Supply policy is the part of a crypto asset that is genuinely knowable in advance, which makes it unusually valuable and unusually easy to misuse. The rules are published, the arithmetic is simple, and almost every claim built on top of them extends far beyond what they support. Separating the mechanism from the conclusion is most of the work.
In a traditional monetary system, the quantity of money is a decision taken repeatedly by an institution that weighs conditions and can change its mind. In most crypto systems it is a rule written into the software, executed identically by every participant, and changed only by convincing enough of them to run different software. That difference is the substance of what these systems claim to offer, and it is worth stating precisely rather than as a slogan.
The practical effect is that the schedule is knowable years ahead. Anybody can compute how many units will exist at a given block height, and that computation does not depend on anybody's judgement or intentions. It is one of very few genuinely certain quantities in this field, which is exactly why so much argument gets attached to it.
What the rule does not determine is anything about value. A known quantity of something nobody wants is worth nothing, and a rule that guarantees scarcity guarantees only scarcity. Every step from the schedule to a conclusion about price requires an assumption about demand, and those assumptions are usually left unstated by whoever is making the argument.
The schedule is one of the few genuinely certain quantities in this field. Everything derived from it about value requires an assumption about demand that is usually left unstated.
The first family issues on a decreasing schedule towards a hard ceiling. New units arrive at a rate that steps down at fixed intervals, and the total converges on a number written in the rules and never exceeded. This is the design most people picture when they think about crypto supply, and it is one design among several rather than the definition of the category.
The second family issues continuously with no ceiling, usually at a rate tied to how much of the supply is participating in securing the network. Supply grows indefinitely, which sounds worse and is not automatically so: what matters to a holder is the rate relative to whatever the units are being used for, not the existence of a ceiling in the far future.
The third family combines issuance with a destruction mechanism, so the net change in supply depends on activity rather than only on a schedule. These systems can be net expanding in quiet periods and net contracting in busy ones, which makes their supply a variable rather than a constant, and comparisons with the first two families require care about which quantity is being compared.
Destroying units means sending them somewhere they can never be spent from, usually an address whose private key cannot exist, or removing them from the accounting entirely at the protocol level. Both are verifiable on the chain, and the second is cleaner because it leaves no balance that looks like it might move.
The mechanical effect is precise and narrow: there are fewer units, so every remaining unit represents a larger fraction of the total. If the total value of the network were held constant, each unit would represent more of it, which is the arithmetic behind every argument made about burns and is where most of those arguments stop.
The step that gets skipped is the condition. Nothing holds total value constant, and a burn does not add anything to the system: it redistributes a claim among fewer holders without creating what the claim is on. Whether that matters depends entirely on demand, which the mechanism does not touch, and stating the arithmetic without the condition is the standard way this gets oversold.
An asset is called deflationary when destruction can exceed issuance, so the supply can shrink. That is a statement about the rules and it is checkable. It is not a statement about purchasing power, and the ordinary economic meaning of deflation concerns prices rather than quantities of a token.
Conflating the two produces a claim the mechanism cannot support. Supply shrinking while demand shrinks faster produces a falling price, and there is no rule preventing that. Anybody who has watched an asset with a shrinking supply lose value has watched the mechanism work exactly as designed while the conclusion drawn from it failed.
The honest formulation is that a destruction mechanism removes one source of downward pressure, the arrival of new supply, and does nothing about the others. That is a real property and a modest one, and it is the version that survives contact with what actually happens.
Some networks destroy part of the transaction fee rather than paying all of it to whoever produces the block. The stated reason is usually to make fees more predictable and to align the interests of holders with usage, and both are defensible arguments with real evidence behind them.
It is worth noticing where the money comes from. A burned fee is paid by the person transacting and benefits every holder proportionally, which is a transfer from users to holders. That is not an objection, it is a description, and it is a design choice that reasonable people disagree about rather than a neutral technical detail.
It also creates a specific dependency: the destruction rate is a function of network activity, so a quiet period burns little. Any projection that assumes today's activity continues is a projection about activity dressed up as a projection about supply, and the two are frequently presented as one.
A burned fee is paid by the person transacting and benefits every holder. That is a transfer from users to holders, not a technical detail.
New units are not created for their own sake. They pay the participants who secure the network, and the amount issued is therefore a budget for that security. Reducing issuance reduces the budget, which is a genuine trade-off rather than a free improvement, and it is the trade-off that supply arguments most often ignore.
