The last word on the Bitcoin filter debate

The Great Bitcoin Filter Debate of 2025 rages on, and it’s getting pretty brutal.

Conspiracies are being loosely thrown around on X (formerly Twitter), including accusations of bribery. Some even go so far as to warn the the next version of Bitcoin–Bitcoin Core v30– will “destroy Bitcoin” if run.

The following is an Op-Ed from Charlie Spears, co-founder of Blockspace Media and host of Bitcoin Season 2, a Bitcoin tech podcast you can subscribe to here.

If you’re walking into this a bit flat footed, the online drama in the Bitcoin tech spaces began with Ordinals and Inscriptions, a way to place image data on the Bitcoin blockchain. Also called Bitcoin-NFTs, backlash against these Ordinals grew rapidly from there appearance in 2023 and a growing sect of Bitcoiners have advocated for “filtering” these Bitcoin transactions out of the blockchain.

Filter would simply prevent Ordinals from reaching Bitcoin miners, who create blocks of transactions that are then validated by Bitcoin nodes. The debate over whether NFTs belong on Bitcoin or whether they are spam has become heated, passionate, and often virulent. And it’s reaching a crescendo.

Previously, I had tried to somewhat neutrally cover the topic, although I will be forward and say I’ve never personally been neutral. I believe the attempts to filter economically demanded transactions on Bitcoin are at best misguided, but the overwhelming majority of those pushing the narrative are disingenuous and hostile actors.

Your attempts to filter Bitcoin are futile

Some claim filters work. But they do not when a certain type of Bitcoin transaction has economic demand and those transactions are already being mined.

This is the same principle that ensures that a nation state has an asymmetric challenge censoring transactions on Bitcoin: it is incredibly difficult to prevent a transaction from propagating. And while a nation may ban mining, you cannot control others from mining it. For example, if the U.S. were to ban Bitcoin transactions, they would find it difficult to censor the global network of miners outside there jurisdiction, say in Iran where Bitcoin mining is widespread.

Read More: Inside Iran’s thriving, underground bitcoin mining scene

1 sat/vbyte summer

What if we could change Bitcoin’s rules to filter out certain transactions? Let’s take a look at an example from this summer: the filtering of 1 sat/vbyte transactions. Despite 100% of users filtering transactions with extremely low fee rates, those transactions were still being mined in over half of the blocks.

How many nodes would have to filter transactions to impose a reliable cost (i.e., a disincentive to send a filtered transaction)? One Bitcoin developer who goes by the online pseudonym “PortlandHODL” demonstrates that the number is upwards of 98% of all nodes to impose a modest cost. (Note: mining node policy is historically quite effective at preventing certain transaction types).

Story Continues In other words, even with 100% filtering we still see the iron laws of economics prevail: if there’s a paying customer, someone will cash the check.

Fees are the filters, and it works

Bitcoin spam itself is a subjective evaluation: one man’s spam is another man’s economic transaction.

The best way to disincentivize transactions the majority of users consider to be spam, then, it is to assign a cost to transactions that are a function of the amount of data: make them pay for it! Empirically, fees are highly effective at disincentivizing spam. For example, image-based, data-heavy Inscriptions are less likely to be pushed across the network when fee rates are high, due to the higher cost of sending the data, as shown by Data Always.

What about “banning” useless data? Unfortunately, that falls flat on its face. Users can insert arbitrary data on Bitcoin through other means like public keys–albeit at twice the cost–as Bitcoin developer Andrew Poelstra puts it:

“If 2x is enough to disincentivize storage, then there’s no need to have this discussion because they will be forced to stop due to fee market competition anyway. And if not, it means there is little demand for Bitcoin blockspace, so what’s the problem with paying miners to fill it with data that validators don’t even need…?”

The ugly side of Bitcoin development

It’s worth mentioning that the debate has gotten pretty ugly, with claims of tacit or explicit support for child exploitation, not-so-veiled death threats, and delusional conspiracy floating around Bitcoin discourse.

I won’t link to examples because I do not want to give those posts views. I do want to be very clear: these are not niche narratives from a single account. These are the prevailing attitudes from the filter camp whether it be obsessive posting about child exploitation. Some pundits go so far as to claim that Inscriptions are a form of “violence”akin to “attacking [people] in their homes.”

One of the reasons I don’t try to refute these accusations directly is because these users purposely do not want to have a good faith argument. They’d rather drag you down into the repugnance of these topics, trying to disgust you and everyone reading on social media. You’re wrestling a pig in the mud.

Some time in the future when the debate has lost steam, many of the leaders of this movement may claim these were never actually mainstream narratives that they drove. It needs to be documented that these are actually the things they are saying and encouraging today. Without proper accountability, the Bitcoin economy will have a difficult time in the future finding developers willing to place their reputations on the line.

But, what about…?

As a prominent media voice on the topic, I’ve had to endure hundreds of replies with the same exhausting and reductionist takes. So, I ask you too, dear reader, to read all of these before you reply:

• Articles on mempool policy & filtering: 1, 2, 3, 4, 5, 6

• Exhaustive refutation of OPRETURN and policy misinformation

• Another 50+ tweet exchange refuting spurious claims

• Datacarriersize was not a bug

• Do we really think a hypothetical state actor cares about plaintext vs “encrypted”?

• OPRETURN vs Inscription cost

  View Comments