Four point three six percent. That is the number HSP-029 puts on six years of NEAR emission cuts, and it is the number every summary of the proposal will repeat back to you, unchanged, for the next two quarters. It is also a fraction of a token supply that will never exist.
I ran the proposal's arithmetic before I read its conclusions — a habit that killed a €500k seed round in 2017 and has not once made me popular at conferences. The staking-yield identity reconciles to the third decimal. The supply math reconciles to within two basis points of a rounding decision. And the headline number is computed against a counterfactual: the NEAR supply that would have existed if the protocol had done nothing at all. Measured against the supply that actually exists, the cut is closer to 5.0%. The proposal is not lying to you. It is choosing a denominator, and in a market that has spent five months going nowhere, the choice of denominator is the entire argument.
There is a second thing I found, and it is the reason I am writing this at all. The proposal's implementation parameter — a reduction of 3.75 basis points "per epoch" — is not an epoch parameter. Ninety basis points divided by twenty-four steps is 3.75. It is a monthly ladder, labeled as an epoch ladder. Anyone who ships it literally would collapse issuance from 2.5% to 1.6% in roughly twelve days.
HSP-029 proposes a 24-month linear glide from 2.5% annual issuance to 1.6%, with 90% of issuance routed to stakers against a current staking ratio of 42.71%. Those are the four numbers that matter. Everything else in the document is commentary on them.
The context is worth stating precisely, because the debate around it has been sloppy. NEAR is a sharded L1 whose security budget is paid in newly minted tokens rather than in fee burn. Issuance is therefore not a monetary abstraction — it is the line item that keeps validators solvent and the line item that dilutes everyone who is not staking. The 90/10 split between stakers and the protocol bucket means that when you cut issuance, you are cutting validator revenue almost one-for-one. The 42.71% staking ratio is the denominator that converts an aggregate reward pool into a per-token yield, and it is the number the entire proposal treats as a constant.
That last assumption deserves a full section, and it gets one below. But first, the basic mechanics. A standard L1 staking APR is not a policy variable. It is an equilibrium output: annual inflation, multiplied by the share of issuance paid to stakers, divided by the share of supply actually staked. The protocol sets the numerator. The market sets the denominator. Proposals that confuse the two are not economic policy; they are arithmetic with a press release attached.
The glide path itself is unremarkable, which is precisely why it is interesting. Twenty-four months of linearity is a governance artifact, not an economic one. It exists so that the change can be described as gradual, so that no single cohort can be identified as the loser, and so that the chart in the governance forum renders as a straight line descending to the right. Liquidity doesn't read governance forums. It reads the second derivative.
So let me give you the four checks I ran, in the order I ran them, including the one that does not reconcile.
Check one: the staking yield identity reconciles perfectly, and that is the problem. At 2.5% issuance, 90% to stakers, 42.71% staked, the APR is 5.268%. At 1.6%, holding the other two inputs fixed, it is 3.372%. The proposal reports both to three decimals and both reproduce exactly. This is not a validation of the forecast. It is a demonstration that the forecast contains no forecast — it is a static substitution into an identity that holds only if no staker ever reacts. Three point three seven two percent is the yield in a world where the cut is announced and forty-two point seven one percent of supply stays staked out of politeness. The protocol does not control per-staker yield. It controls the aggregate reward pool, and it is cutting that pool by 36% in nominal terms. Where the per-token yield lands depends entirely on how many tokens show up to compete for it. If the staking ratio decays to 38%, the realized APR at 1.6% issuance is 3.79%. At 35%, it is 4.11%. The advertised 190 basis point haircut is probably a 120 to 150 basis point haircut, and the proposal's own model is structurally incapable of saying so.
Check two: the total supply reverse-engineers cleanly to roughly 1.31 billion NEAR. Two years of linear glide averages 2.05% annual inflation, which is 0.45 percentage points per year below the status quo, which the proposal converts into approximately 11.8 million tokens, which implies a supply base near 1.31 billion. That is the right order of magnitude for NEAR. Fine. Note what just happened, though: the supply figure is not an input to the proposal, it is an output of it. The document never states its own denominator. It back-solves for it and moves on.
Check three: the 4.36% headline is a denominator switch, and it compounds. Here the arithmetic is internally perfect and externally misleading in a way that is entirely standard for this genre. Take a 1.3 billion base. Run the counterfactual — six years at 2.5% — and you get 1.508 billion. Now run the actual glide path: two years averaging 2.05%, then four years at 1.6%, which compounds to 1.4425 billion. The gap is roughly 65.5 million tokens, which is 4.34% of the counterfactual supply and 5.08% of today's supply. The proposal quotes 4.36%. Everyone quotes 4.36%. Nobody quotes 5.08%, because 5.08% is measured against a number that exists and 4.36% is measured against a number that doesn't. This is not fraud. It is the standard practice of every tokenomics document ever written, and the only defense is to read the denominator before you read the percentage.
