If you want to change the world, don't protest. Write code!
Prof. Eli Ben Sasson joined (finally!) the Freedom of speech protocol.
Censorship resistance for the win.
Nostr is and will be a critical piece of the Integrity Web of tomorrow. So it’s a natural fit for someone who pioneered and contributed so much to the realms of verifiable computation to embrace it and support it!
Loading quoted post…
7/30/2026, 4 engagements
Ok I will come back and post more often on Nostr. It’s a shame I don’t manage to maintain consistency here. Now there is a real chance with all the buzz on Buzz to make Nostr getting finally the real traction it deserves. Freedom of speech protocol. And much more actually. The future of agentic commerce might happen on Nostr. Decentralized marketplace of services and billions of agents exchanging services over Nostr DVMs. Let’s go
7/30/2026, 5 engagements
I trained a GPT across 4 workers using Nostr as the communication layer.
No central server. No coordinator. Workers exchange compressed pseudo-gradients as signed Nostr events through public relays.
This is "nostrain", distributed ML training over Nostr.
How it works:
This is DiLoCo, the same algorithm Google used to train LLMs across poorly-connected datacenters.
Except the "datacenter interconnect" here are censorphip resistant Nostr WebSocket relays.
The repo ships a char-level GPT demo (834K params, 4 layers, 4 heads).
4 workers each get a different slice of Shakespeare (inspired by Karpathy nanoGPT).
No worker sees the full text.
Text evolution over 5 rounds:
Round 0: ROMEO:2NYPp@JTb<;2..qce[vP Round 2: ROMEO: Maf sthat pine ted mes I chat Round 5: ROMEO: d I shes mear to the ce withat, tre so ther wisho
Loss: 4.5 → 2.4. The workers that saw completely different text are converging to the same model through the relay.
Why Nostr?
Because Open Source / Decentralized / Sovereign AI must win.
Public relay infrastructure already exists. WebSocket pub/sub at scale, with cryptographic identity built in. Every gradient event is Schnorr-signed and verifiable by any subscriber.
You don't deploy servers. You don't manage infrastructure.
You just point workers at relays that already exist.
The relay doesn't know or care that it's carrying ML gradients.
And it is censorship resistant.
3/27/2026, 18 engagements
Do meta prompting more. Less complex setup and orchestration. Rely more on self reflect capacity of the models. That’s it.
1/11/2026, 3 engagements
fgu >>>>> ngu freedom tech will save the world
12/12/2025, 6 engagements
Time to shake the balance of power in the world honestly
12/8/2025, 2 engagements
Global majority ftw
zero knowledge proofs are truly another 0 to 1. it's a complete paradigm shift. a new way to verify integrity of computation. on bitcoin and other blockchains the way to verify integrity of the network is by re-executing everything. you run your full node (you should! it's the only way to have true sovereignty), and you validate everything, the consensus rules are respect, execution is correct, etc. but of course this process is hard to scale, because of the tension between making this verification accessible to anyone (and hence to low hardware and devices), and increasing throughput / computation power. zkp breaks this dichotomy. you can preserve fully the ability to verify cheaply, efficiently, even on small devices, without even re-executing blocks. it feels magical, but this is really what it brings. you, as an individual, even with a raspberry pi, can verify fully the actions performed by big machines, even super computers, even if they have 1000000x more compute power than you. we are yet to see the implications of such a new paradigm really. it started with blockchains, for scaling and privacy, because right now this is the only "industry" that is sufficiently paranoid to desire trustlessness, but you can be certain that zero knowledge proofs will shake many more industries (healthcare, military / defense, logistics, ai, you name it). it's a whole new world of possibilities. btw, starks are the endgame of zero knowledge proofs. truly scalable, no trusted setup, quantum resistant, pure mathematical power. those things are not just moon math anymore. they work, in real world. starknet proving system alone has secured over $1.5 trillion (yes trillions!) of cumulative trading volume over 5 years running in production, has processed over a billion transactions. it's the most battle tested zkp system in the world. zkps will change the world, first because they will turn bitcoin into truly unstoppable private and scalable p2p cash for the planet. and remember, if you fix the money, you fix the world (i truly believe in this!). but then they will also change many more industries. prove the world, one stark at a time!
