Verified Platforms
Quick Links
Where to Stay Secure
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
The number on your staking dashboard is not a rate you are paid. It is what is left after arithmetic you never agreed to. Open an Ethereum staking page today and you will see something close to 2.6%, printed to two decimals with the calm of a bank statement. Nobody is promising you that figure. There is no counterparty standing behind it, no contract, no desk that owes you the difference when it moves. The number is the output of a formula, and formulas belong to whoever can rewrite them.
On 4 August, six researchers filed a draft proposal to rewrite this one.
Eunha teaches the seam where structure meets feeling, and she works by question rather than by answer. She does not open with the number. "Who owes you the 2.6%?" she asks. "Name them."
You reach for a name and find none. The validator is you, or someone you hired. The exchange passes through whatever arrives. There is no institution anywhere in the chain that owes you a percentage.
Eunha starts at the source, because that is where the answer is hiding.
Ethereum's consensus layer currently issues about 1,054,000 ETH a year to the validator set. Measured against the staked pool, that works out to roughly 2.62%. A second and much smaller stream arrives from the execution layer, priority fees and MEV, which together add at most about 0.20% and rise and fall with how busy the chain is. Issuance accounts for at least 93% of what a staker earns.
That word, issued, is carrying more weight than it looks.
When a bond pays a coupon, money leaves an issuer's account and arrives in yours, and if it fails to arrive you have a claim against a named party. When Ethereum pays a validator, no account is debited. The protocol creates new ETH and assigns it to you. Supply expands by exactly the amount you were credited, and nobody's balance went down to make room for it.
"A payment has a payer," Eunha says. "This has a printer."
So the question of whether staking yield is guaranteed answers itself before you get anywhere near the percentage. A guarantee needs somebody capable of being held to it. There is no such person inside issuance.
Ethereum's issuance rises with the square root of the amount staked, not in step with it. Double the staked pool and total issuance climbs by about 41%, not 100%, and that larger pot is now spread across twice as many validators. Per staked ETH, the rate falls with the inverse square root of the staking ratio. Multiply the pool by four and your rate halves.
This is live today. It is not part of any proposal.
Nothing you do touches it. You can run perfect uptime, never miss an attestation, never get slashed, and still watch your annual rate slide, because the slide has nothing to do with your behaviour. The denominator moved. Somewhere a fund deposited 40,000 ETH, and every validator on the network absorbed a fraction of that dilution in the same instant, quietly, with no notification and no way to opt out.
Roughly 39 million ETH is staked against a supply of about 120.7 million, so the network sits near a third of the way along that curve. The remaining two thirds is the part that has not happened yet. Every institution still deciding whether to stake is deciding, without knowing it, what your rate will be next year.
Eunha calls this the first owner of your yield: everyone else.
Follow that curve far enough and something strange happens. It stops falling.
Even if every ETH in existence were staked, the present formula still pays roughly 1.5%. The rate bends toward that level and settles on it. No economic force produces the number and no equilibrium is being discovered. The floor is there because the issuance curve was drawn to keep it there, which makes it a decision rather than a discovery. The same shape governs the UNI fee switch, where whether a token earns anything at all comes down to a setting somebody is permitted to flip.
A parameter that has never changed is not the same thing as a parameter that cannot change.
"Ask who wrote it," Eunha says. "Then ask what it would take for them to write it differently."
On 4 August, six people answered the second question.
EIP-8361, titled Tapered Issuance Burn, is a draft. Standards Track, Core category, requires a hard fork, filed by six authors including the Ethereum Foundation's Justin Drake alongside Jérôme de Tychey, Anders Elowsson, Ladislaus von Daniels, pintail and pa7x1. The Defiant reported it at "Proposed for Inclusion," the weakest stage Ethereum has, which commits no client team to anything.
What it does is narrow. Rewards keep being calculated exactly as they are calculated now. Then a share of them is destroyed before it reaches anyone, and that share grows as the staking ratio grows, scaling with the ratio raised to the power of 1.5. The burn reaches 100% at a fixed saturation balance of 60,250,000 ETH, roughly half the current supply of 120.7 million. At that level a validator performing every duty flawlessly earns zero net consensus yield. At today's staking ratio the authors put the effect at about 2.6% falling to about 1.2%, phased in across 18 months and 123,300 epochs through a linearly decaying factor.
Slow down for this part, because it is the misread waiting to happen: the burn never touches your stake. It applies to newly issued rewards and to nothing else. Your 32 ETH does not shrink. What shrinks is ETH that would have been created and handed to you, and that ETH does not exist yet, so nothing is taken from you in the ordinary sense of the word.
The objections arrived fast, and they are not the same objection. Greg Koumoutsos went at the process, noting the proposal landed 48 hours before the submission deadline, which "clearly doesn't leave adequate time for community review of a monetary policy change of this magnitude." Aave's Stani Kulechov went at the economics, running it at the current 39 million ETH staked and getting a 48% cut to validator income, from 2.862% to 1.476%. His framing was sharper than his arithmetic: "A zero-yield regime accelerates the capture it means to deter." ether.fi's Mike Silagadze argued the mechanism would strip out the solo stakers it claims to defend, concentrating the validator set among whoever holds the cheapest capital. Jérôme de Tychey, defending it publicly, points at the 18-month taper plus roughly six months of fork lead time and answers that nobody needs protecting from a change with a two-year runway.
