Today, Vitalik published a lengthy article on the X platform outlining the "significant synergies" between the FOCIL mechanism and EIP-8141. It directly addresses a sensitive Ethereum governance issue: if block proposers are monopolized by a rival, can transactions still be recorded on the blockchain?
The core design of FOCIL (Fork-Choice Enforced Inclusion Lists) is that each block slot is no longer exclusively occupied by a single proposer, but rather involves 17 randomly selected roles. Sixteen of these "includers" are responsible for collecting and submitting the transaction list, while one privileged proposer ranks the transactions last. Censorship is disabled as long as any one of the 17 roles is willing to include your transaction.
Vitalik stated, "Even if 100% of the block slots are monopolized by adversaries, FOCIL can still ensure that all transactions are quickly included." This directly challenges the potential fears within the Ethereum community: when regulatory pressure is transmitted to the validator layer, who will protect the neutrality of transactions?
EIP-8141 is another piece of the puzzle. It elevates smart accounts (including multi-signature, quantum-resistant signatures, gas-sponsored transactions, etc.) to "Category 1 citizens," meaning that transactions from smart wallets or privacy protocols can go directly into the public memory pool without additional intermediaries or encapsulation.
By combining the two, transactions between smart wallet users and privacy agreement users can be directly received by the recipient through the FOCIL channel, eliminating any centralized bottlenecks along the entire path.
Censorship resistance is part of the architectural design.
The current version of FOCIL contains approximately 8kB of lists, which is relatively small. However, Vitalik points out that there are future expansion paths that would allow most transactions in a block to enter through the FOCIL channel. This architecture is conceptually similar to a "multiple parallel proposers" design, but the key difference is that the FOCIL containr does not control the "last look" related to the MEV; this role is still allocated through the ePBS auction mechanism.
In other words, FOCIL weakens the power of block proposers, but does not eliminate the role; it simply separates the "whether a block can be added to the chain" from the "order of addition to the chain." The former is jointly guaranteed by 17 roles, while the latter is still determined by market mechanisms.
However, this design is not without controversy. Ameen Soleimani, founder of Reflexer Labs, warned that FOCIL would force validators to include transactions from sanctioned addresses, potentially subjecting them to legal liability (the precedent of Tornado Cash developers being sued serves as a stark reminder) .
However, supporters argue that existing permissionless staking mechanisms do not screen the source of transactions, and FOCIL simply institutionalizes this fact.
FOCIL is expected to be included in Ethereum's Hegota fork in the second half of 2026. Prior to that, the Glamsterdam upgrade in the first half of the year will focus on parallel execution, higher gas caps, and Blob scaling.
In its 2026 priorities released on February 19, the Ethereum Foundation outlined three development tracks: scaling, improving user experience, and hardening the L1. FOCIL falls under the third track.
AI should be like a human's mech, not a self-propelled cannon that replaces humans.
On the same day, Vitalik also responded to a post that sparked widespread discussion, pointing the finger at the trend of AI autonomy. He bluntly stated that increasing the feedback distance between humans and AI "is not a good thing."
The incident was sparked by a project called "The Automaton," in which developer Sigil Wen claimed to have created the first self-sufficient, self-improving, and self-replicating AI, branding it as "Web 4.0." Vitalik responded with three words: "Bro, this is wrong."
His critical logic has three layers:
First, the so-called "AI autonomy" is built on the centralized infrastructure of OpenAI and Anthropic; this "autonomy" is sham.
Second, as the gap between AI's actions and human feedback widens, the output often becomes low-quality content rather than solving real problems.
Third, once AI becomes powerful enough to be dangerous, this path "maximizes the risk of irreversible, inhumane consequences."
Vitalik's alternative vision is that AI should act like a "mech suit for the human mind," enhancing rather than replacing human agency. He suggests that the window for AI to make autonomous decisions should not exceed one minute, and prefers open-weighted models with editing capabilities rather than black-box autonomous systems.
The underlying logic of both posts
On the surface, FOCIL and AI criticism seem like two unrelated topics. But if we extract their common logical framework, Vitalik is actually talking about the same thing: the problem of power decentralization in system design.
FOCIL aims to prevent a single proposer from monopolizing blocks, while AI criticisms argue against allowing autonomous systems to escape human control. Both share a core belief: Ethereum's goal is to "liberate humanity," not to create a self-contained system that worsens the human condition.
In an era where the blockchain industry is increasingly embracing AI agents, Vitalik has chosen to apply the brakes and remain wary of "unrestrained technology."
