download an episode every week maybe 10 or 20 of them are very interesting". The Bitcoin builder crowd is smaller than the Bitcoin investor crowd, but "the builder spirit the hacker spirit is much more powerful in that community".
Max had a joke about meetups: the best way to organize one is to trick other people into doing the work. His stance on the larger project is more earnest: "we need more freedom tech builders for sure". I told him about a time I translated a Bitcoin book into Romanian for a publisher who paid far less than the job deserved, which is how I learned that some projects only exist because somebody cares more than the budget allows.
Toward the end I asked about the older plumbing again, the stuff that predates all of this, and Max pointed out something that stuck with me about how we communicate in the first place. The protocols we lean on for private conversation, he said, "were actually designed for delivering emails".
I closed by asking if there was one idea he wanted to leave people with. He went back to the simplest version of what he builds, the thing a person types into when they want to reach somebody privately without asking anyone's permission, and said the work is far from done.
Listen to Max Hillebrand on Spotify, Apple Podcasts, YouTube, and More
Time stamps:
- 00:01:07– Max Hillebrand's Background
- 00:02:47– Post-Wasabi Wallet Era and Privacy Challenges
- 00:04:28– Nostr Protocol Overview
- 00:05:58– Early Nostr Experiences and Key Loss
- 00:08:51– Nostr Client Progress and Edge Cases
- 00:09:58– Nostr's Relation to Bitcoin
- 00:13:53– Hardware Wallets for Nostr Keys
- 00:18:06– Nostr Key Backups and Security
- 00:21:53– Nostr Search and Web of Trust
- 00:24:21– Ads: Bitcoin.com News and NoOnes.com
- 00:26:12– White Noise Protocol Introduction
- 00:27:52– NIP 44 and Gift Wraps for Encryption
- 00:30:20– Signal Protocol and MLS Evolution
- 00:32:16– MLS Protocol Details
- 00:33:23– Nostr Privacy Limitations
- 00:35:17– Trusted Relays and Metadata
- 00:37:25– Nostr as Data Delivery Layer
- 00:38:41– Self-Authentication in MLS
- 00:40:02– Group Creation and Key Packages
- 00:42:38– MLS vs. Major League Soccer Joke
- 00:42:53– Diffie-Hellman and Scaling to Millions
- 00:44:24– MLS Key Tree Structure
- 00:46:59– Scalability Benefits of MLS
- 01:18:00– Marmot and Nostr Tools
- 01:25:30– The Zcash Pump Dance
- 03:09:25– Angor Protocol for Recurring Payments
- 03:14:02– Bitcoin Script and Smart Contracts
- 03:15:32– Great Script Restoration Proposal
- 03:21:03– Builder Spirit in Bitcoin vs. Ethereum
- 03:25:19– Bills of Exchange and Peer-to-Peer Credit
- 03:33:09– Historical Finance and Farming
- 03:38:35– Starting Bitcoin Meetups
- 03:41:43– Localizing Bitcoin Content
- 03:44:17– Recommended Podcasts
- 03:47:48– Podcast Benefits and Guest Interactions
- 03:54:39– Audience Quality and Impact
- 03:55:36– White Noise Future and Building Freedom Tech
- 03:58:51– AI Coding and Proof of Concepts
- 04:00:38– Nym Mixnet Integration
- 04:01:54– Closing Remarks and Future Sessions
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
d99ebc05940787851bbd8ff449625c49c04faa8f9de34986759ef666fa7e3f32- Proof
- 2025-10-03_s16_e47_max_hillebrand_on_white_noise_private_messaging_on_nostr.mp3.ots · all episode hashes (proof)
- Status
- Anchored in Bitcoin block 974,312
- Timestamped
- After release. The episode came out on 3 Oct 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 2025-10-03_s16_e47_max_hillebrand_on_white_noise_private_messaging_on_nostr.mp3(on a Mac:shasum -a 256 2025-10-03_s16_e47_max_hillebrand_on_white_noise_private_messaging_on_nostr.mp3). The result must match the SHA-256 above. - Install the OpenTimestamps client (
pip install opentimestamps-client) and runots verify 2025-10-03_s16_e47_max_hillebrand_on_white_noise_private_messaging_on_nostr.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 2025-10-03_s16_e47_max_hillebrand_on_white_noise_private_messaging_on_nostr.mp3.otsshows the block height to look up in any block explorer.
Related
S16 E18: Rockstar Dev & Max Hillebrand on Privacy
Rockstar Dev and Max Hillebrand on CoinJoin after zkSNACKs, Nostr, vibe coding, Drivechains and how to compartmentalize your…
S15 E45: Lucas Ontivero & Turbolay on Wasabi Wallet's Future
Lucas Ontivero & Turbolay on Wasabi after zkSNACKs: community coordinators, WabiSabi, unfuckable CoinJoin, daemon integrations &…
S13 E3: Casey Rodarmor on Ordinals, Inscriptions & Bitcoin Culture
Ordinal inscriptions are all the rage. They obsoleted the concept of NFTs by offering artists a permanent way to display their…
Comments
Comments run on my own server. No trackers, no tracking cookies.