When a dev tells you that something is weird and hard and they have to change behavior that users rely on, the correct response is somewhere between "Cool story, bro." and "Wah, wah, wah, are the bits being mean to you?". Developers serve users, not the other way around.

For this and other reasons, the ord developers recommit to the stability and predictability of inscription numbers, and will not pursue renumbering.

As someone who people in the ordinals community mysteriously pay a lot of attention to, I see myself as having two distinct roles with regards to proposals such as the great renumbering.

One is to have and propose things that I think are good ideas. This will continue, and I don't see any particular reason to self-censor. In other words, the unfettered blog posts will continue. The community needs to understand that these are proposals only, and that I have had and will continue to have wacky ideas.

The other role is as one of the people who helps make decisions about what to implement in ord, the ordinals reference implementation. Raph, as lead maintainer, has the final say, but he hasn't stopped listening to my wacky ideas, so I still have some influence. Those decisions must be made with respect to the userbase of ord and the greater ordinals community.

When I first decided to propose renumbering, I knew that it had to meet a very high bar. Changes to ord and to ordinals must be Pareto improvements, changes to the status quo that leave everyone better off, or extremely close to it. Changes cannot create winners and losers, even if the benefit to the winners might be greater than the harm to the losers.

After reading discussion on GitHub and Twitter, lurking in many a Twitter space, and discussing it with many people, it's clear that the great renumbering does not meet that very high bar, and is better off consigned to the dustbin of history.

A couple of the arguments that I personally found most persuasive:

• Any public discussion will consist only of the most engaged users. Many less engaged users simply won't be aware that a discussion is taking place, or might not participate, regardless of their views. These voices simply won't be heard. This includes people who aren't terminally online, many people who don't speak English as their native language, and people who just don't particularly like arguing online. This means that the bias must be strong towards the status quo,

• A programmer's job, fundamentally, is to drag his (or her!) balls through miles of broken glass if it benefits the user. Unless the status quo is so complex and unmentionable that it is untenable, disruptive changes must be avoided. As a spirited GitHub commenter put it, "Code and protocols get messy as they age. That's reality. Purity is for virgins."

What does this mean, practically?

• ord will commit to the stability of positive inscription numbers. Negative, or cursed, inscription numbers are subject to change at any time, if new cursed inscriptions are recognized.

• At some as-of-yet unannounced block height in the future, new inscriptions which would previously have been cursed will be blessed, and will receive positive inscription numbers as part of the main sequence of inscription numbers. Hopefully, this kind of community wide coordination is not a common occurrence, but it may happen again from time to time, if it is advantageous to recognize new kinds of inscriptions which previously would have been invalid.-

• The commitment to the stability of inscription numbers does not preclude the fixing of bona fide bugs. If there is a legitimate bug, and the only reasonable way to fix it proves disruptive to inscription numbers, we will fix it.

Inscriptions are both a technical artifact, and an artistic technology. Technical considerations must contend on equal footing with aesthetic considerations.

Inscription numbers are here to stay.

Inscription numbers are numbers assigned to inscriptions in the order in which they are created, starting at zero for the genesis inscription.

When inscription numbers were originally added to ord, the Ordinals wallet and explorer that powers ordinals.com, they were intended to be stable and never change.

However, now that we have more experience with the protocol, that seems less tenable and has undesirable consequences, as maintaining stable inscription numbers is a challenge in the face of changes to the inscription protocol.

Consider a simple case, an update that allows multiple inscriptions to be created in a single transaction.

The inscription numbers assigned by an implementation that recognizes multiple inscriptions in a single transaction would differ from those assigned by an implementation that does not.

The former implementation would see T1, creating two new inscriptions, and T2, creating a single new inscription, and assign inscription numbers N, N+1, and N+2, the latter implementation would assign inscription number N to the first and only inscription it recognized in T1, and N+1 to the inscription in T2.

There are many such updates that we would like to make or have made which would introduce discrepancies like these, including:

• Multiple inscriptions per transaction
• Inscriptions in reveal inputs other than the first
• Inscriptions with new even fields
• Multiple inscriptions on the same sat

We've been able to make these changes and keep inscription numbers stable using what I've now come to think of as a regrettable hack: cursed inscriptions. Whenever ord indexes an inscription which would not be recognized by an earlier version, it assigns a negative inscription number. This keeps old inscription numbers stable, while still recognizing new inscriptions.

