I’ve been reading through the EIP-8363 discussion and the earlier work on issuance, solo staking, validator economics, Maximum Viable Security, and formal models of the staking market. EIP-8363 itself is here: https://eips.ethereum.org/EIPS/eip-8363
There is already enough serious work in this area that I don’t think the useful thing is to build another generic model showing that different kinds of stakers respond differently to changes in yield. That question has been studied. Some of the work I’m starting from includes Initial Analysis of Stake Distribution (https://ethresear.ch/t/initial-analysis-of-stake-distribution/19014), Maximum Viable Security: A New Framing for Ethereum Issuance (https://ethresear.ch/t/maximum-viable-security-a-new-framing-for-ethereum-issuance/19992), Impact of Consensus Issuance Yield Curve Changes on Competitive Dynamics in the Ethereum Validator Ecosystem (https://ethresear.ch/t/impact-of-consensus-issuance-yield-curve-changes-on-competitive-dynamics-in-the-ethereum-validator-ecosystem/21617), and Key Insights from a Formal Framework of the Ethereum Staking Market (https://ethresear.ch/t/key-insights-from-a-formal-framework-of-the-ethereum-staking-market/22813).
The part I still don’t think is resolved is what happens to who actually controls Ethereum’s validators if EIP-8363 succeeds at reducing the amount of ETH being staked.
Those are not the same question.
Reducing total stake could reduce concentration. It could also produce a smaller validator set in which the operators with the lowest costs, the strongest MEV economics, the stickiest customers, or the greatest ability to absorb lower margins control a larger share of what remains. If the security argument ultimately depends on decentralization and resistance to capture, I think that difference matters enough to test directly.
I’m sympathetic to the framing Konstantin Lomashuk and the other authors put forward in Maximum Viable Security: https://ethresear.ch/t/maximum-viable-security-a-new-framing-for-ethereum-issuance/19992. That framing makes more sense to me than treating issuance minimization as the objective by itself. I think Ethereum’s decentralization, sovereignty, neutrality, and distribution of independent control are higher-order properties worth protecting.
But I also don’t want to begin a research project whose conclusion has already been decided by that preference. If EIP-8363 improves the distribution of validator control, the work should be able to show that. If it makes control more concentrated, it should be able to show that too. And if the evidence is not strong enough to distinguish between those outcomes, that should remain an acceptable conclusion.
The work I’m considering would start by reproducing the closest existing staking-market research rather than pretending this question begins here. From there, I want to update the model around the actual EIP-8363 proposal and the Ethereum staking environment as it exists now. That means using the exact EIP-8363 issuance curve, including its transition period, rather than substituting an older hypothetical reduction in issuance. It also means recalibrating around the post-Pectra staking environment instead of simply carrying forward assumptions from an earlier validator economy.
The part I think matters most is movement between staking modes. A solo staker who stops running validators does not necessarily stop staking ETH; they might delegate it. Someone using an exchange might move into an LST. An institution might choose a professional operator. Capital can leave one form of staking without leaving staking itself, and that movement is exactly where a change in validator composition can happen.
I also don’t think “staking provider” is a precise enough unit anymore. The person or institution that owns the ETH is not necessarily the entity deciding where it gets staked, and the entity deciding where it gets staked is not necessarily the one that can move it once that choice has been made. The staking product, middleware, custodian, node operator, and validator operator can all sit in different places, so if the question is decentralization, I think we have to follow that chain all the way through: who owns the capital, who decides where it goes, who can move it, who operates the validators, who can coordinate those validators, and ultimately how many genuinely independent centers of control remain.
I also want to avoid assuming that every participant immediately reacts to a change in yield. A great deal of stake is probably sticky for reasons that have little to do with maximizing every basis point of return. Custody relationships, regulation, taxes, institutional mandates, technical ability, switching costs, familiarity, and simple inertia all matter.
The existing research already shows how dangerous it is to pretend we know these parameters more precisely than we do. The Formal Framework work is particularly useful here because the authors explicitly discuss the difficulty of identifying a unique calibration from the staking market we observe: https://ethresear.ch/t/key-insights-from-a-formal-framework-of-the-ethereum-staking-market/22813.
So wherever the evidence is weak, I would rather treat costs, MEV income, taxes, provider fees, alternative yields, minimum acceptable returns, and other uncertain inputs as ranges instead of choosing one number and generating an impressive-looking equilibrium from it.
The result I care about is also not another graph of “solo / LST / CEX share.” I want to know whether the resulting validator set becomes more or less concentrated in actual control. That could involve operator concentration, independent operator share, custodial concentration, or other defensible measures. I’m not attached to any particular metric yet. I would rather use fewer measures that actually capture something meaningful than build a dashboard full of proxies and mistake that for understanding.
There is one outcome I think needs to remain completely acceptable from the beginning: we may not have enough information to answer the question.
Some of the strongest existing work already runs into this problem. Multiple sets of assumptions can reproduce something resembling the staking market we observe today. If two equally plausible models can produce opposite answers about EIP-8363, then I don’t think the responsible move is to choose the model whose answer we prefer. The result would instead be to identify which variables Ethereum does not currently understand well enough, how much the conclusion depends on them, and what would actually have to be measured before anyone could make the stronger claim.
So the research question I want to pursue is fairly simple:
If EIP-8363 lowers the staking ratio, does Ethereum actually become less concentrated in control?
Not just whether issuance falls, whether the aggregate staking ratio falls, or whether one category of staker earns less. I want to know who is still controlling the validator set after participants have had time to respond, and whether the structure of that control is meaningfully better or worse than the one we started with.
Before I start building this, I want to make sure I haven’t missed work that already answers it. If anyone knows of an EIP-8363-specific analysis that combines heterogeneous staker behavior, movement between staking methods, current/post-Pectra economics, and the resulting distribution of validator control, please send it my way.
I would much rather begin by reproducing good existing work and pushing it one step further than spend time rediscovering something Ethereum already knows.