His positive case is worth stating at full strength, because it explains why anyone would want this. Ethereum's supply currently grows about 0.9% a year and drifts toward 1% as the staked pool expands. Halving that to roughly 0.5% withholds on the order of a billion dollars a year of new ETH at a $2,000 price. Today that dilution is paid by everyone holding ETH, to reward the subset who stake it. The disagreement is not really about whether yield falls. It is about which group of holders should be funding the security budget.
So what actually changes here, and what was already true before anyone filed anything?
| Today | Under EIP-8361 | |
|---|---|---|
| What sets your rate | Issuance rises with the square root of the staked pool | The same formula, then a burn scaling with the staking ratio to the power of 1.5 |
| Where it stops falling | A floor near 1.5%, however much ETH is staked | Zero net consensus yield at 60,250,000 ETH staked |
| What you would notice | A rate that slides as others deposit | The same slide, steeper, with nothing underneath it |
The proposal is not what makes your yield uncertain. It only makes the machinery visible. A rate that already drifts toward a floor somebody chose was never contractual, and it does not become more contractual for having survived this long.
What would any of this look like from where you sit, with your wallet open in front of you? Nothing at all. That is the part that transfers to everything else you will ever hold.
A fee is visible. It has a line, a percentage, a place on a statement, and a party attached to the collection of it. You can add fees up. You can compare them between providers. You can be annoyed at somebody specific about them.
A burn at issuance has none of that. The reward is reduced before it exists, which means no transaction, no counterparty, no line item, nothing to reconcile against. Your balance is correct. It was always going to be that number. The only trace left behind is that a rate you saw last year reads lower this year, and rates move for plenty of reasons.
You cannot audit a subtraction that happened upstream of your balance.
The accounting is not settled either. Kulechov asked for written tax opinions from the US, the UK, Germany and Portugal, because a jurisdiction that taxes staking rewards on receipt has to decide what exactly was received when a reward is calculated and then destroyed. Nobody has answered him.
Set that against a fixed rate stablecoin yield, where the number holds still because a counterparty took the other side of it and can be pointed at when it breaks. Staking has nobody on the other side. That absence is the reason the rate moves, and the same reason nobody is able to promise it.
Nobody is able to promise it, and yet the product you are more likely to actually hold prints a number on the front page.
Staking ETPs and ETFs quote yields, and the language wrapped around them is careful. Bitwise's staking guide tells you that "the effective staking rate is variable and depends on factors such as network conditions, protocol-specific reward mechanisms, validator performance, and overall market dynamics." Every word of that is accurate.
Read it again with the curve in hand and it stops sounding like boilerplate. "Protocol-specific reward mechanisms" is the issuance formula. "Network conditions" includes how much ETH strangers deposited this quarter. The disclosure is describing precisely the machinery you now know about, in the general language that disclosure gets written in.
The wrapper does not fix any of it. It sits on top of it. Whatever the protocol issues, less the burn, less the operator's cut, is what reaches the fund before the fund takes its own. ETF reverse splits showed how a wrapper's share price sets what it costs you to trade. The yield runs the same logic in reverse. The number on the factsheet is an output being passed through, not a rate being set.
There is one more asymmetry buried in that arrangement, and it is the one that costs money. The rate can be changed by a process you do not participate in, at a speed set by a hard fork schedule. Your ability to leave runs at a different speed entirely, through an exit queue that lengthens exactly when everyone reaches the same conclusion at once. A wrapper adds its own redemption terms on top of that. So the number can move faster than you can respond to it moving, which is the definition of a position you do not fully control.
Eunha's rule for the whole category fits in a sentence. If the marketing number and the protocol number are the same number, nobody in that chain is guaranteeing anything.
Eunha finishes where she opened, on the question you could not answer.
Every yield you are ever quoted is either contracted or computed. A contracted yield has a counterparty: a bank, an issuer, a treasury, somebody who owes you the number, can be held to it, and whose failure to pay is an event with a name attached. A computed yield has an author: a formula, a curve, a parameter set, and a process by which those get changed. Both can pay you well for years. Only one of them can be promised.
Telling them apart does not take a checklist. It takes Eunha's opening question, asked before you size the position: who owes you this number, and can you name them.
Ethereum answers it in the open, and the answer is nobody: an issuance curve sets the rate, everyone else's deposits move it, and an EIP process six people used on 4 August can rewrite it. Ask the same question at stablecoin yield and the answer comes back different, which is the point of asking it.
None of this makes a staking yield unsafe to hold. It makes it a variable, and variables are fine to hold as long as you knew that was what you were holding. The failure is not the rate moving. The failure is finding out on the day it moves that you had been reading it as a promise.
Eunha's version of this exercise is not about staking at all. Open the simulator, size any position on the $5,000 practice balance, and before you commit, write down the one number you are counting on and the name of whoever is on the hook for it. When that name comes back blank, you have found a computed number, and the next one will be easier to spot.