Cursed inscriptions and negative inscriptions numbers have a number of downsides:

• An inscription number now does not tell you anything about the order in which the inscription was made.
• The logic required to keep track of which inscriptions are cursed is a source of bugs and complexity.
• "Blessing" cursed inscription types, i.e., collectively deciding that after a certain block height, certain cursed inscription types will no longer be assigned negative numbers, and be assigned positive numbers instead, requires coordination.
• Cursed inscription numbers are permanently unstable, so a substantial number of inscription numbers are already unstable, even under the status quo.

In light of the above, I propose that we make inscription numbers permanently unstable, and bless all cursed inscriptions, both retroactively and on an ongoing basis. Cursed inscription numbers would be folded into the main sequence, and, going forward, inscription numbers should not be used in URLs, which is already the case for ord, and inscription numbers would be de-emphasized on /inscription pages.

This would substantially simplify the ord codebase, make it easier to produce an implementation that assigns the same sequence numbers, and make future protocol changes easier.

ord version 0.4.0 has been released. Inscriptions are finally ready for Bitcoin mainnet.

Inscriptions

Inscriptions are digital artifacts native to the Bitcoin blockchain. They are created by inscribing sats with content using ord, and can be viewed with the ordinals explorer. They do not require a separate token, a side chain, or changing Bitcoin.

Inscriptions are created by including content, like an image, text, SVG, or HTML, in an inscription transaction. The content is included in the transaction witness, which normally contains signatures and other data proving that a transaction is authorized.

Along with the content, the inscription transaction contains a content type, also known as a MIME type, identifying the type of content to be inscribed.

When mined, the inscription is made on the first sat of the first output of the transaction, permanently and inexorably marking it, distinguishing it from its fellows. It is no longer just a sat, it is an intertwined component of the long and confusing tale that is human art and culture.

Using ordinal theory, the unspent output containing an inscribed sat can be found, and its movements and ownership tracked across time and transactions, allowing inscriptions to traded, gifted, bought, and sold.

This allows inscriptions quite native to Bitcoin. They can be sent to normal bitcoin addresses, in normal bitcoin transactions, and benefit from timelocks, multisig, and all the rest of Bitcoin's infrastructure. To avoid losing them, a wallet that holds inscriptions must perform sat control, the sizing and alignment of transaction inputs and outputs that controls the destination of individual sats, but aside from that, transactions that transfer inscriptions are quite mundane.

Digital Artifacts

Inscriptions are digital artifacts, and digital artifacts are NFTs, but not all NFTs are digital artifacts. Digital artifacts are NFTs held to a higher standard, closer to their ideal. For an NFT to be a digital artifact, it must be decentralized, immutable, on-chain, and unrestricted. The vast majority of NFTs are not digital artifacts. Their content is stored off-chain and can be lost, they are on centralized chains, and they have back-door admin keys. What's worse, because they are smart contracts, they must be audited on a case-by-case basis to determine their properties.

Inscriptions are unplagued by such flaws. Inscriptions are immutable and on-chain, on the oldest, most decentralized, most secure blockchain in the world. They are not smart contracts, and do not need to be examined individually to determine their properties. They are true digital artifacts.

ord 0.4.0

ord is an open-source binary written in Rust, and developed on GitHub. It implements an ordinal wallet, which can create and transfer inscriptions, and a block explorer. There are public mainnet, signet, and testnet instances.

ord is experimental software, and comes with no warranty or guarantees. ord 0.4.0, the latest release, has been tested carefully, and can now be used to make inscriptions on mainnet, and that those inscriptions will not break due to a future protocol change.

Three mechanisms exist to introduce opt-in changes to the protocol, without breaking existing inscriptions: Versioning, optional fields, and mandatory fields.

Inscriptions can contain a version field. The inscription parser will ignore inscriptions with an unrecognized version. This allows introducing fundamental changes to inscriptions, without disrupting existing inscriptions.

Additionally, fields can be marked as optional or mandatory, which determines how an inscription parser treats an unrecognized field. Unrecognized optional fields are ignored but the inscription is normally, while unrecognized mandatory fields render the whole inscription invalid.

Individual features that can be safely ignored if unsupported can be introduced as optional fields, while features that must be supported in order in order to understand an inscription can be introduced as mandatory fields.

Inscriptions are not finished, but this flexibility gives us confidence that future improvements can be made opt-in and non-disruptive.

Let markets and bazaars where rare sats and inscriptions are traded grow and flourish, and may their wares never crack or vanish.

What's missing?

Inscriptions have many unique benefits and features, but they have not yet reached feature parity with other NFT implementations.

Two key features are missing: provenance and decentralized markets.

