Vitalik: Keep the chain minimalist and don't overload Ethereum's consensus

The consensus mechanism of the Ethereum network is currently one of the most secure encrypted economic systems. A block is confirmed every 6.4 minutes by 18 million ETH (~$34 billion) worth of validators running multiple different protocol implementations for redundancy. If something goes wrong with the cryptoeconomic consensus, whether due to a bug or a deliberate 51% attack, a large community of thousands of developers and many more users carefully monitors to ensure the chain is properly restored. Once the chain is restored, protocol rules will ensure that attackers can be severely punished.

Over the years, there have been many ideas (often in the thought-experiment stage) of leveraging the Ethereum validator ensemble, and even Ethereum social consensus, for other purposes:

The ultimate oracle machine: Propose a proposal that users can vote for the truth by sending ETH, using the SchellingCoin mechanism: everyone who votes for the majority answer can get a proportional share of all the ETH that voted for the minority answer.

The description goes on to state: "So in principle this is a symmetrical game. What breaks the symmetry is that a) the truth is the natural point of coordination, and more importantly b) people who bet on the truth can, if they lose, threaten to Fork Ethereum.”

Re-staking: A set of techniques used by many protocols (including EigenLayer), based on which Ethereum token holders can simultaneously put their shares as a deposit in another protocol. In some cases, their deposits are also penalized if they violate the rules of other protocols. In other cases, there are no in-protocol incentives and the stake is used for voting only.

L1-driven L2 project recovery: It has been raised many times that if L2 has a bug, L1 can fork to restore it. A recent example is this design, using L1 soft forks to recover from L2 failures.

The purpose of this post is to explain in detail why I believe some of these technologies pose a high degree of systemic risk to the ecosystem and should be blocked and resisted.

These proposals are usually made with good intentions, so the goal is not to focus on individuals or projects, but on technology. The general principle that this article will try to defend is this: **While there are some risks associated with dual-use of ETH pledged by validators, it is basically acceptable, however, trying to "recruit" the Ethereum social consensus to serve your The purpose of the application is not desirable. **

Example of difference between validator reuse (low risk) and social consensus overload (high risk)

Alice created a web3 social network, if you can cryptographically prove that you control the keys of an active Ethereum validator, you automatically get the status of "verified". **low risk. ** *Bob proves that he is wealthy enough to satisfy certain legal requirements by cryptographically proving that he controls the keys of ten active Ethereum validators. **low risk. **
Charlie claims that he has disproved the twin primes conjecture and claims to know the largest p such that p and p+2 are both primes. He changes his pledge withdrawal address to a smart contract where anyone can submit a purported counterexample q > p, along with a SNARK proof that q and q+2 are both prime numbers. If someone makes a valid claim, Bob's validator is forced to quit, and the submitter gets Bob's remaining ETH. **low risk. **
Dogecoin decided to move to Proof of Stake, and in order to increase the size of its security pool, it allows Ethereum token holders to "double pledge" while joining its validator set. In order to do this, Ethereum token holders need to change their staking withdrawal addresses to a smart contract where anyone can submit a proof that they violated Dogecoin's staking rules. If someone submits such a proof, the holder's validator will be forced to withdraw, and their remaining ETH will be used to buy and destroy DOGE. **low risk. **
eCash did the same thing as Dogecoin, but the project leaders went a step further by announcing that if a majority of participating ETH validators collude to censor eCash transactions, they expect the Ethereum community to hard fork to remove these validators. They argue that Ethereum is interested in doing this since these validators have proven to be malicious and unreliable. **high risk. **
Fred created an ETH/USD price oracle that works by allowing Ethereum validators to participate and vote, with no incentives. **low risk. **
George created an ETH/USD price oracle that works by allowing ETH holders to participate and vote. To prevent laziness and potential bribery, they added an incentive where participants who give answers within 1% of the median get 1% of the ETH for participants who answer more than 1% of the median. When asked "What if someone credibly offered to bribe all participants and everyone started submitting wrong answers, and the honest ones got 10 million ETH?" George replied: Then Ethereum would have to Stripping away funds from bad players. **high risk. ** *George conspicuously avoids answering medium to high stakes (since the project may create incentives to attempt such a fork, so it is possible to attempt a fork even without formal incentive)
George replied: "Then the attacker wins, and we will forego using this oracle." Low to medium risk (not exactly "low risk", just because the mechanism does create a large number of Incentivize actors to independently advocate forks to protect their deposits)
Hermione created a successful second floor and claims that because her second floor is the largest, it is inherently the most secure, because if there is a mistake and funds are stolen, the loss will be so great that the community has no choice , users' funds can only be recovered through a fork. **high risk. **

