LibraryConfidential compute2019Design paperCorpus record
Ekiden: A Platform for Confidentiality-Preserving, Trustworthy, and Performant Smart Contracts
Oasis. Raymond Cheng, Fan Zhang, Jernej Kos, Ari Juels, Andrew Miller, Dawn Song and collaborators.
The Ekiden paper: smart contracts that execute inside trusted hardware, with the ledger checking attestations rather than re-executing the private code. Oasis Network is the later production system in this line. The paper is the academic statement of the approach.
Ekiden's paper runs a smart contract's private state inside a hardware enclave, and puts the public chain in charge of ordering and of recording the enclave's attestation, so the chain never sees the secret.
- Evidence
- Primary paper
- Re-measured
- No
- Assumptions
- 4
- Records linked
- 3
Code
Source, repository, and what is actually hosted
- arXiv:1804.05141https://arxiv.org/pdf/1804.05141
- 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
The enclave is the computer that is allowed to see the data
A trusted execution environment decrypts state, runs the contract, and produces an attestation. The host operating system is assumed to be unable to read the enclave's memory.
Claim 02 · Paper model
The chain is the ordering and the audit log
Consensus nodes do not execute the private contract. They order requests and store the resulting commitment and attestation. Separation of compute and consensus is the architecture.
Claim 03 · Paper model
Attestation is how a user knows which program ran
The hardware signs a statement that a particular binary, with a particular measurement, produced the output. Trust moves from 'the operator' to 'the hardware and the measurement'.
Claim 04 · Paper model
Keys are a first-class problem
The paper has to say how contract keys are created, sealed to an enclave, and recovered. A lost key is a lost contract. A key that can be extracted is a failed privacy story.
Claim 05 · Paper model
Hardware trust is not 'trustless'
The paper's own model includes a manufacturer and a class of side channels. Calling the result trustless is a misuse of the word. The accurate word is relocated.
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 The enclave is the computer that is allowed to see the data
Observation
What the study says
A trusted execution environment decrypts state, runs the contract, and produces an attestation. The host operating system is assumed to be unable to read the enclave's memory.
Model
What has to hold
Attestation keys and the expected measurement are the ones the verifier has pinned. A verifier who skips the check has trusted the host.
Falsifier
What would retire it
Which enclave, and which measurement, is the verifier supposed to pin?
02 The chain is the ordering and the audit log
Observation
What the study says
Consensus nodes do not execute the private contract. They order requests and store the resulting commitment and attestation. Separation of compute and consensus is the architecture.
Model
What has to hold
Consensus availability is separate. A halted chain freezes the private contract even if the enclave is healthy.
Falsifier
What would retire it
Where are contract keys sealed, and who can recover them?
03 Attestation is how a user knows which program ran
Observation
What the study says
The hardware signs a statement that a particular binary, with a particular measurement, produced the output. Trust moves from 'the operator' to 'the hardware and the measurement'.
Model
What has to hold
Attestation keys and the expected measurement are the ones the verifier has pinned. A verifier who skips the check has trusted the host.
Falsifier
What would retire it
Which enclave, and which measurement, is the verifier supposed to pin?
04 Keys are a first-class problem
Observation
What the study says
The paper has to say how contract keys are created, sealed to an enclave, and recovered. A lost key is a lost contract. A key that can be extracted is a failed privacy story.
Model
What has to hold
The enclave implementation matches the paper's threat model, including the side channels the user cares about.
Falsifier
What would retire it
Where are contract keys sealed, and who can recover them?
05 Hardware trust is not 'trustless'
Observation
What the study says
The paper's own model includes a manufacturer and a class of side channels. Calling the result trustless is a misuse of the word. The accurate word is relocated.
Model
What has to hold
The enclave implementation matches the paper's threat model, including the side channels the user cares about.
Falsifier
What would retire it
What side channels has this implementation actually closed?
03 Sequence
One action, as an operating tape
- 01A contract is loaded into an enclave. The hardware produces an attestation of the binary's measurement.