Provenance is the ability to determine the author of an inscription, or its membership in a set of inscriptions all created by the same person. We have a design that accomplishes this, and its implementation is tracked in issue #783. A transaction creating a new inscription, X, may include an existing inscription, P, in its inputs, which is returned back to the owner in its outputs. Since only the owner of P could have made this transaction, this identifies P as the parent of X. This mechanism is flexible, and can be used to identify an inscription as being created by an individual, or to identify an inscription as being a member of a larger collection. Furthermore, this mechanism is recursive. You can create in inscription that represents your identity, and use that inscription as the parent of an inscription that itself is used as the parent inscription for a collection.

For inscriptions be valuable, there must be venues where they can be bought and sold. We have a sketch for decentralized and trustless offers to buy and sell using partially-signed transactions, and its implementation is tracked in issue #802. The owner of an inscription may offer it for sale by publishing a transaction that includes it as an input, and containing an output that pays to them the sale price. Any third party can take such a transaction, add their own input of at least the sale price, add an output sending the inscription to themselves, and finalize it by broadcasting it to be mined. Offers to buy are similar.

What's next?

Inscriptions have been designed to be native to the web. Inscriptions are byte strings, identified with a content type, and so can be displayed in a browser. HTML, CSS, JavaScript, SVG, MP3, PNG, and JPEG, are supported by the ordinals explorer.

Inscription content is sandboxed so that it cannot make outgoing web requests. However, in the future, this sandboxing will be relaxed to allow inscriptions to use the content of other inscriptions. This will allow for the development of an a modular, on-chain ecosystem of remixing and composition. This is tracked in issue #1082.

Inscriptions are unnamed and untitled, but we hope to give artificers the ability to give their inscriptions globally unique, human-readable names. This is tracked in issue #794.

The future of inscriptions is bright. We hope not only to reach feature parity with other NFT implementations, but to surpass them.

We just released ord version 0.1.0!

ord is an ordinal number wallet and the index that serves ordinals.com.

This release of ord is by no means complete, but it now supports making ordinal-aware transactions on mainnet using the ord wallet send command thanks to @raphjaph, and boasts much faster indexing thanks to @callmeier.

ord wallet send

The ord wallet commands interact with an existing Bitcoin Core wallet using Core's RPC interface, and this release features a new send subcommand that constructs an ordinal-aware transaction to send a particular ORDINAL to a recipient's ADDRESS:

ord wallet send ORDINAL ADDRESS

We write extensive tests alongside every feature, but bugs can always slip through, and the project as a whole is still immature, so restrictions apply when using ord wallet send with Mainnet wallets.

ord wallet send will refuse to interact with a Mainnet wallet with a name other than ord, which does not start with ord-, to encourage users to explicitly mark wallets as being intended for use with ord.

Using bitcoin-cli wallet commands with an ord wallet risks spending your rarest and most exotic sats, and using ord wallet commands with your main wallet risks doxing your main stash if you assocaite a spicy ordinal with your identity.

Index Optimization

ord must build and maintain an index that maps transaction outputs to ordinal ranges. ord 0.1.0 now exploits the fact that many UTXOs are spent soon after they're created, and uses an in-memory cache to avoid huge numbers of database insertions and retrievals. ord uses the redb database, which is already quite fast, but using an in-memory cache avoids reads from and writes to redb's B-tree, which are more expensive than the in-memory cache's hashmap.

Backwards Compatibility

ord is still alpha-quality software, so breaking changes are inevitable, but starting with version 0.1.0, we'll indicate when breaking changes are included in a release by bumping the minor version number (0.x.0).

Try it out!

Check out ord 0.1.0 and let us know what you think! ord is CC-0-licensed open-source software developed on GitHub.

Raph and I have been working hard on ord, the ordinal block explorer and wallet, and have just finished the initial version of the ord wallet send ORDINAL ADDRESS command, which makes a transaction sending ORDINAL to ADDRESS.

Ordinal-aware Bitcoin transactions are rather tricky, and must attend to a number of considerations:

• Rare ordinals shouldn't accidentally be lost to the recipient or the fee.
• The ordinal being sent should be aligned to the beginning of the output going to the recipient.
• Additional postage should be added so that the ordinal can be sent onward without needing to merge in additional UTXOs.
• Excess sats should be stripped from the recipient's output.
• Inputs and outputs must be aligned so that the ordinal being sent actual makes it to the recipient.

This requires a great deal of stressful fiddling with transaction input and output size and order, and a great deal of checks for edge cases. You can see all the gory details in the transaction builder.