**If you are designing a protocol where even if everything goes down completely, the loss is limited to the validators and users who choose to participate and use your protocol, then it is low risk. On the other hand, if you are intentionally introducing broader Ethereum ecosystem social consensus to solve your problems through a fork or reorganization, then this is high stakes, and I think we should strongly resist all attempts to create that expectation. **

Middle ground situations are those that start out in low-risk categories but give their participants an incentive to slide towards higher-risk categories; SchellingCoin-style techniques, especially mechanisms with significant penalties for departing from the majority, are a prime example.

So what's wrong with stretching the Ethereum consensus?

Let’s say it’s 2025 and a group of people, frustrated with the existing options, decide to create a new ETH/USD price oracle that determines the price by allowing validators to vote hourly. If a validator votes, they are unconditionally rewarded with a portion of the system's fees. But before long, actors start to get lazy: they connect to centralized APIs, and when those APIs come under cyber attack, they either quit or start reporting wrong values. To solve this, they introduced incentives: the oracle also votes on the price from a week ago, and if your vote (real-time or retrospective) differs by more than 1% from the median vote, you will be heavily penalized, Penalties will be given to those who voted "correctly".

Within a year, more than 90% of validators participated. Someone asked: What if Lido, together with several other large stakers, conducts a 51% attack on the vote, forces a false ETH/USD price value, and extracts heavy penalties from all those who do not participate in the attack? At this point, proponents of the oracle machine, already deeply invested in the plan, replied: If this happens, Ethereum will definitely fork and kick the bad guys out.

At first, the program was limited to ETH/USD, which seemed pretty stable. But over time, other indices were added: ETH/EUR, ETH/CNY, and finally the exchange rates of all G20 countries.

However, in 2034, things start to go wrong. Brazil has an unexpectedly serious political crisis that has led to disputed elections. One party controls the capital and 75% of the country, but another party controls some northern areas. The main western media thinks that the northern party is the clear winner because its behavior is legal, while the southern party's behavior is illegal (and they are fascist). However, official sources in India and China, as well as Elon Musk, believe that parties in the South actually control most of the land and that the international community should not try to police the world, but should accept this outcome.

At this point, Brazil already has a CBDC that splits into two forks: BRL-N (in the north) and BRL-S (in the south). 60% of Ethereum stakers offered an exchange rate of ETH/BRL-S when the oracle voted. Most community leaders and businesses have condemned stakers' cowardly submission to fascism and proposed to hard fork the chain to only include "good validators" who offer the ETH/BRL-N rate and reduce the balances of other validators to close to zero. In their social media bubbles, they think they will clearly win. However, once the fork occurs, the power of the BRL-S side is unexpectedly strong. What they expected to be a landslide victory was actually an almost 50-50 community split.

At this point, the two parties are each in two separate universes, each with two chains, and can't actually get back together. Ethereum, a global permissionless platform created in part to escape the influence of states and geopolitics, has been split in two due to unexpected serious internal problems in any of the G20 member countries.

This is a good sci-fi story that could even make a good movie. But what can we actually learn from it?

The "purity" of the blockchain, in the sense that it is a purely mathematical construct that only tries to achieve consensus on purely mathematical things, is a huge advantage. As soon as a blockchain tries to "hook" the outside world, conflicts with the outside world begin to affect the blockchain. If a sufficiently extreme political event occurs, in fact, even a monetary oracle could tear communities apart, given that the above story is basically a parody of what actually happened in every major (>25m population) country over the past decade .

Here are some possible scenarios:

The currency tracked by the oracle (possibly even the U.S. dollar) simply hyperinflate and the market collapses to the point where there is no clear market price at some point.

A contentious split like the one in the above story isn’t hypothetical if Ethereum adds a price oracle to another cryptocurrency: it’s something that’s already happened, including the history of Bitcoin and Ethereum itself.

If strict capital controls are imposed, then which price to report between the two currencies as the legitimate market price will become a political issue.