12/3/2025, 4 engagements
Ok I will be back more often. Fucking network effect of X is hard to fight. Time to do my part for real and bring more folks to the freedom of speech protocol
12/3/2025, 6 engagements
Proposal: OP_STARK_VERIFY - Native STARK Proof Verification in Bitcoin Script.
10/14/2025, 4 engagements
"If I wanted to destroy bitcoin I would just fund infinite developers who are very talented and tell them to make it better."
I think he is not completely wrong, there is a risk. A risk of distraction, a risk of wanting to work on cool stuff because they are exciting, etc. those risks exist. BUT, in my opinion, what Michael Saylor actually thinks is closer to this:
“If I wanted to destroy Bitcoin I would just scare people so that no one touches it and address the remaining issues and challenges that prevent Bitcoin to truly become P2P cash at the planet scale. It will make it easier for us to capture it and just hammer the NGU and digital gold narratives. Fuck their cypherpunk values and their hope of freedom.”
Who benefit the most from scaring people from ever touching Bitcoin again to try to improve it and fulfill its promise ? I think that considering Bitcoin final now is the biggest gift to governments or any powerful adversarial forces that would want to capture Bitcoin or prevent it from being adopted widely as money, and not just as digital gold.
Complacency also can kill Bitcoin. Job is not finished. The real fight has barely started.
9/18/2025, 3 engagements
NEW WORLD RECORD.
WE CAN PROVE THE WHOLE BITCOIN BLOCK HEADER CHAIN IN 6.5 HOURS.
FOR $13.5.
YES, WE HAVE PROVEN 16 YEARS OF BLOCKCHAIN HISTORY OF A $2 TRILLION ASSET FOR LESS THAN A BIG MAC COMBO MENU.
STWO, FASTEST PROVER IN THE WORLD, BACKBONE OF STARKNET.
9/9/2025, 7 engagements
One day you will pay using fiat currency for the last time without even realizing it.
Bitcoin.
Hear me out, PONS: Proof Of Non-Spam for Bitcoin!
I was discussing with
about various usage of ZKP for Bitcoin, and he came with this intriguing idea to use Zero Knowledge Proofs to fight spam on Bitcoin (without breaking some designs such as BitVM). Initially I was sceptical, but the more I think about it, the more I am convinced that it could be a very interesting idea to explore.
Just to be concrete, here are some examples of valid / non spam materials that could be theoretically proven via a system like PONS:
As usual, I like to experiment and play with ideas, so I started to implement a proof of concept of this idea, for the minimalistic simplest example here that is: prove that a public key is "real" and not a "fake" public key.
You can see in the demo video below the following flow:
Sorry Luke Dashjr I know that you consider Rust as a woke language haha, but for now we have the STWO prover and verifier only implemented in Rust, so I used it for the demo. We can easily integrate the STWO verifier in Bitcoin Core or Knots via ffi though (as we already did in another POC).
So of course here it's completely overkill to use a STARK proof for this and you could do it much more efficiently without a ZKP, but it is to show how the architecture could look like end to end.
There are many interesting design questions, like how to make nodes / miners play nicely and prioritise transactions with valid proofs, how to incentivise the generation of proofs, how to make it efficient and not bloated, etc.
I am curious to hear your thoughts on this idea, and if you think it could be a good idea to explore further.
Breaking news coming from El Salvador today, let me introduce you the first Starknet powered Bitcoin ATM in the world! Edgar did the first ever Bitcoin purchase from a K1 ATM where the funding is coming from a Starknet wallet via an atomic swap to Lightning!
