EigenLayer "scoured the dictionary" to find "Intersubjective" as a replacement for "social consensus"

Author: ZHIXIONG PAN

EigenLayer has just released the whitepaper for its protocol token EIGEN, introducing many novel and complex concepts such as Intersubjective, Work Token, Token Forking, slashing-by-forking, and Intersubjective staking—terms that have quickly become focal points of community discussion.

The original design intent behind the EigenLayer protocol is to generalize the utility of Ethereum nodes—not only enabling these nodes to execute additional business logic to generate more revenue but also increasing income for users participating in ETH consensus. However, expanding node utility inherently introduces risks. Among these risks are not only directly observable objective risks but also uncertain areas lying between subjective and objective realms. This uncertainty cannot be fully guaranteed through cryptography and mathematics but instead relies on "social consensus." This is precisely what is meant by "Intersubjective," a term I prefer to translate as "social consensus."

Building upon this idea, because it hinges on "social consensus," their protocol token must also be forkable. The ability to fork necessitates a dual-token isolation model and supporting mechanisms like challenge processes for initiating forks. It's as if they've implemented an entire network logic using Ethereum—one with (social) consensus, nodes, and even the ability to fork—yet it isn't a blockchain itself.

Additionally, I recall some old anecdotes related to Work Tokens and Weak Subjectivity, which I include here as supplementary context.

What Is a Work Token?

Work Tokens have been discussed since around 2018, gaining wider recognition largely due to Kyle Samani of Multicoin Capital. In short, under the Work Token model, service providers must stake the network’s native token to earn the right to perform work for the network.

That means nodes must provide both tokens and services to receive rewards. Of course, with the rise of on-chain liquidity and DeFi protocols over recent years, the tokens required for staking can now come from third parties rather than requiring nodes to commit large amounts of capital themselves. This evolution leads us to today’s EigenLayer model: users supply ETH, Liquid Restaking Protocols provide liquidity, node operators supply hardware, and AVSs (Actively Validated Services) provide business logic.

Back in 2018 or earlier, the industry experimented with various classifications of tokens—the most common being store-of-value (e.g., Bitcoin), security tokens, utility tokens, and work tokens.

For those interested in Work Tokens and other token models, I recommend starting with this article from Multicoin: https://multicoin.capital/2018/02/13/new-models-utility-tokens/

What Is Intersubjective?

First, let's define two key terms: Objective and Subjective. Errors in blockchain and decentralized networks can be categorized into four types based on their nature:

• Objective errors: These are based on data and cryptography and can be explicitly verified—for example, execution within the Ethereum Virtual Machine (EVM).

• Intersubjective errors: These involve social consensus among participants. Actions or judgments that violate this shared understanding are considered intersubjective errors.

• Unobservable errors: These are known only to victims and cannot be observed by others.

• Subjective errors: These are entirely based on personal experience and perspective, leading to no shared agreement.

EigenLayer argues that unobservable and subjective errors are difficult to correct, hence proposes using ETH to prevent objective errors and EIGEN to mitigate intersubjective errors.

The concept of Intersubjective can be seen as a state lying between "objective" and "subjective." The term combines "Inter" (meaning "between" or "mutual," as in Interactive or Internet) with "Subjective." Thus, Intersubjective refers to a shared subjective state formed through interaction among individuals—a consensus built within a society.

If you ask ChatGPT what Intersubjective means, it might give you a hard-to-grasp Chinese translation: "inter-subjectivity."

For instance, in financial markets, the claim "1 BTC = 1 USD" is widely rejected; such a flawed consensus would qualify as an intersubjective error. Therefore, to explain Intersubjective, we can understand it as "social consensus"—a collective acceptance of certain ideas or facts within a group.

Although there are subtle differences between "social consensus" and Intersubjective in academic and professional discussions—Intersubjective emphasizes the process of sharing subjective experiences and knowledge among individuals, while "social consensus" focuses more on the outcomes of collective decision-making and action.

ETH Objective, EIGEN Subjective?

To briefly recap EigenLayer’s protocol: users deposit ETH into Liquid Restaking Protocols, which then use that ETH to operate Ethereum validators. These validators simultaneously run various middleware services (AVSs), such as oracles, cross-chain bridges, and data availability layers, providing services to end applications.

For AVSs, they can be divided into two categories: objective and intersubjective. Objective AVSs rely on cryptography and mathematics, making them clearly quantifiable and verifiable. Within EigenLayer’s design, these AVSs can depend on restaked ETH for security. Intersubjective AVSs, like oracles, deal with off-chain data that cannot be validated on-chain, so they must rely on social consensus among nodes—the data accepted by a sufficient number of nodes becomes trusted.

In summary, restaked ETH serves as the work token for objective AVSs in the EigenLayer protocol, while EIGEN acts as the work token for intersubjective AVSs.

Can Protocol Tokens Be Forked?

Token Forking is another novel concept. Typically, when people say blockchains are forkable, they refer to open-source code or the network (i.e., the chain itself). ERC-20 tokens, in theory, are not forkable—at least not natively—because tokens, as smart contracts, fully depend on the objective properties of the EVM.

However, EigenLayer argues that within their framework, token forkability serves as a last-resort safety mechanism—even if it occurs once in a century. If malicious actors exceed a majority within the EigenLayer network, ordinary users could fork the token, allowing all users and AVSs to choose which version aligns with their interests. Essentially, this lets social consensus determine which token is the legitimate one. They call this concept slashing-by-forking, a term originating from a blog post written nine years ago by Vitalik Buterin.

Supporting this forkability requires extensive additional logic. For example, if tokens are forkable, can EIGEN still serve as collateral in lending protocols? To address this, they’ve designed a dual-token isolation model: EIGEN itself cannot be forked, but another token, bEIGEN, can. They’ve also introduced challenge mechanisms for initiating forks and compensation logic.

Ethereum's Weak Subjectivity

Interestingly, Ethereum’s PoS consensus already incorporates a concept called "weak subjectivity"—again coined by Vitalik Buterin—which also exists in the space between "objective" and "subjective." Moreover, only PoS-based blockchains exhibit weak subjectivity.

For PoW networks, mining power competition represents real-world costs, so the longest chain is considered the most secure—essentially fully "objective." In contrast, for PoS networks, block production costs are negligible and attack costs low. Therefore, a node newly joining the network must consult external social information to establish this "weak subjectivity" before securely participating in the PoS consensus process. Hence, there is a degree of "subjectivity" involved before joining the correct network.

However, for nodes already correctly participating in Ethereum’s consensus, all consensus procedures and EVM operations are objective—guaranteed by cryptography and mathematics. Examples include deterministic EVM input/output behavior and clear rules for slashing double-signing events.