Our thoughts

Brave new provable world

The future is exciting, as long as we ensure we’re headed in the right direction. In today’s post, based on Sylve’s presentation Brave new provable world at StarkNetCC 2023, we’ll talk about what we’re excited about: zero-knowledge proofs.

Zero-knowledge proofs: where we’re at

So far, we’ve seen zero-knowledge proofs as the perfect enhancement for blockchain, allowing it to go faster.

Blockchains have two main advantages:

  • Trustlessness: when something works onchain, it generally means it’s irrevocably true
  • Decentralization: while different conditions may apply depending on blockchain, anyone can join the network and contribute their computing power

To paraphrase Louis Guthmann, what zero-knowledge proof rollups bring to the table is the ability to not think about decentralization as a source of trustlessness.

We’re finally starting to have a rudimentary understanding of zero-knowledge proofs. They give us trustlessness without decentralization, something completely new for people coming from the blockchain world, as we’re used to both of these qualities only existing together.

When do you care about censorship resistance?

Currently, in gaming, we do everything onchain. Fight a rabbit? It’s onchain. Get loot? It’s onchain. The computation entirely happens onchain. Does it need to, though?

You can play a game locally and submit a proof once in a while, and that will be enough. You will only have one transaction, with only one transaction’s cost and computational needs.

For gaming, you probably don’t care about censorship resistance at every part of the stack: what you care about is trustlessness.

We need to decouple zero-knowledge proofs and blockchains

Why do we even want to put things onchain? For trustlessness, we have zero-knowledge proofs.

Blockchain is also known as crypto − and it has taken over cryptography, the real crypto, for more than a decade. We encourage you to go back to the original crypto and start thinking outside of the blockchain framework.

Zero-knowledge proofs are starting to have the same problem. They don’t need blockchain to work. But because we see them as a way to achieve privacy and scalability onchain, we forget how great they are by themselves.

Zero-knowledge proofs give you so much more, though, because they don’t need to be decentralized at all. This gives you much more oversight on what’s happening.

Zero-knowledge proofs also allow for centrally executed code: you can easily understand how running on one local machine, instead of 5000 nodes all over the planet, is cheaper and faster than executing onchain.

If you take zero-knowledge proof as the core, and blockchain as an extra layer for security, then you change the entire paradigm. And if you want your results to be censorship-resistant, then you can drop them onchain.

Using zero-knowledge proofs for machine learning

We love Giza’s work on verifiable artificial intelligence through zkML (zero-knowledge machine learning) technology.

They’re using blockchain because there’s a point to it − but they’re not taking on the mammoth task of doing everything onchain. They’re only putting proof that a specific dataset and algorithm were used.

Neural networks are costly to train, and if we were to put everything onchain, not many nodes would be able to replicate and trust the execution. Whereas if the proof is available, then you can just trust the proof without redoing all the computation, especially not onchain.

Zero-knowledge also works for Web2

There could be a world where you can use a peer-to-peer sharing network and prove, in a fully reliable way, that you’re seeding and not just leeching.

And that example is particularly important for the way it infuses trustlessness in normal web2 environments.

We can achieve great things by infusing proofs in existing systems, and it’s much easier than building new web3 alternatives.

It doesn’t require any kind of structural change like putting your entire system onchain, it doesn’t require previous investments to gain some weight in the blockchain ecosystem. 

Taking crypto back to crypto

The crypto in cryptocurrency comes from the ethos of cypherpunks, the folks who actually built so many of the foundations for today’s systems.

People need to care

Back in the 1980s, they were really adamant that people should care about trustlessness and decentralization. To them, it was incredibly important:

  • To have these tools of decentralization and trustlessness, such as having everyone use their own PGP key;
  • To care about them and understand why privacy is important.

Today, we can reasonably say that the tools exist. But people still don’t care − and in the end, maybe in the 2020s, we can feel like that’s okay. If we build these trustless affordances and leverage them without people needing to care, then that might be the best way to go.

If the emergence of the world of crypto-commerce creates systems that are vastly less corruptible, but under a whole mix of different mechanisms and governance regimes such that they’re not always everywhere incorruptible, I think that’s fine.
Mark Miller, quoted in Bitcoin or the End of History

To Mark Miller, and to me, and to Hylé, it’s better to have everyone do a little bit of trustlessness than a handful of people who are really passionate about decentralization and go hardcore.

It’s ok if the system isn’t perfectly secure. 

Raising levels incrementally

There’s a lot of power in providing people tools such that they can successfully start to act more like you would like them to. As opposed to: “You’ve got to come over here where it’s really hard edge encrypted and it works exactly the way we want.” No no, let’s raise people’s levels incrementally, and that’s an improvement in the world.
Dean Tribble, quoted in Bitcoin or the End of History

We can « raise people’s levels incrementally », as Dean Tribble puts it.

We can start by putting zero-knowledge proofs everywhere. On their own, they’re already really great and useful. And when zero-knowledge proofs are a given for everyone’s experience, then maybe we’ll be able to make them understand the point of blockchain. But we don’t need to start with that kind of heavy lifting and to require resources from them until we get there.