The following is a guest post by John deVadoss, co-founder of InterWork Alliancez.
The crypto world is currently obsessed with generative AI, with the concept of “agents” powered by crypto “rails” and coordinated via on-chain smart contracts. This is not a good idea for the simple reason that you cannot build “agents” on the stochastic quicksand that is a large-scale language model (LLM). The LLM has a role to play when it comes to the creative generation of ideas and content (such as code). But the same creativity manifests itself through potentially malicious behavior: deception. Beware of vendors.
It has also become fashionable to discuss quantum computing towards post-quantum cryptography or “future-proof” cryptographic protocols. Elliptic curve cryptography as currently implemented has risks. But much of the remaining surface area is only threatened by polynomial scale at best, and access to quantum will probably end up lifting all boats (e.g. Proof of Work will become much more difficult for everyone). ). But here's the problem. Real-world quantum computing is decades away. Hakuna Matata.
While we are distracted by these shiny new objects, our core design priorities, choices, and trade-offs are accumulating rust and need to be actively revisited. We are in danger of becoming “good enough.” We list 10 of them below.
social agreement. If ever there was an anachronism in the cryptocurrency ecosystem, this concept of “social consensus” exemplifies it. Social consensus is how so-called community leaders run their clans. It does not exist in cryptographic protocols in 2025. On-chain governance. What happened to on-chain governance after gaining social consensus? Is it too difficult? Have we just given up? Still, we believe that AI agents can be managed on-chain. The value that miners can extract. Are miners and block proponents now able to siphon revenue by manipulating how transactions are prioritized, excluded, rearranged, or modified within blocks?The Oracle Problem. Has it become common knowledge that the Oracle problem is an economic problem and no longer a technical problem? Is this collateral damage from the move to proof-of-stake? And isn't this a slippery slope back to pseudo-centralization? Centralized stablecoins. Talking about centralization, aren't centralized stablecoins essentially CBDC-lite? Why ( Will there be a backlash against central banks from both sides? Payment layer. There is no such thing as a payment layer, no such thing as L1 vs. L2. Any chain (including so-called L1s and alternate L1s) can become the L2 of another chain by posting ledger data and deploying bridge contracts. We need to stop confusing ourselves and clean up our terminology. privacy. Somewhere along the way, we lost the cypherpunk spirit and the importance of privacy. Perhaps the concept of privacy pools is how cryptographic protocols ultimately balance regulatory compliance and privacy. Well, this would be a first-class use of zero-knowledge proofs. Rollups. In reality, a rollup done correctly is a mini-blockchain. Unfortunately, they have largely flown under the radar and avoided many problems, from multisig rug pulls to centralized sequencers MEV and CR and everything in between. Rollup terminology and execution semantics need to be cleaned up across the board. Centralized staking and block building. Due to the mandated transition to proof-of-stake, we are experiencing increased consolidation of both staking and block building. As private order flows become more dominant, this creates less and less resistance to censorship. This takes us around a full circle and returns us to where we started. Is it permissionless and trustless? Or do you care if the numbers go up?Public goods funding. As numbers increase, the long-term challenges and problems of funding public goods arise. This opportunity remains at the forefront for crypto protocols to play a unique and meaningful role in funding public goods. You must remember that this is a high priority outstanding item.
