Over the years and across multiple discussion channels, we have gathered feedback and encountered several strong concerns about a change of this kind. This post addresses the most common ones up front, so that readers can also understand the reasoning behind our design choices.
Disclaimer: It is split in several parts due to char. limitation per post and has been AI formatted and spelling proofed so please excuse its form which can be off putting. This is meant to signal where this EIP stands on recurring questions and comments.
Individual responses to everyone chiming in the thread are deserved and will come asap.
By all means, I concure with @pintail regarding Anders FAQ as an excellent read.
The post is organised in two parts: objections about the consequences of the change (questions 1–5, as previously asked by @CelticWarrior), and objections about whether we have enough evidence to make it (questions 6–13 collected by hanniabu notably and appearing in some of the early replies here). Bracketed references [1]–[14] point to the bibliography at the end. The topic is issuance is dense and we encourage everyone to give it a deep dive.
Part 1: Objections about consequences
An often shared premise worth challenging: 0 yield at 50%
Four of the five questions in this part assume that under this proposal “yields tend towards zero.” … They don’t. Under the taper, net yield reaches zero only at the 50% saturation ratio and that point is the off-switch, not the destination [1]. The staking market equilibrates where net yield equals the marginal staker’s risk premium, which is strictly positive, and therefore at a staking ratio strictly below 50%. Yields tend towards the market price of staking risk. This will be the the first time this yield is set by supply and demand rather than by a curve’s floor.
An important piece of shared machinery: the EIP’s per-duty burn belongs to the issuance-offset family that Elowsson’s June 2026 offsets post analyses as a safe way to implement low issuance [3]. It preserves the relative reward/penalty balance between validator roles (the offset cancels from any comparison between two actions), which is what protects against minority discouragement attacks (the failure mode that makes naive low-issuance designs dangerous). The mechanism behind the taper is not a new idea; it builds on where the low-issuance research has converged.
The counterfactual to keep in frame throughout: “do nothing” is not a neutral baseline. With the entry queue saturated at maximum churn, stake grows ~1.75M ETH per month toward >70M ETH (>55% of supply) by January 1, 2028 [7]. Elowsson’s FAQ names the threshold at which “do nothing” stops being tenable — around 33.6M ETH staked (2^25) [2]. We passed it.
Each answer below therefore ends with the status-quo comparison, because the honest question is never “taper vs. today” but it is rather “taper vs. that trajectory.”
1. What is the appropriate level of staked economic security for a multitrillion-dollar onchain economy?
The steelman: if Ethereum secures trillions in value, surely security should scale with the value secured. Cutting issuance cuts the stake that provides it (For reference, this was the core framing of Maximum Viable Security [6]).
The answer: security does not scale with TVL, and it is not what stake-weighted intuition suggests.
-
Validity is not at stake. A coalition controlling any amount of stake cannot forge signatures, mint ETH, steal from contracts, or push an invalid block past full nodes. What stake controls is ordering, liveness, and finality [1]. The multitrillion-dollar economy is protected first by validity rules that stake cannot break, and only second by consensus honesty. The relevant question is not “how much stake per dollar of TVL?” but “what does an ordering attack earn, and what does it cost?”
-
The cost side is already vast and researchers have named numbers. An attacker needs to acquire and then forfeit a large fraction of slashable stake (≥1/3 to disturb finality). Justin Drake has suggested 1/4 of supply (30M ETH) is appropriate; Vitalik has concurred while noting even 1/8 (15M ETH) would be fine [2]. At any equilibrium in the tens of percent of supply, the slashable stock is tens of millions of ETH — and acquiring it moves the price against the attacker long before they get there [1].
-
Security is stake × value, not stake alone. Elowsson’s 2016 comparison makes it vivid: doubling participation from 30% to 60% back then would have raised the finality-delay threshold from ~$70M to ~$140M which is trivial either way, because the price was the binding term, at a market cap ~500× smaller than today’s [2]. Long-term economic security is inherently linked to ETH retaining its value; a policy that protects the monetary premium (by bounding dilution) protects the value term [1, 2]. Ever-growing issuance erodes it.
-
The marginal unit of stake buys almost nothing and costs real welfare. Moving from 33% to 80% staked does not meaningfully change the cost of an attack that was already prohibitively expensive but it does concentrate custody, weaken the social-layer backstop, and force out independent validators [1, 8]. Elowsson quantifies the other side of the ledger: maintaining the current curve compels users to incur unnecessary costs notably in terms of hardware, risk, illiquidity, taxes, plausibly more than a billion dollars per year of welfare loss for security nobody needs [2].
-
Whose principal is at risk matters, not just how much is staked. A solo staker’s own 32 ETH (or whatever multiple of it they run) is what gets slashed if they misbehave or go offline. They are principal and agent at once, and the capital securing the chain is their own. A delegated validator is operated by a firm whose exposure is fee revenue and reputation, while the ETH that would be slashed belongs to somebody else. That gap is precisely the moral hazard the EIP’s Motivation describes [1], and it is a property of who the validator set is made of, not how large it is. Stake acquired by shifting the set from principals to agents does not buy the alignment that a headline staking ratio implies.
-
The floor is adjustable; the ratchet is not. If evidence ever showed the equilibrium stake to be genuinely insufficient, issuance can be raised by a future fork (an easy case to make when security demonstrably needs it). The reverse move, clawing back an overshoot after most of the supply is staked through intermediaries who profit from the status quo, is the politically hard one.
And is the status quo acceptable?
Not on this question’s own terms. Do nothing, and by early 2028 the worst-case trajectory puts >70M ETH at stake [7]: security expenditure keeps rising (≈1.4M ETH issued per year at 70M staked, versus a peak of ~0.5% of supply under the taper [1]) while buying nothing because an attack was already unaffordable. In the mean time, Ethereum which actually underwrites a multitrillion-dollar economy (a neutral, credibly scarce reserve asset and a social layer capable of intervening) is precisely what the overshoot degrades. Elowsson’s FAQ puts it directly: beyond certain limits even short-term security degrades, because the social layer loses its neutrality and credibility as the ultimate arbitrator against dominant SSPs [2].