All that fiddling has paid off, and today we made three transactions on signet:

• This transaction is simple send of mvzwjdjcofe. The first output is change, added to ensure that mvzwjdjcofe is the first ordinal in the output sent to the recipient:

  ord wallet send mvzwjdjcofe tb1qaykcryeds0kmz3azv5tn46dyt3wu5l5qc3djz0

• This transaction sends ordinal nphhwymyrjx to tb1qew25cz5603xqzkw66m5yq9ruflkkpgjyprcdc5:

  ord wallet send nphhwymyrjx tb1qew25cz5603xqzkw66m5yq9ruflkkpgjyprcdc5

  nphhwymyrjx is first in the input, so no alignment output is needed to make it first in the recipient's output.

  However, the input is quite large, so an additional output is added that brings the size of the recipient's output down to 10,000 sats. We could reduce the size of recipient's output to the dust limit, however in that case, UTXOs would need to be merged in to pay for fees when sending the ordinal onwards.

• This transaction sends ordinal 139761296637063 to tb1q02yk08hkp9pkyhwdsdfd0ra437tjcdnrjs0sxv:

  ord wallet send 139761296637063 tb1q02yk08hkp9pkyhwdsdfd0ra437tjcdnrjs0sxv

  There's a lot going on here! Let's break down the inputs and outputs:

  Inputs:

  1. Contains 139761296637063 at offset 7,500.
  2. Needed to pay for all outputs

  Outputs:

  1. Change containing 7,500 sats to make 139761296637063 the first ordinal of the recipient's output.
  2. The recipient's output containing 139761296637063 at offset 0. Worth 10,000, so that the recipient can forward 139761296637063 without needing to merge in additional UTXOs.
  3. Change containing 16,006 sats, to reduce the recipient's output to 10,000 sats.

ord wallet send will hopefully be much easier to use compared to using bitcoin-cli to manually construct ordinal-aware transactions. Give it a shot and let us know what you think!

I've been working on a numbering scheme for satoshis that allows tracking and transferring individual sats. These numbers are called ordinals, and constitute a numeric namespace for Bitcoin. Satoshis are numbered in the order in which they're mined, and transferred from transaction inputs to transaction outputs in first-in-first-out order. More details are available in the BIP.

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

Ordinals can be represented in a few ways:

With raw notation, like so 1905530482684727°. The number is the ordinal number, and the "°" is the Romance language ordinal symbol.

With decimal notation, like so 738848.482684727°. The first number is the block height, and the second is the index of the ordinal within the block.

With degree notation, like so 0°108848′992″482684727‴. We'll get to that in a moment.

A block explorer is available at ordinals.com. You can explore recent blocks, and look up ordinals by number, decimal, degree, or name.

Arbitrary assets, such as NFTs, security tokens, accounts, or stablecoins can be attached to Ordinals.

Ordinals is an open-source project, developed on GitHub. The project consists of a BIP describing the ordinal scheme, an index that communicates with a Bitcoin Core node to track the location of all ordinals, a wallet that allows making ordinal-aware transactions, a block explorer for interactive exploration of the blockchain, and functionality for minting ordinal NFTs.

Rarity

Since ordinals can be tracked and transferred, people will naturally want to collect them. Ordinal theorists can decide for themselves which sats are rare and desirable, but I wanted to provide some hints.

Bitcoin has periodic events, some frequent, some more uncommon, and these naturally lend themselves to a system of rarity. These periodic events are:

• Blocks: A new block is mined approximately every 10 minutes, from now until the end of time.
• Difficulty adjustments: Every 2016 blocks, or approximately every two weeks, the Bitcoin network responds to changes in hashrate by adjusting the difficulty target which blocks must meet in order to be accepted.
• Halvings: Every 210,000 blocks, or roughly every four years, the amount of new sats created in every block is cut in half.
• Cycles: Every six halvings, something magical happens: the halving and the difficulty adjustment coincide. This is called a conjunction, and the time period between conjunctions a cycle. A conjunction occurs roughly every 24 years. The first conjunction should happen some time in 2032.

This gives us the following rarity levels:

• common: Any sat that is not the first sat of its block
• uncommon: The first sat of each block
• rare: The first sat of each difficulty adjustment period
• epic: The first sat of each halving epoch
• legendary: The first sat of each cycle
• mythic: The first sat of the genesis block

Which brings us to degree notation, which unambiguously represents an ordinal in a way that makes rarity easy to see at a glance:

A°B′C″D‴
│ │ │ ╰─ Index of sat in the block
│ │ ╰─── Index of block in difficulty adjustment period
│ ╰───── Index of block in halving epoch
╰─────── Cycle, numbered starting from 0

Ordinal theorists often use the terms "hour", "minute", "second", and "third" for A, B, C, and D, respectively.

