The decision to reduce issuance / implement stake capping should be accompanied with a well-analyzed model for economic security. As is presented now, the current thinking is that we should be okay with 1/4 staking ratio, and presented arguments are [1], [2], [3].
With the current stake distribution these arguments may seem reasonable. However, a severe change in the institutional adoption is happening right now, with the advent of ETFs:
Bitcoin ETFs are rapidily growing, currently custoding 3.7% of BTC [4]
8 out of 11 ETF issuers use Coinbase as BTC custodian [5]
As of Q4’23 Coinbase custodies 17.5m ETH, of which 4.4m ETH is staked. So now Coinbase custodies ~14.5% of all ETH in circulation [6]
ETH ETFs are expected to launch as soon as in May, and there is little doubt they will attract lots of institutional flows. Coinbase will likely service most of ETF issuers.
Given (1-4), it is quite likely that Coinbase’s share of ETH custodied will grow significantly, up to 30m ETH.
The market forces imply that all the ETH in Coinbase Custody, Coinbase Prime and future Coinbase ETF Custody will be long-term staked.
This means that of the future 30m ETH custodied by Coinbase, easily half of that – 15m ETH – will be staked, if not more.
If the staking ratio 1/4 is targeted (~30m ETH), we arrive at the very much possible scenario of Coinbase controlling 51% of staked ETH.
I understand that the proposed curve doesn’t strictly cap the issuance and doesn’t target the 1/4 level. However, it is a move in this direction, and I think the premise of this direction is under-analyzed. In particular, this proposed lowering of issuance has the repercussions discussed by VS above.
Bottom line:
A. I think that the premise of the staking endgame (capping and targeting 1/4 level) is dangerous and may allow 51% attacks in the future, in the view of very strong institutional flows into staking market.
B. Most importantly, partially in the view of A, I think the proposal lacks analysis from all the relevant viewpoints, and requires significantly more scrutiny.
I think it makes most sense to not add this into Electra fork, and wait for more analysis and clarity on these issues. We should push for more external research on issuance change impact on stake distribution. From our end, we have recently launched a grant specifically dedicated to this.
Changing our monetary policy represents a significant hit to credible neutrality.
Why not wait and see
How LST competition continues evolving
Whether or not restaking delivers yield at scale, and
How 7251’s max effective balance change plays out?
I’m not a researcher and have only a welcome appreciation for the depth of research here, but I can’t at this time see the community supporting this change.
ETH holders will always seek yield, adjusting Ethereum’s issuance curve with the intent of decreasing the % of staked ETH will likely push ETH holders further up the risk curve leading to higher restaking concentration on Eigenlayer. This is irresponsible and will lead to new concentration risks. Any DeFi user knows this, LSTs were also obvious to predict, capital efficiency dictates the rules. Changing monetary policy arbitrarily seems very credibly neutral lol
From my experience working with solo stakers and the staking market, I believe that if applied, this proposal will result in the following:
Institutional ETH will end up staked anyway
Staking APR will reduce to the values that are even less appealing for the solo stakers
Even if the ETH price goes up significantly, the entry barrier for the new solo stakers will get even higher
As a result, we will see almost no net new solo stakers
The staking market will be dominated by the vertically integrated parties who can afford to run validators at a loss by funding it from other revenue sources like selling blockspace, offering other services, etc.
The research behind the proposal is incomplete and based on assumptions that do not include incoming institutional capital.
I’m firmly opposed to MVI as a direction, and think the impact of this change would be very negative for the health of the Ethereum network. This would:
As a counterproposal, if the problem to be solved is LSTs becoming the base asset for the network, I think we should fix the primary downside of native staking, by allowing a mechanism for someone to exit their stake at finality for an escalating basefee. There is a large amount of risk-averse delegators that currently hold >32 eth worth of LSTs and do nothing with it, purely so they have the optionality to exit promptly if they so choose. The idea that the exit queue could become months long is unacceptable to this group, and so they hold derivatives.