7/30/2025, 5 engagements
Spending some time with the great folks at
working on such an important mission: Bitcoin education in El Salvador and all over the world. As I was told here: Mi Primer Bitcoin is a MOVEMENT!
7/22/2025, 1 engagement
Yosh Nostriches, I am in San Salvador this week. Any recommendations for Bitcoiners to meet there or cool places for Bitcoiners in general.
I was chatting with the Alby wallet team about their Lightning MCP server using Nostr Wallet Connect. Forked it to integrate Lightning <> Starknet atomic swaps via Atomiq Labs SDK. In the demo, I fired up Claude Code (also did it with Goose) and asked it to handle an end-to-end swap: "swap 311 sats to STRK". It executed perfectly, pulling from my Primal wallet via NWC connection string and performing the swap.
Fully trustless (HTLC reverse submarine swap).
Exciting times for building Bitcoin apps with AI. 🧡
#Bitcoin #BitcoinAI #ClaudeCode #Goose
7/17/2025, 1 engagement
Bitcoin covenant toolkit: A deep dive into CTV and CSFS
https://hackmd.io/@AbdelStark/bitcoin-covenant-toolkit-ctv-csfs
7/15/2025, 2 engagements
Built a proof of concept of Bitcoin vaults to showcase CTV & CSFS on Mutinynet signet.
Basic vault: Funds locked to a P2TR output with a single leaf CTV script committing to a trigger tx template. Trigger moves to a conditional P2TR: IF branch requires CSV delay + hot key sig; ELSE is CTV to cold address (immediate, no sig). NUMS internal key forces script spends.
Hybrid vault extends to two leaves: One for CTV path (same as basic). Second for CSFS: Verifies a Schnorr sig over a delegation message (e.g., "EMERGENCY_DELEGATION:AMOUNT=...:RECIPIENT=...:EXPIRY=...:VAULT=...") from treasurer key, allowing direct spend to recipient (escape hatch).
Demo flow 1: Cold clawback via CTV. Fund vault P2TR. Broadcast trigger (satisfies CTV from vault). Then broadcast cold tx from trigger (ELSE branch, satisfies CTV to cold). No delay. Works as emergency response if trigger detected (e.g., hot key compromise).
Demo flow 2: Delegation via CSFS. Treasurer signs message authorizing ops team to spend X sats to Y by Z block (though the amount and other dynamic transaction params are not enforced in the demo, would require OP_AMOUNT, might be possible forcing the spender to provide the transaction data in the witness, check the signature of the tx data with CSFS, and then introspect the transaction elements we want like the amount, but not sure the parsing / splitting of the transaction elements would be practical). Spend vault directly via CSFS leaf: Witness includes sig, message hash, treasurer pubkey, script, control. Bypasses CTV/timelock for authorized emergencies.
7/11/2025, 1 engagement
Running Bitcoin ZK Knots, baby!
We've successfully run a POC verifying a STARK proof of Bitcoin's mainnet block #1 inside a modified Bitcoin Knots client.
The proof validates critical consensus rules:
Here’s the technical flow:
The main purpose of this POC was to test the integration of our Rust-based proving stack into the C++ codebase.
The ultimate vision? To improve Initial Block Download (IBD) dramatically by allowing the entire chain's validity to be verified with a single recursive proof.
Once Raito is fully equipped with recursive proving and strong performance, we'll be back to implement a full, real-world IBD with ZK verification enabled. Stay tuned! Follow our StarkWare Bitcoin Github for updates: https://github.com/starkware-bitcoin/
Decentralized mining on Bitcoin is an important area. We've developed a prototype within Stratum v2 that leverages STARK proofs to achieve a more trustless system and better preserve the strategies of individual miners.
Our demo shows the individual miner (JD client) generating a STARK proof for the validity of the coinbase transaction and the total fees of the block template, all without disclosing the list of transactions. The JD server can then verify this STARK proof, confirming the block template's validity and the claimed sum of fees, without gaining any knowledge about the specific transactions.