- 02A user encrypts inputs to the contract key and submits them. Consensus orders the request.
- 03The enclave decrypts the relevant state, executes, and emits a new state commitment plus an attestation.
- 04Consensus nodes record that output. They do not receive the plaintext state.
- 05A later enclave with the same contract key can pick up the sealed state. A verifier checks the attestation against the measurement they expected.
04 Load-bearing
The argument, and where a pitch drops it
The manufacturer is in the trust root
The cut
Attestations are worth what the hardware signature is worth. A broken enclave, a leaked attestation key, or a compelled manufacturer changes the meaning of every proof. The paper does not hide this. Summaries often do.
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
Trusted hardware is a trust assumption. Calling the result trustless is inaccurate.
Side channels are in scope
The cut
A correct attestation does not stop an observer from learning secrets from timing, memory access or power, if the implementation leaks them. The paper's security theorem and a careless implementation are different objects. A review should ask what leakage was actually measured.
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 is not an endorsement of any particular chip, and not a current Oasis status report.
Private state still has public effects
The cut
The output that goes back to the chain, the size of a transaction, and the pattern of calls can describe the secret. Confidentiality of state is not confidentiality of the fact that a particular user called. Application design has to minimise what the output says.
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
Confidentiality of state is not confidentiality against a user who is supposed to receive the output.
05 Register
What has to be true
Model · Not re-measured
The enclave implementation matches the paper's threat model, including the side channels the user cares about.
Cryptography · Not re-measured
Attestation keys and the expected measurement are the ones the verifier has pinned. A verifier who skips the check has trusted the host.
Cryptography · Not re-measured
Key backup and recovery do not create a plaintext copy 'for safety' outside the enclave.
Data · Not re-measured
Consensus availability is separate. A halted chain freezes the private contract even if the enclave is healthy.
06 Divergence
What happened after the paper
Oasis and other systems drew on Ekiden's split between enclave execution and a consensus log. Hardware attestation has also had a rough public decade, which is a reason to read the threat model carefully rather than a reason to pretend the paper claimed trustlessness. The paper is the checklist: who computes, what the chain verifies, and which manufacturer those verifiers must trust.
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
Re-executing a contract on every node publishes the contract's inputs, and multi-party computation is expensive for general programs. Ekiden's bet is that a small set of compute nodes can run the contract inside a trusted execution environment, and that the chain only needs to check that the right binary, on plausible hardware, produced the output.
What the design proposes
- Compute nodes produce attestations. Consensus nodes order and record results. The roles are split.
- The ledger does not see the secret state. It sees a commitment and a hardware attestation.
- Key management and attestation freshness are part of the security argument, not operational trivia.
How the mechanism is specified
- Trust moves to the hardware manufacturer and to the attestation verification, plus the usual consensus assumptions.
- A side-channel or a broken enclave breaks the privacy claim even if the ledger is honest. The paper's model has to be read with that in mind.
- Oasis's later ParaTime architecture is an implementation choice beyond the paper.
What this page does not treat as proven
- Trusted hardware is a trust assumption. Calling the result trustless is inaccurate.
- The paper is not an endorsement of any particular chip, and not a current Oasis status report.
- Confidentiality of state is not confidentiality against a user who is supposed to receive the output.
Why the desk still reads it
When a venture proposes 'confidential contracts' for businesses, Ekiden is the checklist: who computes, what the chain actually verifies, and which manufacturer those verifiers must trust. If the pitch cannot name the enclave, it does not have this design.
09 Lexicon
Terms, opened into the record
- Enclave
- A hardware-isolated execution environment that is supposed to hide its memory from the host.
- Attestation
- A hardware-backed statement that a particular program produced a result.
- Sealed state
- Encrypted contract state that an enclave with the right key can reopen and the chain cannot read.
- Measurement
- The hash of the program the enclave claims to be running. Verifiers pin this value.
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.