Modifying EIP7002 to process forced-exits independently from the free queue would make native staking a viable option for this risk averse part of the market. They would no longer be exposed to smart contract upgrade risk, nor the oracle risk of LSPs. This could become the safest way to delegate stake. LSTs would then be primarily needed for fractional stake (<32 eth), and for those that legitimately want to put their collateral to use, e.g. in DeFi. (LSPs can benefit from this feature too for that matter). I think this modification should be adopted for Electra/Pectra as opposed to the 7002 design as currently specified. Happy to work with the original authors on getting this proposal to technically complete to make that happen.
If a target was to be set, the nominal yield at this point (let’s say 1/4, though this number is still very much to be discussed) would be quite different from what it is today, meaning that Coinbase depositors would be faced with a new decision of whether to stake or not, and Coinbase itself would need to decide what to do with custodied ETF ETH, so I am questioning here the assumption “The market forces imply that all the ETH in Coinbase Custody, Coinbase Prime and future Coinbase ETF Custody will be long-term staked”.
A new equilibrium would be found, where the share of Coinbase in the staking set would likely not simply be their custodied amount at stake under the previous configuration with non-targeting issuance curve (so 15m staked, say) divided by the target (30m). I would expect the net effect to be a reduction of Coinbase’s nominal amount of ETH at stake. This demands of course more analysis to understand the magnitude of the effect, but is in my opinion a more reasonable hypothesis than the nominal amount at stake remaining constant.
edit: Reminding myself that this post is actually not about targeting, better discussed there, but the argument also stands for a reduction in yield with this new moderate curve.
I believe that modifying the issuance curve could potentially lead to issues with Ethereum’s trustworthiness and neutrality. Furthermore, if modifications are made, there is no guarantee that it will be a one-time adjustment.
Large corporate staking has a higher chance of proposing blocks, and they are more likely to gain MEV benefits compared to solo stakers. Therefore, I think that solo stakers are the ones who would be disadvantaged and in the event of a reduction in profits, it is anticipated that the solo stakers would be the ones to go extinct.
While there is an urgent need for updates, in order to approach this carefully, I believe that a simple and certain way to limit this would be to temporarily adjust the maximum staking amount dynamically according to the total supply quantity, and to establish a queue for entry stability.
My biggest critique on the insurance curve tweaking analysis is that maybe we should use the intersection of “real total staking yield” with the “hypothetical supply curve” to get the equilibrium rate.
assuming in the long run the market is rational, not acting based on face value.
In this case, it seems it actually will lower the real yield for all stakers (which is fine). Just wanted to point out.
While the current curve is definitely bad and staking ratio is already that big that it should be addressed, decreasing ROI together with issuance (making staking less attractive at all) will hurt solo stakers more than LSTs. Applying alone at the fork it will axe solo-stakers completely which will mean lower security for the protocol.
We need some measures along with this to benefit solo-stakers. It could be:
a. Spread rewards according to the validator ordinal number. If we should have 2.5% reward, it will mean #1 validator receives 4% and #1,000,000 1% (ordinal is not current id, it doesn’t count gaps). This change cannot be applied together with validator indexes reuse. It benefits not only solo-stakers but at least it benefits them more and enough to stay. And it’s deterministic. Until we have a way to change existing validator private key, selling existing validators is not an issue. It is also close to the concept “reward rate is fixed forever at the time of joining”.
b. Every attestation includes voluntary reward decrement with number of validators on this node. Honest nodes count unique validator ids per node and if decrement is lower than the real number of validators, such nodes are banned.
c. anything else benefiting solo-stakers
While I dislike both example measures, at least they directly address solo-stakers’ importance. One curve for all will always benefit LSTs more than solo-stakers whatever it looks like. There was no way found in the real world to tax bigger parties and small subjects with the same rate, while benefitting smaller subjects more.
In regards to institutional stakers, they have a structural advantage in the staking ecosystem due to their ability to spread costs at scale, making even minimal yields economically viable for them. IMO, as long as yields remain greater than zero, institutions will continue to stake, independent of the rewards’ size. This can lead to an increased concentration of staking power among these entities.
