David Tse

Today we release BABE, a new Groth16 proof verification protocol for Bitcoin. It improves the state-of-the-art by three orders of magnitude in setup and storage costs. eprint.iacr.org/2026/065.pdf BABE (BAbylon-BErkeley) is a synthesis of two key ideas: Witness encryption on linear pairing, and Argo MAC, a recently introduced garbling primitive. Witness encryption on linear pairing reduces the complex pairing operations in Groth16 verification to a single scalar multiplication on the BN254 elliptic curve. The single scalar multiplication can further be transformed into a vector homomorphic MAC, which can be efficiently computed by Argo MAC. BABE will be launched as part of Babylon's alpha-testnet for the Trustless Bitcoin Vault in February. We thank: - our Berkeley collaborators @SanjamGarg and Dimitris Kolonelos for teaching us so much about witness encryption - @liameagen and @therealyingtong for sharing their amazing work on Argo MAC - the Babylon engineering team, for turning our theoretical ideas into a real system with demonstrable performance gains Check out the paper and give us feedback!

Ethereum's long-term goal: Everything should be proven, nothing trusted. Every transaction, every operation, formally verified as correct. They're starting with ZK proof systems. The ambition is remarkable even if fully achieving it remains distant.

Many don’t know this. Most DeFi protocols have a backdoor called "upgrade keys" that developers control. They're not truly trustless. Bitcoin has no such concept. No upgrade keys, no developer backdoors, no exceptions. This is why Babylon chose to build on Bitcoin's security model rather than typical smart contracts.

Every time you move to a new research area, you lose everything except your thinking. No reputation. No relationships. No credibility. Just the problem in front of you.

Four years to find product-market fit. Four years to continuously improve it. That's the honest timeline for most successful protocols. Anyone claiming a shorter path is measuring something else.

Ep 3 of Beyond Digital Gold is out! Will the real Bitcoin DeFi Please Stand Up?? Everyone disagrees with what Bitcoin DeFi is. Some say it already exists through wrapped Bitcoin on Ethereum. Others argue it will happen on Bitcoin Layer 2s, while disagreeing on what Bitcoin Layer 2s are. And some believe the only real Bitcoin DeFi must happen directly on Bitcoin itself. In our latest episode in our Beyond Digital Gold documentary series with @StarkWareLtd , we map the entire ecosystem and break down the six competing models trying to unlock Bitcoin’s trillion-dollar liquidity. Watch here 👇🏻 Full episode: youtu.be/cBHh1Fq9GjU?si…

I'm learning cryptography through two methods: 1. Shamelessly asking our collaborator cryptographers "stupid" questions 2. ChatGPT as a 24/7 tutor After Dan Boneh's course, I can't ask him questions daily. But collaborators are captive audiences, and ChatGPT never sleeps.

Hackers perform cost-benefit analysis: How much benefit versus how much effort to find the exploit? Smart contract vulnerabilities offer better ROI than Bitcoin protocol attacks. This economic reality provides an additional security layer beyond cryptographic guarantees.

AI learns from human-written papers. That means it also inherits our bad habits. The most common one in academic writing: when you're not sure about something, you cover it with language. "This is obvious" or "it follows that." No proof. No explanation. AI is doing the same thing. Confident hand-waving at the exact step where rigor matters most. If AI is trained on human shortcuts, how does it learn to do better than us?

Our 10.000x reduction in BitVM proof verification transforms three critical dimensions: 1/ Speed: Engineering cycles accelerate from months to weeks 2/ Cost: Operating expenses drop from $14,000 to $37 per transaction 3/ Sovereignty: What was technically possible but economically impractical becomes viable for everyday use

We gave an AI system our new 60-page protocol paper and asked it to find bugs. It found several. Ones we had independently found ourselves. But it missed the most important one. The kind a professional cryptographer catches immediately: we had used the wrong security notion. Standard security when adaptive security was required. Humans are still winning. For now.

We haven't seen attempted hacks on Babylon yet. One reason: No smart contracts means most hackers' training doesn't apply. Hackers do cost-benefit analysis. Learning entirely new attack vectors for Bitcoin-native protocols changes that calculation significantly.

Cryptographic protocols are like Barcelona's Sagrada Família: never truly finished. We wrote a 60-page paper, but security analysis continuously improves. You refine the attacker model, add new dimensions, expand possible attack vectors. Building for hundreds of years...

Formal analysis is the process of writing a computer program to prove your theorems. The code is the proof. We're running this process on our new protocol. AI assists. A team of cryptographers checks. The combination is more thorough than either alone. Still not infallible. But closer.

Aave is launching V4 as our first vault application. Starting with the largest TVL DeFi protocol across the entire blockchain ecosystem proves the trustless vault model before expanding. Success with Aave makes every other conversation easier.

When I gave my first blockchain talk, a professor stopped me five minutes in. One slide in. Done. He didn't know me. Hadn't seen me in this field before. His assumption was immediate: this person doesn't know what they're talking about. Academia is tribal. Every new field has its gatekeepers. To contribute, you have to prove yourself from zero. No reputation carries over. I've done this multiple times now. It never gets comfortable. But first principles thinking works in every field. That's the trade.

Exciting to have Ledger on board!

The gap between "technically possible" and "practically useful" often spans orders of magnitude. BABE's 10,000x improvement in proof verification closes that gap, transforming Bitcoin sovereignty from an expensive ideal into an economically viable reality.

Linear improvements (10%, 50%, even 2x) rarely change behavior. But 1000x improvements fundamentally alter what's possible. Just the past week we got up to a 10,000x improvement. I can't think of another time in my career where I've witnessed this pace of improvement.

Babylon's motto: "Don't lose people's bitcoin." Google once had "Don't be evil." The problem with aspirational mottos is that removing them creates more suspicion than never having them at all. Babylon is building $5 billion in TVL on a principle we can never walk away from.

Saw a cartoon that said "Zero days since the last BitVM breakthrough." With so many teams claiming innovations, it's hard for people to distinguish real breakthroughs from noise.... Excited for the experts to dig through our BABE paper and confirm it's a real one.