Source: Blockchain Knights

The cryptoasset community is currently obsessed with generative artificial intelligence and the concept of “agents,” which are purportedly powered by cryptoasset “rails.” , and coordinated through smart contracts on the chain.

This is not a good approach for a simple reason: we cannot build "agents" on the random quicksand of large language models (LLMs).

LLM can be useful in creatively generating ideas and content (such as code), however that same creativity can manifest itself through malicious underlying behavior (i.e., deception).

In addition, it has become fashionable to discuss quantum computing (that is, moving towards post-quantum Crypto and "future-oriented Crypto protocols").

Elliptic curve Crypto technology currently implemented does have risks.

However, much of the remaining surface area is at best threatened at polynomial scale, and the application of quantum technology is likely to cause a rising tide to lift all boats.

The key here is this: Real-world quantum computing is still decades away.

While we are seduced by these shiny new things, core design priorities, choices, and trade-offs continue to rust and threaten to become “good enough” when they should be proactive Revisit.

Ten of them are listed below:

1. Social consensus. If there is anything out of place in the Crypto ecosystem, the concept of “social consensus” is the perfect example. Social consensus is how so-called community leaders govern their tribe; it has no place in the Crypto protocol of 2025.

2. On-chain governance. After social consensus, what happened to on-chain governance? Is it too difficult? Do we just give up? However, do we believe we can govern AI agents on-chain?

3. The value that the miner can extract. Is it now acceptable for miners and block proposers to extract revenue by manipulating the priority, exclusion, rearrangement or changes of transactions in blocks?

4. Oracle problem. Has it become conventional wisdom 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”? Isn’t this a slippery slope back to pseudo-centralization?

5. Centralized stable currency. Speaking of centralization, aren’t centralized stablecoins essentially lightweight CBDCs? Since the wheels of Crypto assets are lubricated by (private) centralized stablecoins, why should we double-deal with (private) central banks?

6. There is no so-called settlement layer, and there is no so-called L1 and L2. Any chain (including so-called L1 and alternative L1) can become the L2 of another chain by publishing ledger data and deploying a bridge contract. We have to stop confusing and clean up the terminology.

7. Privacy. NoDo you know when we started to lose the need for privacy protection? Perhaps the concept of “privacy pools” is how the Crypto protocol finally balances regulatory compliance and privacy.

8.Rollup. In fact, Rollups are mini-blockchains. Unfortunately, Rollups are mostly ignored and suffer from a range of issues: from multi-sig rug-pulls to centralized sequencers MEV and CR, and everything in between. We need to completely clean up the terminology and execution semantics around Rollup.

9. Centralized accounting and block construction. As Proof of Stake technology develops, we will have to face more and more mergers. As private order flows dominate, this will increasingly weaken resistance to censorship. This brings us back to the original point: where can we go without permission and trust? Or do we not care as long as the numbers increase?

10. Public welfare funding. Rising numbers raise long-term challenges and issues in financing public goods. Crypto protocols play a unique and meaningful role in funding public goods, and this opportunity remains paramount. We need to remind ourselves that this is a high priority backlog item.