Numbering 2.1 Quadrillion Satoshis: Casey Rodarmor, Ordinals, and Bitcoin's Block Space Civil War

1. The Temple of Boring Money

By the early 2020s, Bitcoin's identity had hardened into doctrine: the ledger was for money, block space was scarce, and everything else belonged on other chains. The culture had watched NFTs take off on Ethereum in 2017 and again in 2021 and passed on both booms. Nothing in the protocol forbade putting a picture on the blockchain; the barrier was assumption, not code.^[1]

Rodarmor encountered Ethereum NFTs in 2017 and experimented again in 2021. He disliked the tooling, varying ERC-721 semantics, and off-chain artwork on a chain he already distrusted. Rather than dismiss NFTs, he set out to build them on Bitcoin, UTXO-based and using Bitcoin's own script and cryptography.^[1]

The obstacle was that UTXOs are ephemeral: they pop into existence when created and vanish when spent — poor homes for permanent artifacts. Rodarmor's answer was to track the coins themselves, an idea first committed to a repository named bitcoin-atoms in December 2021. In a detail the saga would find poetic, the project's script-based NFTs were originally called runes; the team switched to inscriptions, and runes went back on the shelf.^[1]

The rollout was quiet. On 2022-12-14, Rodarmor made the first inscription on Bitcoin mainnet, quickly followed by a second from another developer. The ord wallet's inscribe command stayed disabled on mainnet until 2023-01-09; on 2023-01-20, he announced that inscriptions were ready. In five weeks a private experiment became public infrastructure: no fork, no activation vote, no permission.^[1]

2. Numbering Satoshis Without Asking Permission

Ordinal theory, the conceptual core, is almost insultingly simple. Satoshis are numbered in the order they are mined, and when a transaction spends inputs and creates outputs, satoshis move from inputs to outputs in first-in-first-out order. Follow the two rules and every satoshi acquires a serial number and a paper trail: a specific coin can be identified, tracked, and hoarded as a discrete object rather than an anonymous fraction of a balance. Nothing about the convention changes the ledger — it is a way of reading it.^[2]

For a community whose reflexive question about anything new was always the same — does this change Bitcoin? — his rebuttal fit in one sentence:^[2]

Ordinals don't require a separate token, another blockchain, or any changes to Bitcoin. They work right now.^[2]

— Casey Rodarmor

The mechanics came from parts Bitcoin already shipped. Inscription content is stored on-chain, in Taproot script-path spend scripts — which face very few restrictions on their content and receive the witness discount, making data storage relatively economical. Each inscription is a two-phase procedure: a commit transaction creates a Taproot output pledged to a script containing the content, and a reveal transaction spends that output, exposing the data on-chain. The payload sits inside envelopes, data pushes wrapped in OP_FALSE OP_IF … OP_ENDIF conditionals that never execute and leave the script's semantics untouched.^[3]

3. The Spam Wars

The backlash began inside Bitcoin Core's own repository, technical before it was cultural. Core ships a policy option called datacarriersize, which caps how much extra data a transaction may carry for relaying and mining. An issue titled 'Witness scripts being abused to bypass datacarriersize limit' argued that inscription users were obfuscating data inside OP_FALSE OP_IF patterns to slip past the cap instead of using the standardized OP_RETURN — spam, in that framing, under active exploitation.^[7]

The war had two arenas: pull request 28408, which proposed that default software match and restrict more data-carrying scripts, and issue 29187, where the spam label itself was disputed thread by thread. Both were fights over policy — what nodes relay and miners include by default — not over consensus rules, and nobody proposed revoking a miner's freedom to include any valid transaction. That distinction kept the war cold: the weapon was default software, not the protocol.^[6][7]

The sharpest reply came from Peter Todd, who opened his review with a concept NACK. The transactions it targeted were a very significant source of fee revenue for miners, he observed, and miners do not volunteer to surrender revenue. His warning about the cure was blunt:^[6]

Censoring those transactions would simply encourage the development of private mempools - harmful to small miners - while making fee estimation less reliable.^[6]

— Peter Todd

4. Runes and the Halving Block

In September 2023, Rodarmor turned to the corner of the market even inscription enthusiasts found embarrassing: fungible tokens. On 2023-09-25 he published his rationale for Runes, opening unsure whether Bitcoin should host one at all. The case was harm reduction: a well-designed protocol could bring fee revenue, developer mindshare, and users to Bitcoin while keeping a small on-chain footprint and encouraging responsible UTXO management — whereas the popular incumbent, BRC-20, had the undesirable consequence of UTXO proliferation. His opening line was built to disarm:^[4]

Fungible tokens are 99.9% scams and memes.^[4]

— Casey Rodarmor

The design answered complaints the inscription era had generated. Rune balances are held by UTXOs, avoiding the junk outputs that bloat chain state; protocol messages ride in OP_RETURN data pushes rather than script envelopes. Even the specification made a statement: its documentation is titled 'Runes Does Not Have a Specification' and defers to the ord code, which is normative — the prose is merely a guide. The launch was timed for theater: the protocol activates on block 840,000 and ignores runestones in any earlier block.^[4][5]

Block 840,000 was the fourth halving block. At timestamp 1713571767 it carried 3,050 transactions. Miners collected 3,762,561,499 sats in fees — 37.62561499 BTC — on top of the freshly halved 3.125 BTC subsidy, for a total reward of 4,075,061,499 sats; fees alone outweighed the subsidy more than twelvefold. The Runes protocol activated on this exact halving block where these total fees were recorded, marking a historic coincidence of protocol activation and block congestion.^[5][8]

5. What the Ledger Answers To

Block 840,000 recorded 37.62561499 BTC in fees at the halving and Runes activation, showing intense demand for block space without allocating fees by cause. Supporters cited miner revenue; critics in pull request 28408 and issue 29187 framed the congestion as speculative traffic displacing monetary transactions.^[6][7][8]

The policy war ended in stalemate — itself a verdict of sorts. Pull request 28408 was closed unmerged in January 2024, and the spam framing of issue 29187 never hardened into a default filter. No consensus rule changed, and no censorship was implemented. What survived was the dispute: what the default software of a permissionless network should decline to carry.^[6][7]

What shifted was the burden of debate. Before Ordinals, anyone writing non-monetary data onto Bitcoin had to justify it against community conventions. After block 840,000 recorded 37.62561499 BTC in fees, filtering advocates faced the reality that miners valued this revenue. Crucially, the filtering proposals concerned only default relay and mining policies in node software; consensus validity and the ability of individual miners to include any valid transaction remained entirely unchanged.^[6][7][8]

Rodarmor's legacy is a demonstration more than a product. Ordinals and Runes forked nothing, changed no consensus rule, and asked no one's leave. They numbered satoshis, reused the witness structure Bitcoin already had, and let the fee market respond. The fourth halving recorded 3,762,561,499 sats in total block fees, but a single block fee episode cannot settle long-term security-budget effects or determine the permanent composition of Bitcoin's traffic. That open question remains for the network's future to answer.^[2][5][8]