Should Lido on Ethereum be limited to some fixed % of stake?

Trying to understand your position better. If Lido were to achieve, say, 80% share of all staked ETH, is that something you would view as undesirable? Or do you think it’s perfectly fine (even desirable?) given the right governance rules?

If undesirable, then what would be the ‘ideal’ % share, in your mind?

If desirable, what would the governance rules need to look like to fully secure the entire Ethereum network from an attack?

1 Like

I think that it is for the most part fine. It’s like saying “is it ok for weth to be 90% of all wrapped Ethereum?

What matters is the details.

stETH doesn’t represent Ethereum that is staked with one single host - and that’s critical.

I think that if Lido continues to focus on ensuring that there are safety measures on having ETH staked with as many diverse entities as possible, incentivizes governance participation, and perhaps adds some level of fail safes to ensure each validator host can temporarily halt a potentially harmful action short term, then there is no reason we can’t be comfortable with it having the vast majority of staked Ethereum.

In its current implementation, I already feel mostly comfortable with it, but I am sure there are things that can be improved significantly still. Those can and should be explored.

None of those seem to rise to the level of needing to self-stunt growth.

4 Likes

Thanks for clarifying

The way I see it, if there were a lot more node operators, and they were permissionless (with respect to Lido’s governance), I would feel much better about it.

But if it happened before these things, then it would represent an attack vector for someone who buys up 51% of LDO supply, which is pretty accessible to a lot of individuals, hedge funds, governments, etc.

Imagine a hedge fund that shorts ETH and stETH, spends $200M to secure majority vote of the Lido DAO, and then starts rotating in their own node operators. Even if it’s a slow process due to no withdrawals, stETH would still tank and they’d make a profit.

Or is that an impossible scenario (sincerely asking, I just made it up, maybe I’m not thinking through it clearly)?

1 Like

What would be the worst case scenarios an exploit at the smart contract level lead to? (Asking as a non-technical user of stETH, holder of LDO and long ETH)
Can an attacker hold ETH’s staked through Lido hostage?
Could he trigger validators’ exit and steal the ETH?
Could he trigger a mass slashing?

What could an attacker do with 51% of the LDO?

My main concern is avoiding a black swan event in which an exploit or hostile takeover threatens all of the ETH network.

1 Like

I think the main problem with current proposed solutions is that they either harm Lido disproportionately, or are fundamentally unfair (increasing the performance fee while withdrawals are impossible).

Alternative solution: introduce a small deposit fee when Lido’s share is above some threshold - charged in ETH during a deposit.
A linear deposit fee that starts at 0% at a 22% share in total staked share, and ends up at 100% at (obviously impossible) 100% achieves the following:

  • generates a liquid ETH treasury right now. Potentially could be used to burn LDO, or distributed to LDO stakers. Distribution details are beyond the scope of this thread.
  • substantially helps the stETH peg without spending LDO to defend it, because with such a fee it makes sense to buy stETH even slightly above 1.
  • they don’t unfairly change the terms for existing stakers that can’t exit.

I think the second point is strong enough to think about a deposit fee regardless of the stance on the centralization issue - although maybe with less aggressive fee levels.

Why should something be done?

I’m not worried about LIDO stakers being hostile, but the issue is that they aren’t anonymous, leaving them vulnerable to hostile governmental regulations.
The main risk is that perceived vulnerability to such regulations makes their existence more likely - no government wants the humiliation of legislating a law that can’t be realistically enforced. While I think that even with much higher LIDO concentration the worst case scenario is a period of chaos and rapid disappearance of custodial staking - such regulation would greatly harm the whole ecosystem, including the value of ETH and especially LDO.

In the medium term, I think staking pool dominance is a temporary consequence of no withdrawals, and eventually the issue will solve itself. Small stakers with less than 32 ETH are still going to use them in the future - but larger stakers are going to solo stake to not pay any performance fees.

What about competition from exchanges?
I agree that growth of stETH is preferable from dominance of exchange staking. The main advantage of stETH is its value as collateral and high liquidity - which isn’t easy to compete with at this point. Even if it starts to happen, it’s not going to happen overnight - and if they actually start growing too fast all limits can be removed.

