LibraryScaling2018Design paperCorpus record
Arbitrum: Scalable, private smart contracts
Arbitrum. Harry Kalodner, Steven Goldfeder, Xiaoqi Chen, S. Matthew Weinberg, Edward W. Felten.
The 2018 USENIX paper on Arbitrum: a verifier that checks a manager's execution of a virtual machine by bisecting disputes, instead of re-executing every instruction. Offchain Labs' later Nitro stack is a descendant, not this paper line for line.
The Arbitrum paper describes a dispute protocol in which a manager need not re-execute a contract. If someone cheats, a bisection game narrows the lie to one step a small checker can verify.
- Evidence
- Primary paper
- Re-measured
- No
- Assumptions
- 4
- Records linked
- 6
Code
Source, repository, and what is actually hosted
- USENIX Security 2018https://www.usenix.org/system/files/conference/usenixsecurity18/sec18-kalodner.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
Execution is someone else's computer
Managers assert a result. Honest parties re-execute only if they disagree. The common case is one assertion and silence.
Claim 02 · Paper model
Bisection finds the step
A dispute splits the computation in half, again and again, until a single instruction is left. The contract checks that instruction, not the whole programme.
Claim 03 · Paper model
Any one honest challenger is the model
Safety requires that someone honest and interested challenges a false assertion before the deadline. It does not require a global vote on every step.
Claim 04 · Paper model
Privacy in the 2018 paper is a separate claim
The original system discusses virtual machines whose state need not be public to every verifier. The later Ethereum rollup also called Arbitrum made different choices about publishing data. Do not merge them.
Claim 05 · Paper model
The deadline is a security parameter
A longer dispute window gives honest challengers more time and users less speed. The paper's security is this trade, written as a timeout.
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 Execution is someone else's computer
Observation
What the study says
Managers assert a result. Honest parties re-execute only if they disagree. The common case is one assertion and silence.
Model
What has to hold
At least one honest party will see a false assertion and can get a challenge transaction included in time.
Falsifier
What would retire it
Who is allowed to assert, and who is allowed to challenge?
02 Bisection finds the step
Observation
What the study says
A dispute splits the computation in half, again and again, until a single instruction is left. The contract checks that instruction, not the whole programme.
Model
What has to hold
Stakes are large enough that grieving the dispute game is not a profitable way to stall.
Falsifier
What would retire it
How long is the window, and can a challenge transaction be censored for that long?
03 Any one honest challenger is the model
Observation
What the study says
Safety requires that someone honest and interested challenges a false assertion before the deadline. It does not require a global vote on every step.
Model
What has to hold
At least one honest party will see a false assertion and can get a challenge transaction included in time.
Falsifier
What would retire it
What machine does the one-step checker implement?
04 Privacy in the 2018 paper is a separate claim
Observation
What the study says
The original system discusses virtual machines whose state need not be public to every verifier. The later Ethereum rollup also called Arbitrum made different choices about publishing data. Do not merge them.
Model
What has to hold
Users who need speed either wait out the window or trust a liquidity provider who is taking that risk.
05 The deadline is a security parameter
Observation
What the study says
A longer dispute window gives honest challengers more time and users less speed. The paper's security is this trade, written as a timeout.
Model
What has to hold
Users who need speed either wait out the window or trust a liquidity provider who is taking that risk.
Falsifier
What would retire it
How long is the window, and can a challenge transaction be censored for that long?
03 Sequence
One action, as an operating tape
- 01A party submits an assertion of a contract's new state, with a stake.
- 02During the challenge window, anyone can dispute by staking against that assertion.
- 03The protocol asks the two sides to bisect the execution trace. Each step, one half is abandoned.
- 04When one instruction remains, a checker contract runs it and sees who was right.
- 05The loser is slashed. If nobody disputes, the assertion is confirmed when the window ends.
04 Load-bearing
The argument, and where a pitch drops it
Optimistic means innocent until challenged
The cut
The system is cheap because false statements are expected to be rare and punishable, not because false statements are impossible. A chain with no one watching is an honour box. The paper's honesty is the existence of the challenger assumption.
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 optimistic system with no watcher is an honour system. The paper is explicit that verifiers matter.
One-step proofs need a precise machine
The cut
The checker must implement the same instruction the parties mean. Ambiguity in the virtual machine is a bug in the court, not in the application. Later rollup implementations had to freeze a precise ISA for this reason.
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 later client, parameter or reward formula is a different object from this paragraph.
The 2018 design and the 2020 rollup share a surname
The cut
Both use interactive disputes. They differ on who may assert, how data is published, and how Ethereum is used as the court. A citation should name the document. 'Arbitrum' alone is two eras.
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 this PDF as a description of Arbitrum One's current Nitro release.
05 Register
What has to be true
Adversary · Not re-measured
At least one honest party will see a false assertion and can get a challenge transaction included in time.
Model · Not re-measured
The one-step checker matches the off-chain machine.
Stake and threshold · Not re-measured
Stakes are large enough that grieving the dispute game is not a profitable way to stall.
Model · Not re-measured
Users who need speed either wait out the window or trust a liquidity provider who is taking that risk.
06 Divergence
What happened after the paper
Arbitrum One and later chains took the dispute-game idea onto Ethereum as an optimistic rollup, with a public assertion and a delay on withdrawals. The 2018 paper is the bisection argument and includes privacy goals the rollup did not centre. Read the delay and the data-posting rule from the system you actually mean.
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
Publishing every contract step on a base chain is expensive, and publishing only a result forces verifiers to re-run the whole computation to argue. Arbitrum's protocol makes a challenger and a manager narrow a disagreement to a single instruction.
What the design proposes
- Managers stake on a claimed execution. Verifiers can challenge.
- Bisection finds one step. The base layer adjudicates that step, not the entire program.
- Privacy in the paper's title refers to keeping the computation off the public critical path, under the protocol's assumptions, not to zero-knowledge receipts.
How the mechanism is specified
- Honest verifiers must be able to see enough of the execution to challenge. Data availability is assumed in the way the paper states.
- The economic assumption is that a dishonest manager will be challenged because someone's stake is at risk.
- The virtual machine in the paper is not the later WASM or EVM compatibility layer.
What this page does not treat as proven
- Do not cite this PDF as a description of Arbitrum One's current Nitro release.
- An optimistic system with no watcher is an honour system. The paper is explicit that verifiers matter.
- Dispute delay is part of the user experience. The paper does not set it to zero.
Why the desk still reads it
This is the paper to hand a team that says 'we'll post the result and anyone can challenge' without a bisection protocol, a stake, or a watcher. Those three are the design, not the slogan.
09 Lexicon
Terms, opened into the record
- Assertion
- A staked claim that a virtual machine reached a particular state.
- Bisection
- Splitting an execution trace in half until the disagreement is a single step.
- Challenge window
- The time during which a false assertion can be disputed, after which it is treated as true.
- One-step proof
- The on-chain check of the single instruction the bisection isolated.
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.
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.