Filecoin: A Decentralized Storage Network

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.

1. Claim 01 · Paper model

   Storage and retrieval are different jobs

   A miner can prove a disk still holds a sector and still be a terrible place to download it from. The paper splits the markets so the prices can differ.

2. Claim 02 · Paper model

   Proof of replication is about a unique copy

   The proof is meant to show the miner encoded this client's data, not that they are claiming one shared copy as if it were many. That is the answer to a miner who outsources or deduplicates against the contract.

3. Claim 03 · Paper model

   Proof of spacetime stretches the claim

   A proof at one moment does not stop the miner deleting the file afterwards. Repeated proofs are the paper's way of charging for duration.

4. Claim 04 · Paper model

   Consensus is tied to useful storage

   The paper wants power in the network to follow committed storage rather than an empty hash race. Whether that weighting matches the later implementation is a separate check.

5. Claim 05 · Paper model

   The 2014 sketch in the historic library is not this paper

   An earlier Filecoin note exists on the old site. The 2017 document is the one with the market and the proofs described here. Do not cite them as the same text.

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.

1. 01 Storage and retrieval are different jobs

   Observation

   What the study says

   A miner can prove a disk still holds a sector and still be a terrible place to download it from. The paper splits the markets so the prices can differ.

   Model

   What has to hold

   The proof system matches the paper's replication and spacetime claims. Later proof changes need their own citation.

   Falsifier

   What would retire it

   Is the deal a recorded storage deal with a duration and collateral?

2. 02 Proof of replication is about a unique copy

   Observation

   What the study says

   The proof is meant to show the miner encoded this client's data, not that they are claiming one shared copy as if it were many. That is the answer to a miner who outsources or deduplicates against the contract.

   Model

   What has to hold

   The proof system matches the paper's replication and spacetime claims. Later proof changes need their own citation.

   Falsifier

   What would retire it

   Which proof is required, replication, spacetime, or a later replacement?

3. 03 Proof of spacetime stretches the claim

   Observation

   What the study says

   A proof at one moment does not stop the miner deleting the file afterwards. Repeated proofs are the paper's way of charging for duration.

   Model

   What has to hold

   The proof system matches the paper's replication and spacetime claims. Later proof changes need their own citation.

   Falsifier

   What would retire it

   Which proof is required, replication, spacetime, or a later replacement?

4. 04 Consensus is tied to useful storage

   Observation

   What the study says

   The paper wants power in the network to follow committed storage rather than an empty hash race. Whether that weighting matches the later implementation is a separate check.

   Model

   What has to hold

   The proof system matches the paper's replication and spacetime claims. Later proof changes need their own citation.

   Falsifier

   What would retire it

   Is the deal a recorded storage deal with a duration and collateral?

5. 05 The 2014 sketch in the historic library is not this paper

   Observation

   What the study says

   An earlier Filecoin note exists on the old site. The 2017 document is the one with the market and the proofs described here. Do not cite them as the same text.

   Model

   What has to hold

   The proof system matches the paper's replication and spacetime claims. Later proof changes need their own citation.

   Falsifier

   What would retire it

   Is the citation the 2017 paper or the historic 2014 sketch?

03 Sequence

One action, as an operating tape

1. 01A client and a miner agree a deal: price, duration, and the piece of data.
2. 02The miner seals a unique replica and posts a proof that the sealing was done.
3. 03Over the deal's life the miner posts proofs of spacetime. Missing a window is a fault.
4. 04The client, or another party, pays a retrieval provider to serve the bytes. That payment is not the storage deal.
5. 05If the miner fails the proofs, collateral in the paper's design is slashed and the client can treat the deal as broken.

04 Load-bearing

The argument, and where a pitch drops it

1. A proof of storage is not a content-delivery network

   The cut

   The paper is careful, and summaries are not. Retrieval has its own market because serving a file quickly is a bandwidth problem, not a sealing problem. A product that cites Filecoin for 'decentralised Dropbox' has to say who is paid to hand the bytes over this afternoon.

   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

   A proof of storage does not make retrieval fast, and the paper separates the markets for that reason.

2. Sealing is expensive on purpose

   The cut

   The encoding that makes a replica unique also makes it costly to pretend. That cost is why miners specialise, and why a casual laptop is not the miner the paper is about. Hardware specialisation is a consequence, not a betrayal.

   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

   Do not cite the 2014 sketch in the historic library as if it were this paper.

3. Deals are protocol objects

   The cut

   Price, duration and collateral are on the record. A handshake in a chat is not a Filecoin deal. The formality is what lets a proof be about a specific promise.

   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

   A proof of storage does not make retrieval fast, and the paper separates the markets for that reason.

05 Register

What has to be true

• Cryptography · Not re-measured

  The proof system matches the paper's replication and spacetime claims. Later proof changes need their own citation.

• Model · Not re-measured

  Collateral is large enough that deleting a paid sector is a loss.

• Model · Not re-measured

  Someone will retrieve. A stored file with no retrieval path is an archive, not a service.

• Model · Not re-measured

  Clients verify deals rather than trusting a miner's website.

06 Divergence

What happened after the paper

Filecoin mainnet launched after this paper, with proof constructions and deal-making that continued to evolve. The storage-versus-retrieval split is the lesson that still separates a serious deployment from a slogan. The historic 2014 PDF is a different, earlier sketch.

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

1. Kill test 1Is the deal a recorded storage deal with a duration and collateral?
2. Kill test 2Which proof is required, replication, spacetime, or a later replacement?
3. Kill test 3Who is paid to serve the file, as distinct from storing it?
4. Kill test 4Is the citation the 2017 paper or the historic 2014 sketch?
5. 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

09 Lexicon

Terms, opened into the record

Proof of replication

A proof that a miner encoded a unique copy of a client's data, rather than pointing at someone else's copy.

Proof of spacetime

A sequence of proofs that the replica persisted for the deal's duration.

Retrieval market

A separate payment for serving bytes. It is not implied by a successful storage proof.

Sealing

The expensive encoding step that creates the unique replica the proofs are about.

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.

11 Related

The next papers, as theses

• Arweave · 2018

  Arweave yellow paper

  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.

• Celestia · 2019

  LazyLedger: A Distributed Data Availability Ledger With Client-Side Smart Contracts

  Al-Bassam's LazyLedger paper, the research origin of Celestia. The base layer orders and makes data available. It does not execute application transitions. Clients check availability with sampling, and applications execute on their own.

• Golem · 2016

  The Golem Project: crowdfunding white paper

  Golem Factory's 2016 paper for a marketplace of spare computer power. Requestors split tasks. Providers run them. A reputation and payment layer is supposed to make the exchange work without a single render farm. The paper is explicit that it was also a crowdfunding document. This page uses the technical design and ignores the sale.