What is the safe dominance level?
Outside of specific risks, any single staking pool becoming too big is simply bad marketing for ethereum.
Given that value of LDO strongly depends on the value of ETH, I think regardless of technical/legal arguments, there’s a an optimum point beyond which growth of LIDO eth staking (as a fraction) directly harms LIDO itself.

I think everyone can agree that stETH growing to (virtually impossible) 100% of staked ETH would be very bad and should be avoided.
At 0% the problem of course doesn’t exist.
Therefore, there’s some dominance level between 0% and 100% at which LIDO should start to self limit its share via some mechanism.
Which means that instead of a binary question: should LIDO try to self-limit now, I think there are two separate questions to ask: what should the limiting mechanism be and at what dominance level should it activate.

A linear deposit fee is a very simple mechanism with one parameter: dominance level at which fee starts being positive. This makes it easy to vote on - a weighted average vote for the minimum percentage value.

4 Likes

A few questions to identify.

  1. why should lido restrict itself, while other Cexchanges and holders do not?
  2. Is it safer to restrict lido oneself and then give up the shares to the Cexchange?
  3. what if the interests of the lido holders suffer greatly after the restriction?
    let people make their choice,if they like lido,welcome,if not,find anothers
1 Like

Is CEX safer than LDO?i dont think so,be youself,Lido

1 Like

Adam, this has been a good debate, and I appreciate your perspective. Yet, I want to be clear here that my only specific interest is in helping to prevent a Lido majority/supermajority.

Lido is a great team and an amazing product. But, it seems that ETH holders would rather not exist in the timeline where Lido runs 30%, 50%, 70%+ of stake for years to come.

The ethereum community has spent years investing in maximizing credible neutrality via client diversity, our research community, etc. We think the liquid staking industry deserves a real chance to become an oligopoly instead of a monopoly.

Unfortunately, nobody has a better idea to curtail Lido’s hyper-dominance with certainty other than pleading with the Lido community to altruistically and voluntarily self-limit market share for a temporary period of two years while competitors catch up.

Thanks to Vasiliy and the whole Lido team and community.

6 Likes

first, lido is not only a staking solution but also builds a defi token to release the locked liquidity of eth2.0, which means it’s an independent project in the eth ecosystem like uniswap, aave and others. Limiting market share of lido equals to considering lido as an affiliate to eth — designers want to limit lido’s share to guarantee eth safer, which will reversely prove that eth system’s antifragile is very weak.
Second, to easily change/fix a project’s market share is also a centralized behaviour, targeting only at the symptoms but not the cause. if it happens, others can create lido’s fork, or through some bussiness manners, to create more projects simlar to lido. Will it make sense?
Third, users will choose the safest solution to stake their eth, it’s a normal market behavior. In the past, the choice was CEX, which using revenue and management skills to convince sustomers. Now, we have lido, the kernel of which is tech and team. Instead of directly limiting its shares, maybe we could design some mechanisms to make eth staking follow the law of diminishing marginal utility.

3 Likes

Accumulating > 50% of total possible voting power (which is max supply less locked tokens less tokens that can’t vote (e.g. tokens in the treasury) is not that simple).
a) This amount of tokens isn’t actually circulating, so you can’t just go buy them off the market in one go
b) The capital required to purchase all these tokens is not equal to current price * number of tokens you want to buy. When you purchase large volumes (especially of an illiquid token), price per token shoots up dramatically.

2 Likes

On a high level, I have no problem with a single liquid staking protocol having any share of staking in a protocol as long as the following holds:

  1. it delivers a decent validator set
  2. there’s a possibility to opt out of the protocol for the stakers which is simple and near-free to execute (in other words, there is a free market of solutions)
  3. it’s under the governance of the protocol in question (basically, is onchain and doesn’t have defensive mechanisms against protocol governance takeover)
  4. there are enough safeguards in a protocol that any possible changes to the 3 points above are telegraphed early enough for an opt-out/protocol governance takeover to be possible in a reasonable time
  5. implementation risks of 1-4 are negligible

Even more, I think that this situation is not just okay, but is desirable because I believe it to be the only economic equilibrium that is good for the protocol in question. I think that forced equity of protocols that is propped up by a private, untransparent “informal working group of liquid staking oligopolists” (that does not, to my knowledge, include a single person from the market leader) is not an equilibruim. There will be centralization in staking in some layer of the stack, and it’s in everyone’s best interest to have the biggest entity maximally constrained on-chain (as opposed to constrained by the fickle things like regulations or social pressure). Smart contract-based constraints on an incentive-compatible solution are stronger than hopes and prayers and senseless shaming on Twitter.

