I realized this recently and it really made sense. You know, you are already working with some assumption for a blockchain to work well, right? 2/3rd will be honest, etc.? This assumption is left untouched within the consensus layer and is not reused for any layers above. I am not spilling more beans yet, so let me just leave it at this.

One thing we do at Fermah is that the whole team doubles as DevRel to make customers happy. Ask @zksync .

.@vanishree_xyz on tg is not me.

We typically build for the current world. What are the apps that have high pmf now. Let's optimize our infra for those apps. This is a surefire path to being obsolete soon. Make programmability the core. Take bets on the tooling you provide though. Don't bake in too many decisions into the core.

This is what we realized at Fermah. Modular architectures are good and useful. We usually think about unifying APIs across different elements in the stack. What tends to get missed is understanding the unifying guarantees we want to give to the user. The result? A product that stitched together a bunch of networks and have no idea what kind of guarantees they can offer. UX is a priority. Understanding how graceful the product can be when it breaks is a priority, right below UX.

Why crypto AI hasn't managed to catch up with Global AI, even to a decent extent? I have been thinking about this and talked to quite a few people about this. Before I present my summary, CT, what do you think? What's missing in Crypto AI that makes is less usable than Global AI?

Fermah's Marketplace Fermah is revolutionizing zero-knowledge proof (ZKP) generation by offering a decentralized, efficient, and cost-effective alternative to traditional methods, much like an e-commerce platform simplifies shopping compared to visiting a physical market. Challenges of Traditional ZKP Generation (The "Physical Market" Analogy) When seekers (those needing ZKPs for blockchain applications like zkRollups, zkEVMs, or privacy-focused systems) opt to generate proofs themselves, they face significant hurdles, akin to the inconveniences of physically going to a market: 1. Resource Intensive: 👉Problem: ZKP generation demands substantial computational power, requiring specialized hardware like GPUs or FPGAs. Setting up and managing this infrastructure is complex, costly, and technically challenging for developers. 2. Low Utilization: 👉Problem: When seekers build their own proving infrastructure, it’s typically dedicated to their exclusive use. This leads to underutilization, as the hardware sits idle when not actively generating proofs, resulting in inefficiency and high costs per proof. 3. Poor Incentive Structure: 👉Problem: Centralized proving systems limit access to a small group of seekers, reducing opportunities to incentivize provers (those providing computational power). This lack of competition stifles innovation and keeps costs high. Fermah’s Solution: The E-Commerce" of ZKP Generation Fermah’s Marketplace acts as a decentralized platform that connects seekers with a network of provers, mirroring the convenience of e-commerce. Here’s how Fermah addresses the above challenges: 1. Decentralized Network of Provers: 👉How It Works: Fermah operates as a marketplace where seekers submit proof generation requests, and a distributed network of provers (using GPUs, FPGAs, etc.) compete to fulfill these jobs. This leverages EigenLayer operators for fault tolerance and scalability, ensuring no single point of failure. Benefit: Seekers avoid the need to set up and manage their own costly infrastructure, reducing technical complexity and eliminating the need for physical hardware management. This is like shopping online from home, with no need to travel to a market. 2. Cost and Time Efficiency: 👉How It Works: By aggregating demand from multiple seekers and matching it with a competitive supply of provers, Fermah optimizes resource utilization. Provers can operate at scale, driving down costs through economies of scale and competitive pricing. Benefit: Seekers save significantly on costs compared to maintaining dedicated infrastructure. The marketplace’s dynamic orchestration ensures proofs are generated quickly, reducing latency and saving time, much like having goods delivered swiftly to your doorstep via e-commerce. 3. Improved Incentives and Accessibility: 👉How It Works: Fermah’s open marketplace fosters competition among provers, creating a robust incentive structure. Provers are motivated to offer reliable, cost-effective services to attract more jobs, while seekers gain access to a wide pool of computational resources. Benefit: This decentralized approach enhances accessibility, allowing a broader range of users—from individual developers to large enterprises—to leverage ZKP technology. It’s akin to an e-commerce platform where multiple vendors compete, driving innovation and affordability. In essence, Fermah’s Marketplace makes ZKP generation as user-friendly as shopping on an e-commerce platform, democratizing access to this powerful cryptographic tool for blockchain scalability and privacy. @fermah_xyz @vanishree_rao @fliing_

Excellent @Rowleysghost

Ok, who opened the pandora box?! Another golden nugget: you gotta start thinking at the UX layer and then make lower level decisions. We need ultra-fast trading --> fast consensus Typing on glass should feel effortless -> autocorrect engine, touch correction buffer

Ok, one more thing. Every toggle you remove is like 10x UX. This is exactly how we built our mechanism design. Not publishing it yet ;)