

✓CS Prof. Security and applied cryptography. Some highlights: Zerocash (zcash, et al. ), Zexe (Aleo, Aztec, etc ), zk-creds/zk-promises(...)




Arguably, threshold sigs should die for blockchains, and we can use a migration to hash sigs/lattice sigs as an option to replace them with executable policies. If you assume zk proofs, which likely need to anyway, you can program more realistic policies at low cost on-chain. At the same time, we know from many of the bridge hacks that simple k of n sigs is not a major roadblock to an attacker.

Today in Peer to Peer, @secparam and I take on policy issues surrounding AI Agents. How they can rebalance power between internet platforms and their users, and how their abuse could be an excuse to impose KYC/AML style controls across the internet. p2p.coincenter.org/p/how-age…


Crypto is speed running the lessons of both finance and cybersecurity. Today’s lesson from the Chainalysis breach: “there’s no such thing as a NOBUS (no one but us) surveillance capability.” Chainalysis is, in effect, the blockchain search engine for banks, police, and spies. It’s used to analyze and annotate what’s on chain. Hard to know what was exposed at this point. Hopefully not much, so this is simply a warning. arstechnica.com/security/202…




Independent price and inflation data from Bloomberg. Would be interesting to make this cryptographically verifiable. Minimally you can commit to data as it's collected. But for prices taken off websites, you can go slightly further and get some evidence it was actually from the site via zk tls or attestations. Thanks @austincampbell finding this.

ZK-Tea: a website were verified experts in SNARKs can anonymously spill the beans on whats wrong with the benchmarks in competing papers. (note, i'm not one of the in the weeds snark experts, I just occasionally hear comments and wish this was more public because I don't always know who to believe.)

So Anthropic has a paper on SNARKs for proving the correctness of LLM inference. The idea of snarks for LLMs is that you can check a response came from the claimed model without the operator exposing the model’s weights. It's Cool work. TL;DR: it’s about 70-95x slower than their chosen baseline, which is very impressive. I’m still reading, but my biggest question is whether that overhead is really ~100x. if I spent a few million tokens’ worth of engineering effort, would the baseline improve more than their SNARK scheme? Of course, at some point this doesn’t matter. The point is cost for the capability. If you want verifiable LLM results, on a non-local model, and can’t use a TEE, is 100x overhead acceptable? Note: they do this with a mixture-of-experts model and prove after the fact, in batches, so the prover only does work for the experts that actually ran. That doesn’t seem to translate to 100 tokens/second for other models. But like I said, I’m still reading. anthropic.com/research/prova…


Zcash cannot become a science fair project or a governance fair project. The point is privacy. This was true in 2018, when I said it at Zcon0 and it's true today. We still have science fair ideas from folks who should know better. And we are still trying to figure out governance, which is fine, but we should keep the purpose in mind and build something resilient. As I said at Zcon0, Zcash cannot become " a science fair project [to] ... do cool crypto for the sake of crypto. That is not the point. The point is to do privacy .... [Zcash] can't become, full lack of a better term, a governance fair project. [Zcash] is not an experiment on how you run a cryptocurrency. It's not a way to test this for other cryptocurrencies. It's not a way to indulge your fantasies about how we are going to govern cryptocurrencies. it's a means to an end. we have to build something that is actually worth governing and useful and has a functioning governance structure but the point is not the governance structure. "

Worth some thought of what would happen to Zcash governance and coin holder voting if a large exchange hack happens. No, it wouldn't be catastrophic. In particular, a single hack only gives a small amount of influence, but one might imagine a world of multiple hacks over a few years, at which point the vote looks way less credible. Even then, Zcash can adapt, add checks and balances, etc. So this is a serious thing to consider, but not necessarily a reason to just reject coin holder voting outright. But running the thought experiment will be useful, since it will tell folks a lot of what they want from a resilient and decentralized governance mechanism. For example, it looks very different if coinholder voting retroactively decides funding prioritization vs if it’s deciding core protocol decisions.