Now, of the five points above Lido admittedly crosses only 1 and 3, and crossing 2 is technically impossible at this time. It might be a prudent tactical decision to have some form or limit or anti-incentive on staking, even if there’s no sense in having a strategic decision like that (to reiterate, that’s a personal opinion).

From a practical perspective, at the current moment, Lido is not a risk to Ethereum and is in fact reducing the risk. We contribute a lot to increasing the Nakamoto coefficient of the chain, which we estimate to be quite low at this point, and extremely low when we started out. We get a lot of well-earned flak for having a whitelist of operators instead of a trust-minimized protocol, but the decision to use a whitelist has boons as well drawbacks that y’all have to consider. There’s no technical way for Lido to influence consensus - we don’t run validators. There’s also no possible way Lido can coerce over 20 independent, well-established, reputable, and dependable professional node operators to do anything untoward to the chain they run. At the moment, Lido is not a risk to the beacon chain.

Lido is a risk for its users and integrations (both in terms of implementation and governance risk), but that is a question between Lido, its users, and integrations. The day when informal working groups push Ethereum social consensus to actively govern applications on Ethereum will be the end of Ethereum we know and love.

Lido is a potential risk to the protocol: if the governance goes rogue when withdrawals and, thus, rotating validators is possible, it would be able (if the protocol remains as it is now) to rotate validators to a single malicious entity or a cartel of them; a less realistic attack vector would be to redirect all the new stake to the bad operators. We intend to cross out at least points 2 and 4 above by the time withdrawals are enabled, but of course, that’s an intent, not a given fact. You’d be in your right to not believe it’s definitely going to happen.

At this moment I am undecided if tactical limit/anti-incentive is a good idea or not; leaning it’s not a good idea, but I can see the arguments for Lido being very big but not ossified enough.

I also can see the possibility for Lido to self-limit based on community alignment, even if we do not believe it’s for the best. That said, I do not think there’s been a clear demonstration that the Ethereum community wants Lido to self-limit.

It’s visible now that some part of the Ethereum Foundation is in favor, as well as folks close to Rocketpool and Stakewise. There’s been very little said on the subject by the application and L2 developers, core devs, ETH holders and VCs in the ecosystem, Ethereum node operators, CEXes, and other important parts of the community. E.g. claiming that ETH holders would rather not exist in the timeline where Lido runs a lot of stake is premature. Anecdotical evidence - ETH holders don’t mind staking with Lido - could be interpreted the other way.

16 Likes

Two main thoughts, and the second topic I feel much more strongly about:

(1)

Regarding selt-limiting, my weakly-held opinion is: any solution that requests protocols or products to handicap themselves or make their product worse is not really a solution. Blockchains cannot rely on altruistic and mission-aligned actors in order to maintain a healthy ecosystem.

Ethereum is fragile if it needs people to behave themselves to survive, rather than the incentives at L1 providing healthy coordination conditions. The cultural solution of “Vitalik tweeting to request market share reduction” will not work if the founders/builders transition away from being purpose-aligned to being more mercenary tradfi institutions.

That said, while I believe that a liquid staking solution could safely have a large portion (>50%) of staking market share, I also believe that Lido as an incentives/coordination layer protocol is not there yet and there are a lot of improvements to be made prior to that not being an undesirable reality post-merge. So I could see how limiting growth prior to the ability for users to exit (for example) is not an unreasonable suggestion in the circumstances.

In general, I don’t have a strongly-held opinion on this topic and think both sides have decent arguments.

(2)

There has been suggestions by observers to increase fees on Lido now to decrease market share. I strongly believe this is a mistake, knee-jerk reaction without proper considerations. Increasing fees on existing Lido stakers prior to their ability to exit is not just unprincipled, I believe it is perhaps even fraudulent.

Users staked with Lido with particular terms and agreements: one of those terms was a specific fee structure. Unilaterally changing those conditions on billions of dollars of user-owned funds should not even be under consideration until every single users has the ability to protest the fee by no longer being a user of Lido.

The ability to sell stETH on Curve is not sufficient. It does not allow every user to exit, and it could cause those that do want to exit to take ETH losses due to slippage.