Now for some examples. This ordinal is common:

1°1′1″1‴
│ │ │ ╰─ Not first sat in block
│ │ ╰─── Not first block in difficutly adjustment period
│ ╰───── Not first block in halving epoch
╰─────── Second cycle

This ordinal is uncommon:

1°1′1″0‴
│ │ │ ╰─ First sat in block
│ │ ╰─── Not first block in difficutly adjustment period
│ ╰───── Not first block in halving epoch
╰─────── Second cycle

This ordinal is rare:

1°1′0″0‴
│ │ │ ╰─ First sat in block
│ │ ╰─── First block in difficulty adjustment period
│ ╰───── Not the first block in halving epoch
╰─────── Second cycle

This ordinal is epic:

1°0′1″0‴
│ │ │ ╰─ First sat in block
│ │ ╰─── Not first block in difficulty adjustment period
│ ╰───── First block in halving epoch
╰─────── Second cycle

This ordinal is legendary:

1°0′0″0‴
│ │ │ ╰─ First sat in block
│ │ ╰─── First block in difficulty adjustment period
│ ╰───── First block in halving epoch
╰─────── Second cycle

And this ordinal is mythic:

0°0′0″0‴
│ │ │ ╰─ First sat in block
│ │ ╰─── First block in difficulty adjustment period
│ ╰───── First block in halving epoch
╰─────── First cycle

If the block offset is zero, it may be omitted. This is the uncommon ordinal from above:

1°1′1″
│ │ ╰─ Not first block in difficutly adjustment period
│ ╰─── Not first block in halving epoch
╰───── Second cycle

Supply

Total Supply

• common: 2.1 quadrillion
• uncommon: 6,929,999
• rare: 3437
• epic: 32
• legendary: 5
• mythic: 1

Current Supply

• common: 1.9 quadrillion
• uncommon: 745,855
• rare: 369
• epic: 3
• legendary: 0
• mythic: 1

At the moment, even uncommon ordinals are quite rare. As of this writing, 745,855 uncommon ordinals have been mined - one per 25.6 bitcoin in circulation.

Names

Each ordinal has a name, consisting of the letters A through Z, that get shorter the larger the ordinal is. They could start short and get longer, but then all the good, short names would be trapped in the unspendable genesis block.

As an example, 1905530482684727°'s name is "iaiufjszmoba". The name of the last ordinal to be mined is "a". Every combination of 10 characters or less is out there, or will be out there, some day.

Exotics

Ordinals may be prized for reasons other than their name or rarity. This might be due to a quality of the number itself, like having an integer square or cube root. Or it might be due to a connection to a historical event, such as ordinals from block 477,120, the block in which SegWit activated, or ordinal 2099999997689999°, the last ordinal that will ever be mined.

Such ordinals are termed "exotic". Which ordinals are exotic and what makes them so is subjective. Ordinal theorists are are encouraged to seek out exotics based on criteria of their own devising.

Archaeology

A lively community of archaeologists devoted to cataloging and collecting early NFTs has sprung up. Here's a great summary of historical NFTs by Chainleft.

A commonly accepted cut-off for early NFTs is March 19th, 2018, the date the first ERC-721 contract, SU SQUARES, was deployed on Ethereum.

Whether or not ordinals are of interest to NFT archaeologists is an open question! In one sense, ordinals were created in early 2022, when I finalized the Ordinals specification. In this sense, they are not of historical interest.

In another sense though, ordinals were in fact created by Satoshi Nakamoto in 2009 when he mined the Bitcoin genesis block. In this sense, ordinals, and especially early ordinals, are certainly of historical interest.

I personally favor the latter view. This is not least because the ordinals were independently discovered on at least two separate occasions, long before the era of modern NFTs began.

On August 21st, 2012, Charlie Lee posted a proposal to add proof-of-stake to Bitcoin to the Bitocin Talk forum. This wasn't an asset scheme, but did use the ordinal algorithm, and was implemented but never deployed.

On October 8th, 2012, jl2012 posted a scheme to the the same forum which uses decimal notation and has all the important properties of ordinals. The scheme was discussed but never implemented.

These independent inventions of ordinals indicate in some way that ordinals were discovered, or rediscovered, and not invented. The ordinals are an inevitability of the mathematics of Bitcoin, stemming not from their modern documentation, but from their ancient genesis. They are the culmination of a sequence of events set in motion with the mining of the first block, so many years ago.