Systems with a hard ceiling face this explicitly at the end of their schedule: once issuance stops, security must be paid for entirely by transaction fees. Whether fees will be sufficient is an open question that has been argued about for years without resolution, and the honest position is that it is unresolved rather than settled in either direction.
Systems without a ceiling have made the opposite choice, accepting perpetual issuance in exchange for a security budget that does not depend on fee revenue. Neither approach is obviously right, and presenting one as the responsible design and the other as the reckless one is a preference stated as a fact.
For any system with both mechanisms, the number that matters is the net change over a period: units issued minus units destroyed. That figure is computable from public data, it is published by several independent trackers, and it frequently disagrees with the impression created by announcements about either half.
The common error is quoting one side. A large destruction figure looks impressive and means nothing without the issuance over the same period, and a low issuance rate means nothing without knowing whether destruction is happening at all. Both halves are always available and only one is usually cited.
Expressing the net as an annualised percentage of supply makes it comparable across systems, which is the form that supports an actual comparison. It also usually deflates the rhetoric considerably, because the numbers involved are smaller than the language used about them.
The schedule is in the protocol specification and, ultimately, in the code that every participant runs. That is the authoritative source, and it is more accessible than people assume: the relevant parameters are usually a handful of constants that any reader can locate.
Aggregator sites are the common secondary source and they carry two recurring errors. They quote one supply measure while labelling it another, and they carry figures that were correct at some past date and were never updated. Checking one number against the chain itself takes minutes and catches both.
The third source is announcements from the project, which describe intentions and are not the rule. An intention to reduce issuance is not a reduced issuance, and the gap between announcement and implementation is a period during which many people believe a change has happened that has not.
The entire argument for supply rules rests on their being hard to change, so the question of who can change them and how is not a footnote. Some systems require overwhelming agreement among independent participants; others can be altered by a vote among token holders or by a small group holding administrative keys.
Those are radically different guarantees wearing similar language. A schedule that a committee can revise is a policy, and calling it a rule borrows credibility from systems that earned it differently. Determining which situation applies is a matter of reading the governance arrangement, and it is published in every case worth considering.
It is also worth checking whether the rule has been changed before. A system that has revised its supply schedule has demonstrated that it can, whatever the current documentation says about permanence, and that history is a better guide to the future than any statement of intent.
Supply policy is one input among many and it is the one most amenable to confident-sounding argument, which is exactly why it dominates discussion out of proportion to its weight. It is knowable, it is arithmetic, and it therefore attracts people who want certainty in a field that offers very little.
What it cannot do is tell you what anything is worth. Demand is the other half of every price and no supply rule constrains it, which is why assets with identical supply designs have had completely different outcomes and always will. Any argument that reaches a conclusion about value from a schedule alone has skipped the half that decides.
The useful posture is to treat supply policy as one checkable fact among several: know the schedule, know the net change, know who can alter it, and then stop, because the next step is not supported by any of them. Knowing where an argument stops being evidence is more valuable than the argument.
Issuance in most crypto systems is a rule written into software and executed identically by every participant, changed only by convincing enough of them to run different software. Traditional monetary policy is a decision taken repeatedly by an institution that weighs conditions and can change its mind.
No. A known quantity of something nobody wants is worth nothing, and a rule that guarantees scarcity guarantees only scarcity. Every step from a schedule to a conclusion about price requires an assumption about demand, and those assumptions are usually left unstated.
It removes units from circulation permanently, so each remaining unit represents a larger fraction of the total. That is the whole mechanical effect. It adds nothing to the system: it redistributes a claim among fewer holders without creating what the claim is on.
No. Deflationary is a statement about the rules, that destruction can exceed issuance so supply can shrink. It is not a statement about purchasing power. Supply shrinking while demand shrinks faster produces a falling price, and no rule prevents that.
The person transacting, and every holder benefits proportionally. It is a transfer from users to holders, which is a design choice reasonable people disagree about rather than a neutral technical detail. It also means the destruction rate depends on activity, so a quiet period burns little.
It pays the participants who secure the network, so the amount issued is a security budget. Reducing issuance reduces that budget, which is a genuine trade-off. Systems with a hard ceiling must eventually fund security entirely from fees, and whether that suffices is an open question.
Units issued minus units destroyed over the same period, expressed as an annualised percentage of supply so it is comparable across systems. The common error is quoting one side: a large burn figure means nothing without issuance, and low issuance means nothing without knowing whether destruction happens.
It depends entirely on governance, and the difference is radical. Some systems require overwhelming agreement among independent participants; others can be altered by a token holder vote or by administrative keys. A schedule a committee can revise is a policy, and it is worth checking whether it has been revised before.