Not only is it harmful to users, to which Lido has a primary responsibility, it is also a bad suggestion!

Since ETH cannot exit Lido prior to tx on eth2, modified fee-structures cannot reduce the amount of ETH in Lido… even at 100% annual fee!

Instead, it can only slow new growth by making it less attractive to future stakers. Hostage-taking increased revenue hurts existing users in order to dissuade new users. I believe this is unacceptable. When it becomes possible for users to unstake, suggestions like this could be entertained (given sufficient notification periods on major changes like fees).

Until then, if Lido wanted to dissuade new stake growth, it should do so in a way that does not betray existing users, for example, by adding a staking “entrance fee” (stake 1 eth, receive 0.95 stETH).

Any suggestions to unilaterally increase fees on all stETH holders should be rejected.

Lido has a responsibility to its users and cannot change the financial terms of the agreement until every user is able to safely exit the agreement in response.

27 Likes

Im going to vote NO. I think that if LIDO self-regulated, it would be much easier for a centralized actor with risks of legal and regulatory pressure, to take control of a majority percentage of the staked ether.

An institutional staking solution like Alluvial, backed by Coinbase, could easily grow rapidly by ignoring the recommendations of not exceeding a percentage of the participation of the staked ether. A solution with economic priorities, without decentralized governance.

Once the regulation of the definition of $ETH is defined in the USA, it will be easy for institutions to deploy capital in $ETH and in institutional staking solutions. That capital can easily exceed the billions LIDO manages.

I believe in LIDO’s way of supporting on-chain development and ETH staking tokens that fuel the DEFI economy, although I have reservations with the leverage we see in AAVE or through INSTADAPP, which shouldn’t be supported; spending time and resources developing technologies, boosting ethereum developers and public goods.

For all that I reaffirm my negative vote, there is much to improve but LIDO can work with the ethereum community, the holders of stETH, the holders of LDO, to do it.

1 Like

Like @cobie my opinion is weakly held, but I lean towards exploring some mechanism for limiting Lido dominance.

As @djrtwo pointed out, Lido dominance is worrisome because of tail-risk scenarios. Not because of the normal operation of Lido validators or normal operations of the Lido DAO (both of whom have been excellent to date, including in the decision to have this discussion publicly).

Ethereum so far has implicitly (and possibly unintentionally) had soft rules around governance. This implies (as an example of tail-risk thinking) certain challenges with over-regulating or other adversarial attack. Having a DAO whose largest voters are legible on-chain able to meaningfully shift the validator composition of the chain might change this.

Again, my opinion is weakly held and that’s largely because no one really knows what these tail-risk scenarios (talking about ‘nation state attack’ is a little tired) are. But I lean towards conservatism.

Disclaimer: I hold both LDO and ETH

4 Likes

I’ll also add that these social processes themselves are how ‘Ethereum governance’ has worked in the past, so I don’t think it’s out of sorts to use them for these decisions (i.e. in opposition to what those who think that ‘purely economic/technical’ mechanisms should govern these sorts of parameter changes).

2 Likes

This is what a major proportion of the roadmap aims to minimize. See scorecard.lido.fi.

1 Like

true, but not the case today!

i should say that i’m overall very pro-lido. this group has been very introspective about its position in the market and effect on the broader industry

1 Like

Any self limiting solution is a bad because it assumes and requires that eventually every other staking provider will do the same.

This is not to mention that it doesn’t make much sense for any protocol limit itself without any benefit. There’s a scenario where the ETH community regulates itself and LIDO has no alternative other than limiting itself at the risk of losing even a bigger market share but it’s not realistically gonna happen.

3 Likes

Completely agree with this, was just keeping it simple. There are also considerations which go the other way, such as voter participation, so you don’t need 51% (30% or 20% might be sufficient). You might also be able to bribe some existing LDO holders and purchase their vote. Not sure how that all nets out though, which is why I was simplifying it to keep it more conceptual.

1 Like

This seems to insert a level of moral absolutism into something which is much more economically practical in nature.

If the fee is raised a little bit such that only a small % of users exit and the peg is still maintained, then how is any user worse off? As long as they can exit at 1 ETH per stETH, they are free to rotate into a different liquid staking pool, or stake the ETH themselves.

If the peg is not able to be maintained, then that’s a different story. But that’s why the fee, if raised, should be inched up very slowly.

1 Like