But more importantly, I think there is a Schelling fence: once a blockchain starts incorporating real-world price indices as a layer protocol feature, it can easily succumb to interpreting more and more real-world information. Introducing a layer of price indices also expands blockchain’s legal attack surface: more than a neutral technology platform, it becomes more clearly a financial instrument.

In addition to the price index, what about other risks?

Any expansion of the "responsibility" of Ethereum consensus increases the cost, complexity, and risk of running validators. Validators are required to pay human attention to, and run and update, additional software to ensure they behave correctly according to other protocols being introduced. Other communities gain the ability to externalize their dispute resolution needs to the Ethereum community. Validators and the Ethereum community as a whole are forced to make more decisions, each of which risks splitting the community. Even without a split, the desire to avoid this pressure creates additional incentives to externalize decision-making to centralized entities through staking pools.

The possibility of splitting is also greatly enhanced by the undesirable too-big-to-fail mechanism. There are so many second-layer and application-layer projects on Ethereum that it is impractical for the Ethereum community consensus to be willing to fork to solve all of them. Therefore, larger projects inevitably have a greater chance of being bailed out. This in turn leads to big projects getting the moat: would you rather put your coin on Arbitrum or Optimism, if something goes wrong, Ethereum will fork to save the situation, or on Taiko because it is smaller ( and non-Western, and thus less socially connected in core developer circles), L1-supported rescues are less likely?

However, bugs are a risk and we need better oracles. So what should we do?

I think the best solution to these problems is situation-specific because the various problems are so different in nature. Some solutions include:

Price oracles: Either decentralized oracles that are not fully cryptoeconomic, or validator voting-based oracles that explicitly promise that their emergency recovery strategy is something other than resorting to L1 consensus for recovery (or both) some combination of those). For example, price oracles can rely on a trust assumption that voting participants are slow to corrupt, so users are given advance warning of attacks and can exit any oracle-dependent system. Such an oracle can intentionally award rewards after a long delay, so that if an instance of the protocol goes out of service (for example, because the oracle fails and the community forks to another version), participants are not rewarded.

A more complex truth oracle about facts more subjective than price: some kind of decentralized court system built on a non-full cryptoeconomic DAO.

Layer2****Protocol:

In the short term, rely on part of the training wheels (this post calls it Phase 1)

In the medium term, rely on multiple proof systems. Trusted hardware such as SGX could be included here; I would strongly discourage systems like SGX as the only guarantee of security, but they could be valuable as part of a 2-of-3 system.

In the long term, it is hoped that complex features such as "EVM verification" will eventually be incorporated into the protocol.

Cross-chain bridges: Similar logic to oracles, but also minimize your reliance on bridges: hold assets on their source chains, and use atomic swap protocols to transfer value between different chains.

Use the Ethereum validator set to secure other chains: One reason the (more secure) Dogecoin approach in the example list above may fall short is that while it does prevent 51% finality reversal attacks, it does not prevent 51% censorship attack. However, if you already rely on validators on Ethereum, one possible direction is to stop trying to manage an independent chain and instead become an efficient validator system anchored to Ethereum. If the chain made such a change, its protection against finality reversal attacks would be as strong as Ethereum's, and it would be able to prevent up to 99% of censorship attacks (compared to 49% previously).

in conclusion

Social consensus in a blockchain community is a fragile thing. Because of the need for upgrades, the existence of bugs, and the fact that 51% attacks are always possible, social consensus is necessary, but because it has such a high risk of causing chain splits, we should use it with caution in mature communities. There is a natural urge to extend the core functionality of the blockchain, which holds the greatest economic weight and largest community watchers, but each such extension makes the core itself more vulnerable.

We should be wary of application-layer projects taking actions that might increase the "scope" of blockchain consensus, unless those actions are validating core Ethereum protocol rules. It's natural for application layer projects to try such strategies, often without realizing the risks, but the results can easily be wildly at odds with the goals of the community as a whole. Such a process, without limiting principles, could result in the blockchain community having more and more "duties" over time, pushing it into a situation where it faces a high risk of splits every year and some sort of de facto bureaucratic authority over the chain. An uncomfortable choice between ultimate control.

**We should keep the chain minimalist, support uses for re-staking that don't look like slippery slopes, expand the role of Ethereum consensus, and help developers find other strategies to achieve their security goals. **

View Original