Solo stakers, who are critical to Ethereum’s decentralization, face disproportionate impacts from the reduced issuance. The viability of solo staking is crucial for preventing centralization and ensuring the network remains resilient and censorship-resistant. However, with reduced yields, the economic rationale for solo staking weakens, removing the incentives for new solo operators to come online. No matter the outcome (with or without reducing issuance), addressing this issue may require innovative solutions from the social layer, where community support and initiatives can provide alternative incentives for solo stakers. This should be the last resort, as there is no guarantee these initiatives will take hold.
Given the complexity of the staking ecosystem and the potential long-term implications of the reduced issuance proposal, there’s a clear need for further analysis from third-party researchers and economists. These studies should aim to understand the nuanced impacts of the proposal on both the macro and micro levels of Ethereum’s staking dynamics. This would give us more assurances that any adjustments to the issuance policy are made with a full understanding of their potential consequences. Additionally, any changes to Ethereum’s monetary policy should be done in a way that minimizes the impact on its credible neutrality. In my opinion, this means these decisions should be made over years, not months.
I’m glad we’re getting the conversation started now, and that the community has stepped up and shared their thoughts!
Current proposal VS capping stake low (1/4).
One of the main motivations for the current proposal is a conjunction of two statements: “capping stake low is desirable” AND “reverting staking ratio back might be too hard”. This implies that any argument for/against capping stake low is also an argument for/against the current proposal.
Staked ETH concentration.
Claim is that all the ETH stored long-term with exchanges & custodians will be eventually staked, and the reasons are the specifics of their client base and market structure:
Long-term ETH-depositors don’t have much opportunity costs in terms of yield – they are not willing to go onchain, custodian is their safe place.
There will be ETF competition on the basis of zero VS non-zero yield.
Decision to stake or not to stake may actually end being made not by depositors.
Agree that decreasing issuance should decrease custodians’ nominal amount of ETH at stake, considering all other external conditions are the same. However:
The external conditions (ETF flows) position custodians to grow their nominal ETH at stake significantly.
Centralized custodians will have much lower supply curve than dec pools & solo stakers, due to lower costs and different client base. This implies that custodian’s nominal ETH at stake will not be significantly affected by the issuance decrease, at least relative to dec pools & solo stakers.
Combination of 1 & 2 lead to the possibility of one custodian controlling 51% of staked ETH, as was described in my first reply.
The scenario I described is meant to showcase that 3 is a real possibility, which means that we need to analyze it or wait and see how the situation plays out.
Aside: meta framework.
In addition to analyzing specific arguments, such an important & contentious proposal warrants a framework for how to reconcile and account for all the variables at play. I have sketched a map of arguments below. One starting point for a wholistic meta framework is to understand arguments’ relative importance, and focus first on those that matter most.
Thank you for the meta framework, this is a useful starting point. There will be much more to say about individual items, so I am not intending to reply fully to each of them right now. Here are questions currently on my mind, with some notes on potential approaches:
Assumption of “long-term staked participants” given any PoS reward curve design: Model cost structure of various types of participants (solo stakers/operators, node operators, LST holders, ETF depositors) as well as type-specific reward curves (e.g., solo staker PnL, solo operator part of fractional/DVT pool, node operator in LSP, custodial services). Combining cost structure and reward curves, we obtain type-specific supply curves. Move on to understanding the distribution effects (relative share of each type) across classes of issuance curves, as well as aggregate size of the staking set as determined by type-specific supply curves and issuance curves under consideration. Leading question: Is there a “macro effect” to the size of the staking set? (i.e., analyse further folk arguments that more issuance loosens everyone’s constraints and increases the relative share of certain types)
PoS mechanism improvements: How far can we go with changes to slashing weights/parameters of the mechanism? What else can we consider to improve internal market competitiveness, i.e., “micro effects”, “make the staking set the best it can be”? (see e.g., proposals contained in rainbow staking framework, and @OisinKyne’s earlier answer) Link back to “macro”: What are the impacts of such changes on the supply curve of each type of participant? Is there any reason to think these effects apply differentially given the prevailing staking ratio, i.e., is there still a “macro effect”, or are the two effects separable?
ETH derivatives: What are fundamental differences between “ETH on L2-derivatives” and “ETH in PoS-derivatives (LST)”? What are fundamental differences between “Re-staked ETH derivatives” and “Re-staked staking ETH derivatives”? (possible starting point for a typology of derivatives)
This is a bit tangential, but I’d like to address one of your core assumptions.
