LibraryStorage and networks2018Design paperCorpus record
Arweave yellow paper
Arweave. Sam Williams and the Arweave team.
Arweave's protocol paper for permanent storage: miners store a recall block drawn from the history, an endowment is supposed to prepay storage, and the dataset is content-addressed. The yellow paper is the technical document. Marketing pages are not a substitute.
Arweave's yellow paper aims at permanent storage: a user pays once into an endowment, and miners must be able to recall a random piece of the history in order to produce the next block.
- Evidence
- Primary paper
- Re-measured
- No
- Assumptions
- 4
- Records linked
- 1
Code
Source, repository, and what is actually hosted
- Arweave yellow paperhttps://www.arweave.org/yellow-paper.pdf
- The source URL is not a GitHub repository. No code remote is invented.
- Full text is not mirrored. Public availability is not a licence to copy the file onto this site.
- Add to a reading list
01 Claim ledger
What the paper is allowed to say
Each row is a sentence already in the study. The status is the same on every row: a model claim, not a live measurement.
Claim 01 · Paper model
Ordinary storage deals expire
A contract for a year is a contract for a year. The paper asks what it would take for a link not to depend on next year's invoice.
Claim 02 · Paper model
Recall is the mining requirement
A block points at a historical block the miner must have. Producing the next block is evidence the miner still stores a slice of the dataset. The dataset is content-addressed.
Claim 03 · Paper model
The endowment is the economic bet
The user pays upfront. The endowment is supposed to pay miners for long enough that the declining cost of storage makes the remainder sufficient. That is an economic assumption, not a hash function.
Claim 04 · Paper model
Content addressing is the identifier
The address is a function of the bytes. A changed file is a different address. Permanence, if it works, is permanence of those exact bytes.
Claim 05 · Paper model
The yellow paper is the technical document
Marketing pages that say 'permanent' in a larger font are not a substitute. The endowment's assumptions live in the paper and in later economic notes, and they should be named.
02 Three cuts
Observation, model, falsifier
A desk does not stop at the summary. Each claim is cut three ways, using only this study's own assumptions and checks. Nothing here is a new figure.
01 Ordinary storage deals expire
Observation
What the study says
A contract for a year is a contract for a year. The paper asks what it would take for a link not to depend on next year's invoice.
Model
What has to hold
The endowment's assumptions about storage-cost decline and interest are roughly right, or users accept the residual risk.
Falsifier
What would retire it
What does the endowment assume about the future price of storage?
02 Recall is the mining requirement
Observation
What the study says
A block points at a historical block the miner must have. Producing the next block is evidence the miner still stores a slice of the dataset. The dataset is content-addressed.
Model
What has to hold
Miners store enough of the weave that a random recall is available to honest producers.
Falsifier
What would retire it
How is a miner required to show they still hold history?
03 The endowment is the economic bet
Observation
What the study says
The user pays upfront. The endowment is supposed to pay miners for long enough that the declining cost of storage makes the remainder sufficient. That is an economic assumption, not a hash function.
Model
What has to hold
The endowment's assumptions about storage-cost decline and interest are roughly right, or users accept the residual risk.
Falsifier
What would retire it
What does the endowment assume about the future price of storage?
04 Content addressing is the identifier
Observation
What the study says
The address is a function of the bytes. A changed file is a different address. Permanence, if it works, is permanence of those exact bytes.
Model
What has to hold
Readers can find a replica. Content addressing checks integrity once a replica is found. It does not search the earth.
Falsifier
What would retire it
Who will serve a replica to a reader, and how is that different from the mining proof?
05 The yellow paper is the technical document
Observation
What the study says
Marketing pages that say 'permanent' in a larger font are not a substitute. The endowment's assumptions live in the paper and in later economic notes, and they should be named.
Model
What has to hold
The endowment's assumptions about storage-cost decline and interest are roughly right, or users accept the residual risk.
Falsifier
What would retire it
Is 'permanent' being used as a marketing word or as a claim under these assumptions?
03 Sequence
One action, as an operating tape
- 01A user pays the fee and inserts data. The transaction's identifier is determined by the content.
- 02The data is woven into the block structure so later recall challenges can point at it.
- 03A miner who wants to produce a block fetches the recall block the protocol names and proves access.
- 04The reward for that block comes from the protocol's issuance and from the endowment's rules, not from a renewal invoice to the original user.
- 05A later reader who has the identifier can fetch the bytes from any node that still stores them, and check the hash.
