Bitcoin Ordinals is a protocol for the Bitcoin blockchain that gives every individual satoshi, the smallest unit of bitcoin, a serial number and lets users attach arbitrary data to it. The attached data is called an inscription and can be text, an image, audio or code. Because a numbered satoshi can be tracked and traded on its own, an inscribed satoshi behaves much like a non-fungible token, but without a separate token contract or a second chain.
The protocol was released in January 2023 by software developer Casey Rodarmor, together with an open-source program called ord that runs alongside a Bitcoin Core node. It requires no change to Bitcoin's consensus rules: it is a convention for reading data that the network already accepts. Ordinals quickly became the basis for most activity in Bitcoin NFTs and for experimental token schemes such as BRC-20, and it set off a long debate about what Bitcoin block space should be used for.
Ordinal theory and satoshi numbering
Bitcoin itself does not track individual satoshis. Balances are held as unspent transaction outputs, each containing some number of satoshis, and the network treats every satoshi as interchangeable. Ordinal theory adds an off-chain numbering scheme on top: satoshis are numbered in the order they were created by the block reward, starting from zero, and a first-in, first-out rule decides which numbered satoshis move to which output when a transaction is spent.
Anyone running the same software reaches the same answer about where a given satoshi sits, so the numbering works as a shared interpretation rather than a rule enforced by miners. Collectors also assign informal rarity labels, for example to the first satoshi of each block or of each halving period. These labels are conventions of the ordinals community and carry no meaning for the Bitcoin protocol.
How inscriptions work
An inscription is created in a two-step process. A first transaction commits to a script containing the content, and a second transaction reveals that script when it spends the output. The content sits inside the witness part of the transaction, wrapped in a so-called envelope that the Bitcoin script never executes. The inscription is then assigned to the first satoshi of the revealing transaction's input and follows that satoshi from then on.
Two earlier upgrades made this practical:
- Segregated Witness. SegWit, activated in 2017, moved signature data into a separate witness section and counts witness bytes at a discount when measuring block size, so large inscriptions cost less than they otherwise would.
- Taproot. Taproot, activated in November 2021, introduced script-path spending that allows large scripts in the witness, which is where the inscription envelope is placed.
Inscriptions are stored in full on chain, unlike many NFTs on other networks whose images live on outside servers and are only referenced by a link.
BRC-20 and other token schemes
In March 2023 a pseudonymous developer proposed BRC-20, which uses inscriptions of small JSON text to deploy, mint and transfer fungible tokens. Balances are not enforced by Bitcoin; they are calculated by indexers that read the inscriptions in order and apply the BRC-20 rules. This makes the system dependent on indexers agreeing with each other. Rodarmor later proposed a separate fungible-token protocol called Runes, which stores token data in a different part of the transaction; it is distinct from ordinals and inscriptions even though it comes from the same circle of developers.
The ordinals model contrasts with token standards on the Ethereum blockchain such as ERC-721, where a smart contract records who owns each token. On Bitcoin, ownership of an inscription is simply ownership of the output holding the inscribed satoshi, which means a wallet that does not understand ordinals can accidentally spend an inscribed satoshi as ordinary fee money.
Fees, block space and criticism
Waves of inscription activity in 2023 filled the mempool and pushed up the transaction fees paid by all users. Supporters argue that inscriptions are valid transactions that pay market fees, bring miners extra revenue as the block subsidy falls, and widen the uses of the network. Critics see them as spam that crowds out monetary payments, enlarges the data every full node must store, and exploits a witness discount that was designed for signatures rather than images.
The debate has also touched node software. Some developers have proposed or released filters that refuse to relay inscription-style transactions, while others argue that such filters are ineffective because miners can include the transactions directly. The protocol's critics and supporters agree on one point: because inscriptions are written into blocks, they cannot be removed once confirmed.
Risks for holders
- Wallet handling. Inscribed satoshis can be lost by sending them with ordinary bitcoin or by spending them as fees, so dedicated wallets track them separately.
- Indexer dependence. Inscription numbers and BRC-20 balances are computed by off-chain software, and differences between indexers have caused disputes.
- Liquidity and pricing. Trading takes place on specialist marketplaces with thin activity, and valuations rest on collector demand rather than any income stream.
Ref 1070001308