LibraryData and agents2020Design paperCorpus record
The Graph: a decentralised query protocol for blockchains
The Graph. Yaniv Tal, Brandon Ramirez and the Graph team.
The Graph's protocol paper for indexing chain data. Indexers stake on serving a subgraph. Curators signal which subgraphs matter. Consumers pay for queries. The problem is read access, not consensus.
The Graph's paper describes a market for indexing chain data: indexers stake, curators signal which subgraphs matter, and consumers pay for queries the base chain is bad at serving.
- Evidence
- Primary paper
- Re-measured
- No
- Assumptions
- 4
- Records linked
- 1
Code
Source, repository, and what is actually hosted
- Graph research papershttps://github.com/graphprotocol/research
- graphprotocol/researchRepository taken from the source URL. No other study cites this repository. This is not an audit.
- 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
Chains are poor databases
A contract stores the state it needs to be correct, not the indexes a website needs to be fast. Reading a rich query from a node is a service, and someone runs it.
Claim 02 · Paper model
A subgraph is a saved transformation
Developers describe how to turn events into entities. An indexer runs that mapping and serves queries. The paper's object is this hosted, checkable transformation.
Claim 03 · Paper model
Stake is how a bad index is supposed to hurt
Indexers lock tokens. The paper's dispute process is meant to slash an indexer who served a result that does not follow from the chain and the mapping.
Claim 04 · Paper model
Curators are a discovery layer
They signal on subgraphs, and the signal is supposed to direct indexer attention. It is also a market that can be gamed, and the paper should be read with that in mind.
Claim 05 · Paper model
The useful tension is verification versus convenience
A consumer can often trust a well-known indexer the way they trust any API. The paper's extra claim is that a fisherman can prove a lie. If disputes never happen, the system is an API with a token.
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 Chains are poor databases
Observation
What the study says
A contract stores the state it needs to be correct, not the indexes a website needs to be fast. Reading a rich query from a node is a service, and someone runs it.
Model
What has to hold
Query receipts are retained by anyone who might dispute.
Falsifier
What would retire it
Is this query served by a staked indexer or by one hosted operator?
02 A subgraph is a saved transformation
Observation
What the study says
Developers describe how to turn events into entities. An indexer runs that mapping and serves queries. The paper's object is this hosted, checkable transformation.
Model
What has to hold
At least one indexer serves the subgraph. A perfect manifest with nobody running it is an empty API.
Falsifier
What would retire it
Is this query served by a staked indexer or by one hosted operator?
03 Stake is how a bad index is supposed to hurt
Observation
What the study says
Indexers lock tokens. The paper's dispute process is meant to slash an indexer who served a result that does not follow from the chain and the mapping.
Model
What has to hold
The chain data the mapping reads is available to indexers and to challengers.
Falsifier
What would retire it
Who wrote the mapping, and is a wrong mapping distinguishable from a wrong index?
04 Curators are a discovery layer
Observation
What the study says
They signal on subgraphs, and the signal is supposed to direct indexer attention. It is also a market that can be gamed, and the paper should be read with that in mind.
Model
What has to hold
At least one indexer serves the subgraph. A perfect manifest with nobody running it is an empty API.
Falsifier
What would retire it
Is this query served by a staked indexer or by one hosted operator?
05 The useful tension is verification versus convenience
Observation
What the study says
A consumer can often trust a well-known indexer the way they trust any API. The paper's extra claim is that a fisherman can prove a lie. If disputes never happen, the system is an API with a token.
Model
What has to hold
At least one indexer serves the subgraph. A perfect manifest with nobody running it is an empty API.
Falsifier
What would retire it
Is this query served by a staked indexer or by one hosted operator?
03 Sequence
One action, as an operating tape
- 01A developer publishes a subgraph manifest: which contracts, which events, which schema.
- 02Curators signal on it, locking tokens as a statement that the subgraph is worth serving.
- 03An indexer runs the mappings against chain history and keeps the store up to date as new blocks arrive.
- 04A consumer, or a gateway in front of them, pays a query fee and receives a signed result.
- 05If the result does not match the mapping, a challenger can dispute it under the paper's rules, with the indexer's stake at issue.
04 Load-bearing
The argument, and where a pitch drops it
Indexing is not consensus
The cut
Two honest indexers running the same mapping against the same chain should agree. A wrong mapping produces a consistently wrong API, and no amount of stake fixes a bad schema. The paper secures execution of a published mapping. It does not decide whether the mapping is the question the user meant to ask.
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
An index is not a source of truth above the chain. If it disagrees with the chain, the chain wins.
Disputes need a referee and a window
The cut
A query result that can be proven false only if someone archives the reply and posts a proof before a deadline is a specific security model. Consumers who do not keep the signed response cannot later complain. The paper's cryptoeconomics begins after the receipt is saved.
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
The paper does not make every possible query cheap.
Curation is advertising with a loss
The cut
Signalling directs indexers and can earn a share of fees. It can also be used to push a useless subgraph onto the market. The paper does not contain a theory of which data ought to exist. It contains a market.
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
The paper does not make every possible query cheap.
05 Register
What has to be true
Data · Not re-measured
The chain data the mapping reads is available to indexers and to challengers.
Model · Not re-measured
At least one indexer serves the subgraph. A perfect manifest with nobody running it is an empty API.
Model · Not re-measured
Query receipts are retained by anyone who might dispute.
Stake and threshold · Not re-measured
The dispute window is long enough to matter and short enough that indexers will lock stake.
06 Divergence
What happened after the paper
The Graph shipped a hosted service first and a decentralised network after, with gateways, curation and dispute parameters that evolved. Many applications still used the hosted path for a long time. The paper describes the staked indexer market. Whether a given query is served by that market or by a single operator is a fact about the deployment.
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
A contract write is verifiable from the chain. A rich read — every trade in a pool, every position in a market — is a query over history that a phone should not reconstruct alone. Centralised indexers fill the gap and become silent dependencies.
What the design proposes
- A subgraph schema defines how chain events become queryable entities.
- Indexers choose subgraphs, stake, and serve queries. Their responses can be checked against the stated indexing rules.
- Curators and delegators, in the paper, allocate attention and stake. They are economic roles, not honourifics.
How the mechanism is specified
- Payment for queries is what makes indexing a job rather than a volunteer mirror.
- Disputes and fisherman-style checks are the paper's answer to a wrong index, and they only work if a wrong index is detectable from public data.
- Hosted services and the later decentralised network should not be cited as if they were the same deployment.
What this page does not treat as proven
- An index is not a source of truth above the chain. If it disagrees with the chain, the chain wins.
- The paper does not make every possible query cheap.
- We do not describe token rewards as income.
Why the desk still reads it
Agent and analytics ventures usually hide an indexer in the architecture diagram. The Graph is the paper that pulls it into the open and gives it a stake, a schema and a dispute path.
09 Lexicon
Terms, opened into the record
- Subgraph
- A published mapping from chain events to the entities a query will return.
- Indexer
- An operator who runs the mapping, serves queries, and locks stake against being wrong.
- Curator
- A party who signals that a subgraph is worth indexing, with tokens at stake in that signal.
- Fisherman
- A challenger who disputes a query result that does not follow from the chain and the mapping.
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.