This also provides an advantage for mining pools from a regulatory perspective due to plausible deniability, which in turn improves the network's censorship resistance. While this is currently a proof of concept, our goal is to extend it to include full transaction validity in the future. We're keen to collaborate with teams working on decentralized mining and will also be exploring this with Datum.
#Bitcoin #DecentralizedMining #StratumV2 #Datum #STARKs #ZeroKnowledgeProof
Hey, any Bitcoiner in Prague tomorrow? I won’t be there but have few colleagues from StarkWare who will be there and would love to meet more Bitcoiners. If you wanna hang out with them and chat about Bitcoin in general and / or zero knowledge proofs let me know.
6/17/2025, 15 engagements
There is actually no better way to fight spam and mitigate shitcoinery on Bitcoin than actually improving it as money and payment system. CTV + CSFS enhance Lighting and other scaling designs, and it improves self custody for the masses. Stop trying playing cat and mouse with the filters and let’s actually make Bitcoin fully usable for payments, a true p2p electronic cash unstoppable system.
#FixTheMoneyFixTheWorld #Bitcoin #CTV #CSFS
Loading quoted post…
6/9/2025, 10 engagements
Posting here for more visibility on Nostr.
CTV + CSFS: a letter to the technical bitcoin community from 43 engineers
Full text at ctv-csfs.com
Let’s accelerate and improve Bitcoin as money and as a tool that people can use to take back their sovereignty. CTV and CSFS can improve significantly Lightning network and have many other benefits.
Study them, discuss them, share your opinions, debate.
Hyperbitcoinization intensifies.
6/9/2025, 18 engagements
Feeling cute, might find hash collisions later. We can do covenants today, we are just negotiating the price now.
#Bitcoin #Covenant
Stateful computation on Bitcoin mainnet without Fraud Proofs has been achieved internally with ColliderVM . It's a toy implementation to demonstrate the concepts. We can already do covenants on Bitcoin, now we are simply negotiating the price 😘
What are the interesting properties that ColliderVM could bring on the table:
The problem: Bitcoin's scripting language is intentionally limited, making stateful computation challenging. In particular:
In short, what you need to achieve stateful computation on Bitcoin is really those 2 properties:
ColliderVM is an attempt to overcome those challenges in a different manner than BitVM. But before going deeper, let's be clear on the current status of ColliderVM: ColliderVM is currently a research project and should not be used in production environments. The protocol is in active development and exploration phase. While the theoretical foundations are promising, it remains unclear whether ColliderVM will prove practical for meaningful real-world use cases. Significant research and development work is still required to determine its viability.
Ok, now that this important disclaimer is done, let's explore more the context of this demo on mainnet.
In short, ColliderVM combines presigned transactions with hash collision puzzles to enable stateful computation without fraud proofs.
So, what did we do for this Mainnet demo ? Real Bitcoin transactions demonstrating a ColliderVM's two-step range check computation: verifying that 100 < x < 200 across separate onchain transactions. The range check computation was split in 2 functions (F1: x > 100, F2: x < 200). The onchain hash function used is blake3 (BitVM implementation).
Here is the transaction flow of the demo:
ColliderVM parameters used for the demo (very low security):
Total transaction fees: 172,126 sats (~$179 at demo time). ~136 kB of total onchain data.
ColliderVM toy code: https://github.com/AbdelStark/collidervm_toy
ColliderVM paper: https://eprint.iacr.org/2025/591
6/3/2025, 2 engagements
Just run a node anon. And yeah you can choose whatever implementation you want. That’s good. Personally I prefer running Core and I don’t consider Knots being a production grade serious alternative on par with Core in terms of open source and software best practices (and there are obvious reasons for this to be the case, and it does not mean it could not change), but the choice is yours and it’s fine if you switch to Knots as a voice of protest. I think client diversity is a healthy debate to have at the end of the day, so maybe the filters drama was a good excuse to put it on the table I don’t know. I think there is a lot of bad faith arguments from some people pushing users to switch to Knots but that’s fine, Bitcoin is resilient and does not care about those drama. But yeah, just run a node, it really matters.
