We are saddened to see Presearch winding down this week - our partnership & shared mission will not be forgotten, as their platform brought web3 search into the MASQ browser. Arguably, Presearch was one of the earliest "DePIN" style projects with a real-world use-case & survived through many cycles - choosing a difficult sector of Search to build in web3! With giants to compete with, their vision & mission is pretty darn noble - the core team & OG community could have walked away many times! We hope their team & community still continues to use independent search & tech that is pushing back against BigData corpo-cracy. We wish all the team & community all the best @TeamPresearch #presearch

I’ll give a high-level explanation, and happy to point to more formal resources if you want. Since you’re coming from the Zcash perspective, it’s worth noting that Canton’s privacy and integrity guarantees differ because they address a different problem. Zcash aims to fulfill Satoshi’s vision of an encrypted Bitcoin, and ZKPs are really the only practical way to achieve that combination of censorship resistance, public verifiability, and issuer-less assets. The key insight behind Canton is that in almost all real-world use cases, you don’t need public verification. If you assume that every asset has an issuer (an assumption Zcash can’t make by design), you can achieve strong privacy in much simpler and more efficient ways, without the usability, performance, and integrity-bug blast-radius tradeoffs that come with ZKPs. Credit to the Zcash community for being open about these tradeoffs and constantly improving them. In Canton, the issuer (more precisely, the registrar) can be centralized or decentralized, but you don’t get privacy from that issuer. That’s intentional: my bank should know my balance, my custodian should know my holdings, but the public should not. If someone wanted privacy even from the issuer, that’s where ZKPs would come in. It’s technically straightforward to integrate into Canton, but so far there has been no demand for it. Here’s how Canton achieves privacy. Canton's privacy works on a strict 'need-to-know' basis. Every smart contract explicitly lists the finite set of parties who are entitled to see it. Like Zcash, Canton uses a UTXO-style model rather than accounts, so the ledger state is a collection of contract instances (UTXOs), and each participant only sees the subset they are part of. Once you have selective disclosure at the data model level, consensus becomes much simpler. Canton essentially performs a two-phase commit among all stakeholders of a transaction, where each participant only sees the parts they are authorized to view. To go a bit deeper, Canton has formal definitions of the contract language and abstract ledger model, which make privacy a composable property of the system. Each transaction can contain multiple sub-transactions, and each sub-transaction can have its own visibility set. Parties involved in one part of a transaction can see only what they’re entitled to, while still participating in a larger atomic operation that spans others. So, for example, we can swap cash for an asset, and our banks will only see the cash leg while our custodians only see the asset leg. When we first created the language and ledger model, I still assumed we would use ZKPs for consensus. Because of that, Canton’s smart contract language actually provides an excellent high-level abstraction for a ZKP-based system, if anyone ever wants to build that layer. All of these formal definitions are in the open-source repository. Because of these abstractions, we were able to make Canton’s consensus Byzantine fault-tolerant while relatively simple, at least for a single synchronizer (which orders transactions without seeing their contents). Canton can also scale horizontally across multiple synchronizers, but that’s a separate topic.

This is what NEAR AI private inference looks like in production. An @AskVenice ($VVV) user selects a TEE model, sends a prompt, and receives a hardware-signed attestation report. Six cryptographic checks passed and verified in 343ms.

A first just happened in DeFi history. The first PRIVATE decentralized stablecoin minting on @ethereum just executed. Complete capital freedom: borrow against your assets with zero on-chain trace linking back to you. I talked about this vision weeks ago on a space with @jchaskin22 from @ethereumfndn. Now it's real: etherscan.io/tx/0x10f5ca84e4…

BlackRock CEO Larry Fink says demand for computing power is so large that a new asset class will spring up around it.

The goal of Personal AI: civilization where individual humans, augmented by AI, can do consequential work without being captured by extractive institutions. Freedom to write your prompt and own your data. This is the new battleground. 2034 won’t have to be like 1984.

I think a lot of people are missing the point when it comes to why you would want to use OpenClaw or Hermes. I'm seeing more and more posts talking about using one or the other to do things like generating a daily podcast digest or news summary. IMO, this is kinda silly. You can do this with Claude quickly, easily, and you don't even need the horsepower of Opus to do it. What you want to use something like OpenClaw or Hermes for is when you want to do things where: 1. you want the flexibility of Open Source 2. you want to own/protect your data The reality is though, there are a LOT of people who aren't going to leverage the benefits that come with using Open Source, and also, won't be able to properly secure their data, and they're better off letting a company like Anthropic do it for them. Which leads me to the point I think more people should be making. Everyone does not need to be using OpenClaw or Hermes at this stage in the game. People should be using one or the other if it's a priority for them to have a personal assistant that they can understand and fine-tune at the code level (open source) and can have strict, and direct control over their data (privacy). For me personally, I'm not crazy about giving a company like Anthropic or OpenAI access to my email, medical information, etc. That's private data that I'd rather hold and protect myself. I'm looking at sequencing my genome, and that's super private data, so yeah - I'd like to be in control of this data and not worry about another company getting hacked and my data being exposed. The core of the confusion here I think stems from people feeling pressure to use OpenClaw and Hermes, but not necessarily using them out of a need, but instead out of curiosity. And curiosity is honestly a great reason to use something I'd say. But you just have to set your expectations correctly if this is why you're using something like OpenClaw or Hermes, because you're making some serious tradeoffs. So be curious, tinker, or be a power user and update the code to make it your own, secure your data taking every precaution you would hope any company would, but aren't sure they are. Just don't do things with OpenClaw or Hermes, that you can easily do with Claude, and then complain that OpenClaw or Hermes are too technical or hard to get working...you might just not need them, yet.

After almost twenty years on the platform, EFF is logging off of X. This isn’t a decision we made lightly, but it might be overdue. 🧵(1/5)

After almost twenty years on the platform, EFF is logging off of X. This isn’t a decision we made lightly, but it might be overdue. 🧵(1/5)