Vibecoding the Rest of Ethereum's Roadmap

A developer by the name of YQ just finished the rest of the Ethereum roadmap and shipped it as a client. I guess we’re done here! The 5+ year long roadmap, completed in just a few weeks thanks to AI! 😄

“Two weeks ago I made a bet with @VitalikButerin that one person could agentic-code an @ethereum client targeting the 2030+ roadmap. 

So I built ETH2030.

702K lines of Go. 65 roadmap items. Syncs with mainnet.”

Obviously, this will not be going into production, and is really just a Proof of Concept rather than a full-production client, but the PoC really shows that we need to update what we think about the Ethereum roadmap timelines, and the steps it takes to get there. 

Vitalik responded to YQ's effort, saying it was "quite an impressive experiment" while flagging the obvious caveat - that something built in two weeks without finalized EIPs "almost certainly" contains critical bugs and incomplete implementations.

But the core point is about trajectory:

"Six months ago, even this was far outside the realm of possibility."

The presence of the idea of an AI-finished roadmap also shows up in Justin Drake’s Strawmap – a strawman roadmap for the rest of Ethereum, published last week. The strawmap outlines 7 forks through 2029, but explicitly notes that "AI-accelerated R&D could significantly compress schedules.”

Half Speed, Half Security 

The point isn't just that AI can speed up development. Ethereum is the most valuable piece of software on the planet – rushing to its final form isn't what's best. But the same productivity gains can be redirected toward security.

In Vitalik's response to the ETH2030 effort:

The right way to use [AI] is to take half the gains in speed, and half the gains in security: generate more test-cases, formally verify everything, make more multi-implementations of things.

On the security side, I personally am excited about the possibility that bug-free code, long considered an idealistic delusion, will finally become first possible and then a basic expectation. If we care about trustlessness, this is a necessary piece of the puzzle.

No one is suggesting we vibecode our way to the endgame, obviously. But, if AI can compress years of development into weeks, it can do the same for the security apparatus – more test cases, more formal verification, more independent implementations cross-checking each other.

The tools that accelerate building also accelerate hardening.

Smart Contracts Too

Last week, we had Alpin Yukseloglu from Paradigm on the podcast. He discussed their collaboration with OpenAI on EVMbench – a benchmark standard for measuring AI's ability to successfully detect exploits in smart contracts. What happens next, whether the exploit is exploited, or patched, depends on the color of the hat of the person doing the detection.

AI Now Finds 70% of Smart Contract Exploits | Alpin Yukseloglu on Bankless

AI is getting dangerously good at smart contract security. Faster than crypto is ready for.

Bankless

But the assumption is that AI will eventually become ‘perfect’ at detecting exploits, and it will also become ‘perfect’ at preventing them. While it might be a little chaotic while we cross the chasm from here to there, eventually the crypto industry will achieve ‘perfect security’, which will be massively good for the whole ecosystem. 

Staring into the Abyss

This episode with Alpin was truly great, not just because of the education we got about the interaction between AI and smart contracts, but also because of the philosophical discussion that Alpin led, talking about the tension between agency and uncertainty with regards to the AI event horizon. 

As society tries to discuss the AI-driven future, we end up in this place where we are trying to predict a future that is inherently chaotic. Because of AI, the future is highly variant, and trying to reason your way more than a few years into the future becomes “psychosis inducing.” Alpin’s advice is: just don’t do that, and instead get to work. 

My takeaway from this section is that “staring across the AI event horizon is only psychosis-inducing, because the extremely high agency that AI gives humans can be intimidating if you are doing nothing. But, alternatively, if you ‘grab a shovel and start digging’, you won’t get psychosis – you’ll acquire the agency to take on the unknown with your head held high.

Anyway, just a slight detour I found incredibly arming. 

The Big Takeaway

Looking away from the void and instead back to the Ethereum roadmap, I will leave you with this: 

People should be open to the possibility (not certainty! possibility) that the Ethereum roadmap will finish much faster than people expect, at a much higher standard of security than people expect. - Vitalik Buterin.