5/26/2025, 25 engagements
Bitcoin decentralised mining: pools should trust the math, not the miner, and miners should gain privacy for their transaction selection which improves censorship resistance of the network. Stay tuned. Stratum V2 + STARK proofs 👀
5/21/2025, 3 engagements
Gm to those running a well maintained software following best practices of open source development.
Filters are cool but have you tried verifying a STARK proof with OP_CAT on a signet. Peak spam at its finest.
5/9/2025, 1 engagement
“To be truly radical is to make hope possible, rather than despair convincing.” - Raymond Williams
Bitcoin is radical hope for the world, seriously.
Let’s fight, TATAKAE!
5/9/2025, 1 engagement
be me: yesterday I found randomly a paper about 2 Circle STARK friendly hash functions working with Mersenne M31 (RPO-M31 and XHash-M31: Efficient Hash Functions for Circle STARKs by Tomer Ashur1 and Sundas Tariq)
then i was: ohhh wow, looks interesting. then: write a rust implementation
then start thinking: wait, maybe it could be a good candidate for a Circle STARK friendly AND Bitcoin Script friendly hash function
then: start doing analysis and cost estimate
then realise that they would NOT be practical to implement in Bitcoin Script...
It's a pity for this but still it's very interesting hash functions, and the fact that they natively work with M31 is super promising.
Maybe it's worth building a Cairo and Solidity implementation and benchmark the costs. They might be more interesting to use than Poseidon for Circle STARKs M31.
Paper: https://eprint.iacr.org/2024/1635.pdf Rust repo: https://github.com/AbdelStark/rpo-xhash-m31
4/25/2025, 2 engagements
Gm from Hong Kong Nostriches.
4/6/2025, 2 engagements
Bitcoiners in Hong Kong, register and come to our event on Monday. We will discuss covenants, Rollups, STARKs, Cashu, Lightning network and more.
https://lu.ma/3q2zemn9?locale=en-GB
#BitcoinEvent #BitcoinConference #BitcoinDevsGathering
What can you do with Monero style ring signatures for Nostr ? You can build an Anonymous Voting platform using Nostr identities to vote. Thanks to the Back's Linkable Spontaneous Anonymous Group, you can detect double voting without leaking identities. Check Ringable demo 👇
A client-side anonymous voting platform using Nostr-compatible ring signatures.
Ringable allows users to create proposals and vote anonymously as part of a predefined group (a "ring" of public keys). It leverages the cryptographic power of bLSAG ring signatures to ensure that while votes are verified as coming from a valid member of the ring, the specific voter's identity remains hidden.
The cryptographic functions are powered by the Nostringer Rust library, compiled to WebAssembly (WASM) for use in the browser.
Live demo: https://ringable.starknetonbitcoin.com Ringable rep: https://github.com/AbdelStark/ringable Nostringer: https://github.com/AbdelStark/nostringer-rs
4/2/2025, 17 engagements
By popular demand (from 1 person), I added a BLSAG (Back's Linkable Spontaneous Anonymous Group) variant in Nostringer, similar to Monery scheme. bLSAG signatures are linkable but still anonymous. Can detect duplicate voting or track usage of a credential.
Loading quoted post…
3/28/2025, 2 engagements
Nostringer in Typescript was fun, but did it again in a serious language this time: nostringer-rs is born, a blazing fast unlinkable ring signature scheme, written in Rust 🦀 Try it with the cli: cargo install nostringer_cli && nostringer_cli demo
https://github.com/AbdelStark/nostringer-rs
Loading quoted post…
3/28/2025, 4 engagements