04 Load-bearing
The argument, and where a pitch drops it
Endowment math can be wrong without the hash being wrong
The cut
If storage costs do not fall as assumed, or if demand to store data outruns the fund, the 'pay once' sentence fails while the recall mechanism still functions. A study has to keep the cryptographic mechanism and the economic wager in separate paragraphs.
Why it carries weight
If this cut is skipped, the paper's name is being used without the mechanism that makes the name mean anything.
Where it is dropped
Pay-once storage is only as permanent as the endowment and the miner set. The paper does not repeal the cost of disks.
Recall samples storage, it does not magically replicate everywhere
The cut
The protocol pressures miners to hold data that might be asked for. It does not promise that every node holds every byte. Retrieval depends on replication being wide enough in practice. The paper's incentive and the network's actual copies are both part of the claim.
Why it carries weight
If this cut is skipped, the paper's name is being used without the mechanism that makes the name mean anything.
Where it is dropped
Pay-once storage is only as permanent as the endowment and the miner set. The paper does not repeal the cost of disks.
Permanent does not mean lawful or wise
The cut
Bytes that cannot be taken down are a product decision with obvious harms. The paper is a mechanism for persistence. A deployment policy about what should be stored is a human layer on top, and pretending the mechanism contains that policy is false.
Why it carries weight
If this cut is skipped, the paper's name is being used without the mechanism that makes the name mean anything.
Where it is dropped
Pay-once storage is only as permanent as the endowment and the miner set. The paper does not repeal the cost of disks.
05 Register
What has to be true
Model · Not re-measured
The endowment's assumptions about storage-cost decline and interest are roughly right, or users accept the residual risk.
Cryptography · Not re-measured
Miners store enough of the weave that a random recall is available to honest producers.
Model · Not re-measured
Readers can find a replica. Content addressing checks integrity once a replica is found. It does not search the earth.
Data · Not re-measured
The data people pay to include is data they are willing to see persist.
06 Divergence
What happened after the paper
Arweave launched from this design and became a common place for applications to pin data they wanted addressable for a long time. The endowment's performance against its own assumptions is an empirical question that should be reported as such. The yellow paper is the recall-and-endowment construction, not a guarantee letter.
A later client, parameter set, or reward formula is a different object. Cite this paper for the mechanism. Cite a primary release for the network. This desk has not re-run the proof.
07 Pre-mortem
What to check before you use the idea
- 0 of 4 marked on this browser. A mark is a reading note, not a pass, a rating, or a recommendation.
08 Anatomy
The paper, in the order a builder needs
The problem it names
Ordinary storage deals expire. A link rots when whoever paid the invoice stops paying. Arweave's design asks a user to pay once, into an endowment, and asks miners to prove they are holding a randomly recalled piece of the global dataset.
What the design proposes
- A block includes a reference to a recall block. Mining requires access to that historical data.
- Content addressing means the identifier is a function of the bytes.
- The endowment is an economic object. Its adequacy is an assumption about future storage cost, and the paper has to be read that way.
How the mechanism is specified
- Wildfire, in the protocol's terms, is how nodes prefer peers who serve data. Replication is a social and economic outcome of that preference, not a certificate that every byte sits everywhere.
- A successful proof shows the miner could fetch the recall block. It does not show that every node holds every file.
- Permaweb applications sit on top of this dataset. They are not specified by the storage proof alone.
What this page does not treat as proven
- Pay-once storage is only as permanent as the endowment and the miner set. The paper does not repeal the cost of disks.
- The yellow paper version matters. Later updates change mechanics.
- This page does not describe token price.
Why the desk still reads it
When a venture says 'permanent record', Arweave is the design that at least names an endowment and a recall proof. Ask what happens if storage cost does not fall the way the endowment assumes. That question is the design review.
09 Lexicon
Terms, opened into the record
- Recall block
- A historical block a miner must access in order to produce a new block.
- Endowment
- The prepaid pool that is supposed to fund storage after the user has paid once.
- Content address
- An identifier computed from the bytes, so a different file cannot wear the same name.
- Weave
- The paper's structure of blocks and data, which recall challenges range over.
10 Repository
Every linked record on this page
Underlined words open a page that already exists: a concept, a protocol profile, a failure record, or another paper. If a word is not underlined, this desk does not have a record for it.
Concepts
This is Blockchain Lab's reading of a public design paper. It is not the paper, not a copy of it, and not an offer of tokens, equity, custody or a partnership. Later network behaviour can diverge from the text. Nothing here is investment, legal or technical advice.
Research status: Design paper. Last reviewed: 1 October 2026. This is a reading of a public paper, not investment, legal or security advice.