Check four: the per-epoch parameter is mislabeled by a factor of sixty. Ninety basis points of total reduction, spread across twenty-four monthly steps, is exactly 3.75 basis points per step. Spread across NEAR's roughly 12-hour epochs — call it 1,460 of them over 24 months — it is 0.062 basis points per epoch. The stated figure matches the monthly ladder to the decimal and misses the epoch interpretation by an order of magnitude. This is the kind of error that lives harmlessly in a governance forum and lethally in a reference implementation. A parameter that is arithmetically consistent with two different schedules, only one of which is stated, is not a specification. It is a coin flip with a changelog.
The fifth thing I checked is not arithmetic. It is provenance. The proposal leans on a Solana precedent whose cited timestamp reads August 2026, and several of its key figures are internally consistent but externally unverifiable against any source I can reach. I have audited enough of these documents to know what that pattern means. Internal consistency is a property of arithmetic, not of truth. A spreadsheet that agrees with itself is a spreadsheet that agrees with itself. The auditor blinked; the market didn't — it priced the narrative and moved on to the next governance post.
Which brings me to what I think is actually happening here, and it is not monetary policy.
This is a validator-margin event wearing a monetary-policy costume. Cut the reward pool by 36% in nominal terms and you have not changed the security budget's dollar value only if the token price rises to compensate. It will not, on schedule, because price is set by flows and flows are set by macro liquidity, and liquidity does not care about your glide path. What you have done is compress gross margin for every operator whose costs are denominated in euros, dollars, or colocation invoices. The large operators absorb it. The mid-tier operators with a fixed cost base and no treasury runway exit over the following four quarters. Delegation concentrates into the survivors. And then, two years later, the same governance forum hosts a thread asking why the validator set has thinned out — a thread that will contain every one of the numbers in this proposal and none of the causality. The emission schedule is a decentralization parameter. Nobody in this debate is treating it as one.
The second thing the model gets wrong is the functional form, not the assumption. A linear 24-month glide path is a model of human inertia. It assumes that stakers are people who check a dashboard quarterly and react slowly. The marginal NEAR staker in 2026 is not a person. It is an agent executing a yield-threshold rule across a dozen chains, and a 190 basis point nominal change — or a 120 basis point realized change — is not a slow drift, it is a discrete trigger. Agents do not decay linearly. They step. The staking ratio will not walk from 42.71% to 38%; it will gap on the announcement, partially recover on the realized-yield print, and gap again when a competing chain adjusts its own schedule. Every one of those steps feeds back into the APR through the denominator, which means the yield series the proposal publishes and the yield series stakers actually experience will diverge from month one. I have argued for two years that algorithmic actors need to be modeled as a distinct class rather than as fast humans. This document is a clean case study: it models them as slow humans.
And here is the contrarian read on why the proposal exists at all. Emission cuts are not primarily an economic lever for a protocol in a sideways tape — they are a product feature. A predictable, declining issuance curve is a due-diligence line item for any regulated wrapper that wants to market a large-cap L1 to European allocators under a post-MiCA framework. "Supply schedule" is a checkbox. "Deflationary trajectory" is a slide. The 4.36% figure is not aimed at stakers, who can do the denominator math and will be annoyed by it. It is aimed at the compliance officer who needs one number, one date, and one linear chart. That is a real utility. It is just not the utility the proposal claims to be optimizing for, and the gap between the two is where the next twelve months of confusion will live.
The last blind spot is procyclicality. Security spend is already procyclical: when price falls, the dollar value of a fixed nominal reward falls with it. This proposal makes the nominal reward smaller too, permanently, at the bottom of the next drawdown rather than at the top. That is a defensible trade if you believe the drawdown is far away. It is a much harder trade if you have watched what a 60% price decline does to a validator set that has already lost its mid-tier. Liquidity doesn't wait for the six-year chart to finish rendering before it makes that judgment.
So watch the denominator, not the numerator. If the staking ratio holds above 40% through month twelve, the static model survives contact with reality and the 3.372% figure is roughly honest. If it breaks 38% before month eight, the realized yield cut is materially smaller than advertised, the security budget conversation starts early, and the 4.36% becomes the least interesting number in the document. And if the implementation ships with "3.75 basis points per epoch" written into the spec, you will know within two weeks that nobody who wrote it ran the division.
The question worth asking is not whether NEAR should cut issuance. It is what a protocol is actually selling when it sells a supply schedule to an audience that will never stake a single token — and whether the people who do stake were ever the intended customer.