The Fee Is the Filter

Every so often, a familiar proposal returns to Bitcoin. The specifics change, but the shape is always the same. The network is filling with "spam," inscribed images, tokens, arbitrary data, and something must be done at the protocol level before the junk crowds out the money. The most recent version has a number, but the impulse is older than any single proposal and will outlive this one too.

Underneath every version sits one assumption: that without a deliberate intervention, unwanted data will overwhelm Bitcoin. That assumption is wrong, and it is worth explaining exactly why. The reason is not that the data is harmless or that the worry is foolish. It is that Bitcoin already contains a mechanism for this, designed in from the start, that works by price rather than by permission.

What "spam" actually means

Start with the word, because it is doing more work than it looks. "Spam" is a verdict, not a category. One person's worthless picture is another person's art, and the protocol has no way to tell them apart and no business trying.

So define it by function instead of by taste. Spam is data that competes for scarce blockspace while carrying no monetary purpose. That definition has a sharp edge: the moment a use of blockspace carries a monetary return, it stops being spam and becomes paying demand for a scarce resource, which is exactly what a fee market is built to price. What remains under the label is, by construction, the set of uses that do not pay for themselves. Hold onto that, because it is the whole argument in miniature.

The strongest case for stepping in

Before taking the intervention apart, it deserves its strongest form. A weak version of the opposing case is a waste of everyone's time.

Every full node must download and validate every block, forever. When someone stores data by creating outputs that will never be spent, those entries lodge in the UTXO set, the running list of spendable coins that nodes keep on fast storage and cannot simply discard. Unlike prunable data, that cost never fades. It is carried by every node that will ever validate the chain.

Large inscriptions also compete with ordinary payments for room in a block and can push fees up for people who only want to send money. And some of what gets embedded is content a node operator would rather not store at all, from the merely tasteless to the genuinely illegal. If you believe Bitcoin's first purpose is to be money, watching its permanent record fill with data that free-rides on every participant is not a petty complaint. It is a real cost, honestly felt. Anyone who waves it away has not understood the objection.

The disagreement is narrower than it looks. It is not about whether data embedding has costs. It is about what to do about them.

Satoshi already built the filter

Here is what the intervention overlooks. Satoshi did not leave spam unaddressed. He addressed it economically, and the tools are still running.

Two levers work together. The first is a hard limit on how much data a block can hold, roughly one megabyte of base data and up to about four megabytes once the witness discount is counted. The second is a fee market: because blockspace is capped, transactions bid against each other to get in, and the price of inclusion rises as demand rises.

The whitepaper frames fees as an incentive, the reward that keeps miners honest once the block subsidy fades. But Satoshi also spoke about them plainly as a defense. Responding to an actual transaction flood in 2010, he described tightening the priority rules so that free transactions could not be spun in endless circles, adding that if you are using free transactions "you're taking charity and there has to be some limit." The logic then is the logic now: make the scarce resource cost something, and abuse has to justify its own expense.

Picture the everyday version. A spammer who pays nothing to send a million emails will send a million emails. A spammer who must pay one Euro per email, and who turns almost none of them into money, stops. Bitcoin does to blockspace what a per-message charge would do to email. When blocks fill, the fee to enter rises, and any use that produces no monetary return runs into a bill it cannot pay. This is not a patch bolted on after the fact. It is the design.

The chain has proved it, wave after wave

This is not a thought experiment. It is the observed behavior of the network.

When the Ordinals protocol arrived in early 2023 and the BRC-20 token standard followed that spring, inscription activity surged. On May 7, 2023, the fees in a single block exceeded the block subsidy for the first time since 2017. A year later, the April 2024 halving coincided with the launch of the Runes protocol, and the frenzy drove the average transaction fee to nearly 128 Dollar in a single day, with the busiest block collecting more than 37 Bitcoin in fees alone.

Each time, the same thing happened next. As blockspace grew expensive, the activity that depended on cheap space stopped paying off, and it receded. By 2025 fees had fallen more than 80 percent from that peak, and near-empty blocks priced at a single satoshi per virtual byte became common again.

No consensus rule was changed to produce that result. The market cleared it, exactly as designed. The spam did not have to be banned. It priced itself out.

A ban moves the data, it does not remove it

Now suppose you try to solve this with a rule anyway. You forbid the current method of stuffing data into a transaction. What happens to the data? It does not evaporate. It moves.

This is the part that turns a cure into a carrier of the disease. Faced with a ban on one obvious method, a determined actor splits the payload across many small pushes, or disguises it as ordinary financial data spread across many outputs, and those outputs land in the UTXO set, the one place nodes cannot prune. The intervention meant to keep data out of the chain ends up steering it into the most expensive and most permanent place to store it.

The history of the data-limit fight illustrates the trap. OP_RETURN itself was introduced in 2014 precisely as a lesser evil, a prunable place to park data so it would stay out of the UTXO set. The long argument since then over how tightly to cap it keeps rediscovering the same fact: restrict one channel and the water finds another, often a worse one. A protocol-level ban does not remove the demand for data storage. It only changes the shape of the mess. We have made a version of this case before in a different context, that a treatment aimed at a symptom can end up feeding the underlying condition.

The real question is what the rules are for

So the choice underneath the noise is not "spam or no spam." It is a question about what Bitcoin's rules are for. Should the protocol only check whether a transaction is valid under neutral rules that treat every byte the same, or should it additionally decide which purposes are permitted and which are forbidden?

This essay takes a side, and it is worth saying why rather than just asserting it. Neutral rules plus a fee market are more robust than curated permission for one plain reason: nobody has to be trusted to draw the line. A rule that bans "spam" requires someone to define spam, to keep the definition current as techniques change, and to enforce it through a consensus change that every future disagreement can reopen. A fee market requires none of that. It does not care what the data is. It only asks whether a use is worth the price of the space it takes.

That is the deeper point. Price is the one signal that gathers up knowledge scattered across millions of people whom no committee could ever poll, and it settles the question continuously without anyone deciding it. This is the oldest argument in economics against central direction, and it is why Bitcoin's approach holds. The strength of a neutral rule is precisely that it refuses to know what you are doing.

Where the argument stops

An honest argument names where it stops. The fee market is not a solvent for every problem the critics raise. UTXO bloat is real, and coins locked into the set by past data storage will sit there whether or not fees are high today. When blockspace is cheap, as it is now, the cost of embedding data falls, and quiet periods do invite exactly the low-value use that busy periods punish. The mechanism regulates over time and in aggregate. It does not switch spam off in a given week. And it offers no comfort to a node operator disturbed by specific content, because neutrality cuts both ways.

None of that changes the conclusion. The claim here is not that Bitcoin has no problem. It is that the problem does not justify the cure. A recurring, self-limiting cost is a manageable thing. A rushed, low-agreement change to the rules that define Bitcoin, one that relocates the data it meant to remove and can fracture the network in the process, is not a fix for that cost. It is a larger risk taken on to avoid a smaller one.

Bitcoin does not need a new filter. It has had one since the beginning, and it is priced into every block. For the specific proposal that prompted this argument, and the mechanics of how its activation could split the chain, see our analysis in BIP-110 explained.