This is an unpopular opinion in the age of Lido dominance, but I disagree, especially in the long run. Today’s short-term market issues are the result of distortion introduced from immaturity of the existing PoS technology which will be mostly resolved with Dencun and Prague/Electra.
Ultimately, PoS simply needs more technological maturity to enable more dynamic market competition, which obviates the need for trade-offs like this proposal which present serious risks to Ethereum’s inclusive ideals.
To compliment the prior technological arguments from @OisinKyne, here’s an economic argument:
Although they share properties of both, LSTs are more likely to take on the market conditions of commodities rather than money (if staking parameters are ossified – otherwise, the market will continue to be distorted). That is, they will act more as inputs into a complex manufacturing system of financial products built on top, more similar to crude oil or raw ore, than a standard means of exchange. Many LST designs exist, each with different characteristics like risk-return metrics or blended product offerings. These commodities must be produced through an industry which will increasingly have larger barriers to entry due to declining margins, but the even if the cost of node operation continuously declines, that cost will always present a floor for cost of goods sold (LST fees).
Under this environment, competition naturally ensues, with perhaps one or a few dominant players per sub-category of LST but no single actor becoming dominant overall. An analysis of the LST market which doesn’t recognize the unique nature of different systems and their specialized use-cases can easily over-simplify and assume all LSTs are basically the same, but that ignores the reality of differences between osETH and stETH and rETH and frax and mETH and… Not to mention all the new 0-liquidity LST designs popping up with protocols like StakeWise v3. This market is dynamic, with Lido dominance currently problematic but decreasing slowly even as newer LSTs gain traction quickly.
This is a really interesting proposal, but I think we should be cautious about changing the curve before it’s clear what the endgame distribution of staking looks like.
I suspect what we’ll get if we continue down the current path is more LST dominance, but that is inevitable no matter how you adjust the curve. The interesting thing is that LSTs are adding permissionless options to themselves, which preserves:
solo staking viability
greater potential for enforcing client diversity
low-risk LST token holding
In a world in which there is no actual game theoretical reason for solo stakers to exist and minimal reason for client diversity, this seems like a pretty good endgame.
And if it is the how things will continue to pan out, how do these curve adjustments fit in with that world? Are they still necassary?
I think a proposal like this, if implemented, would essentially end solo staking (with the exception of some very large solo stakers, and some altruists willing to lose money for ethos). We have already seen a number of ways where solo stakers are underperforming (see, eg https://timing.pics/). They have more missed proposals, they cannot play timing games effectively, etc. Making staking profits marginal will ensure that the only validators left are those that are most effective at extracting value – this is extremely different from those that are most effective at securing the chain (I would argue there’s a negative correlation).
I wanted to highlight two comments from above that I think really hit the nail on the head:
I have only one thing to say. Solostakers are extremely important for Ethereum. Any proposal that harms them harms Ethereum credibility and makes the network more fragile. Think about a solution how solo staking is being incentivized. At the very least do not harm solo staking incentives compared to what they currently are.
It’s clear that at the equilibrium of ~every ETH being staked, home stakers are further-disadvantaged-than-today relative to holding LSTs (as their hardware costs are not offset by the very low levels of issuance). But this proposal is specifically meant to be a stopgap until a targeting policy is instituted, so this hypothetical equilibrium is beyond the span of time that this proposal aims to address. It seems to me that the correct span of comparison is Ethereum 2-3 years into the future, where this proposal would imply lower real yields.
I believe this is where the concerns from home stakers in this thread come from. They would be seeing their yields compress further in the short/medium term, and rightfully worry that this means less incentive to solo stake. This also means a lower influx of new home stakers (lower yields), and less development of home staking software (worse UX). Staking providers, meanwhile, can continue to scale operations and UX profitably at lower real yields.
(Ultimately, in the equilibrium, it is my belief that Ethereum cannot simultaneously (a) preserve a large base of home stakers (who cannot amortize costs as efficiently as large providers) and (b) reward all stakers equally. This is why I’ve proposed to make staker information legible to the protocol.)
