LibraryConsensus2022Design paperCorpus record
The Aptos Blockchain: Safe, Scalable, and Upgradeable Web3 Infrastructure
Aptos. Aptos Labs.
Aptos Labs' 2022 paper for a Move-based chain that pipelines dissemination, ordering, execution and certification, and that treats protocol upgrades as a first-class feature rather than a hard-fork event.
Aptos takes the Move language and a pipelined Byzantine protocol, and it argues that a block can be spread, ordered, executed and certified as separate stages instead of one slow lockstep.
- Evidence
- Primary paper
- Re-measured
- No
- Assumptions
- 4
- Records linked
- 9
Code
Source, repository, and what is actually hosted
- Aptos white paperhttps://aptosfoundation.org/whitepaper/aptos-whitepaper_en.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
The pipeline is the architecture
Dissemination, ordering, execution and certification do not have to be the same leaders or the same moment. The paper's throughput claim is that these stages overlap.
Claim 02 · Paper model
Move is there for resource safety
Assets are resources the type system is supposed to stop you from copying or quietly dropping. That is a language claim sitting beside the consensus claim. Both are part of why the paper says 'safe'.
Claim 03 · Paper model
Parallel execution needs a memory model
Block-STM, the paper's parallel engine, executes transactions optimistically and re-runs them if they touched the same state. Conflicts are detected rather than declared in advance.
Claim 04 · Paper model
Upgrades are a stated requirement
The text treats on-chain configuration and upgradeability as a feature of the platform, not as an embarrassing back door. Readers should notice who can exercise it.
Claim 05 · Paper model
Diem is the ancestor, not the product
The people, the language and some of the consensus ideas come from the Diem effort. Aptos is a different network with different operators and a public token. Do not cite one white paper as the other's status.
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 pipeline is the architecture
Observation
What the study says
Dissemination, ordering, execution and certification do not have to be the same leaders or the same moment. The paper's throughput claim is that these stages overlap.
Model
What has to hold
Validators have the bandwidth for the dissemination stage. The pipeline does not create bandwidth.
Falsifier
What would retire it
At which pipeline stage does the product call a transaction final?
02 Move is there for resource safety
Observation
What the study says
Assets are resources the type system is supposed to stop you from copying or quietly dropping. That is a language claim sitting beside the consensus claim. Both are part of why the paper says 'safe'.
Model
What has to hold
Move's resource rules are actually what the published bytecode obeys. A native function that breaks them is outside the type system.
Falsifier
What would retire it
What happens to throughput when many transactions touch one resource?
03 Parallel execution needs a memory model
Observation
What the study says
Block-STM, the paper's parallel engine, executes transactions optimistically and re-runs them if they touched the same state. Conflicts are detected rather than declared in advance.
Model
What has to hold
The conflict rate of real traffic is low enough for optimistic execution to help.
Falsifier
What would retire it
What happens to throughput when many transactions touch one resource?
04 Upgrades are a stated requirement
Observation
What the study says
The text treats on-chain configuration and upgradeability as a feature of the platform, not as an embarrassing back door. Readers should notice who can exercise it.
Model
What has to hold
Move's resource rules are actually what the published bytecode obeys. A native function that breaks them is outside the type system.
Falsifier
What would retire it
Which parts of the VM are native code outside Move's checker?
05 Diem is the ancestor, not the product
Observation
What the study says
The people, the language and some of the consensus ideas come from the Diem effort. Aptos is a different network with different operators and a public token. Do not cite one white paper as the other's status.
Model
What has to hold
Validators have the bandwidth for the dissemination stage. The pipeline does not create bandwidth.
Falsifier
What would retire it
At which pipeline stage does the product call a transaction final?
03 Sequence
One action, as an operating tape
- 01A proposer orders a batch of transactions and disseminates the data.
- 02Validators certify the order under the pipelined Byzantine protocol without waiting for execution to finish.
- 03Execution runs in parallel. Transactions that conflict on state are detected and re-executed in a safe order.
- 04A certification of the resulting state is agreed, so a client can trust the output without replaying every transaction.
- 05A configuration change, if the governance path allows it, is itself a transaction in this pipeline rather than a flag day organised in a chat room.
04 Load-bearing
The argument, and where a pitch drops it
Optimistic parallelism fails in a specific way
The cut
If every transaction touches the same account, Block-STM serialises and the speedup disappears. The paper is honest only if its speedup is quoted for a workload with modest contention. A payments-to-one-hot-contract demo will not show the number in the abstract.
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
Parallel execution still serialises on contended state. The paper does not abolish hot accounts.
Separating order from execution changes finality language
The cut
A client may know the order before the state is certified. Products that say 'final' have to say which stage they mean. Shipping a receipt at the ordering stage is a different promise from shipping it after state certification.
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
Aptos is not Diem. It inherits the Move idea and a team history, and it is a different network.
Upgradeability is a trust decision
The cut
A chain that can change its rules quickly can fix bugs and can also change balances' meaning. The paper lists this as a goal. A study should say who holds the upgrade key and what delay exists. Silence on that point is not neutrality.
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
Upgradeability cuts both ways. A chain that can change quickly can change in ways holders did not model.
05 Register
What has to be true
Model · Not re-measured
Validators have the bandwidth for the dissemination stage. The pipeline does not create bandwidth.
Model · Not re-measured
The conflict rate of real traffic is low enough for optimistic execution to help.
Stake and threshold · Not re-measured
State certification quorums match the Byzantine assumption, usually more than two thirds.
Data · Not re-measured
Move's resource rules are actually what the published bytecode obeys. A native function that breaks them is outside the type system.
06 Divergence
What happened after the paper
Aptos launched a public network with this stack and has revised consensus and execution since the paper. The Diem lineage explains the language. It does not mean a Diem payment pilot is live on Aptos. Read parameters from the network, and the safety argument from the paper.
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 monolithic validator that orders, executes and stores in one lockstep wastes the machine, and a chain that cannot upgrade without a political fork will calcify. The paper separates the stages and puts upgradeability in the design.
What the design proposes
- Move is the native language. Resource types are how assets are kept from being copied by accident.
- The data model is accounts and resources, with the claim that parallel execution does not require the sender to pre-declare every read and write.
- A pipelined path lets different stages of different blocks overlap.
How the mechanism is specified
- Consensus orders metadata. Execution can then proceed in parallel where conflicts allow.
- Key rotation and account abstraction are part of the stated user-safety story, not a wallet afterthought.
- The paper's performance language is a design goal. This page does not adopt it as a benchmark.
What this page does not treat as proven
- Aptos is not Diem. It inherits the Move idea and a team history, and it is a different network.
- Upgradeability cuts both ways. A chain that can change quickly can change in ways holders did not model.
- Parallel execution still serialises on contended state. The paper does not abolish hot accounts.
Why the desk still reads it
For a venture building on Move, Aptos is one of two public platform papers (with Sui) that has to be read against the original Move language paper, not instead of it. Ask what is shared with Diem's design and what was replaced.
09 Lexicon
Terms, opened into the record
- Move
- A language in which assets are resources: the type system is designed so they are not copied or dropped by accident.
- Block-STM
- A parallel executor that runs transactions optimistically and retries those that conflicted on memory.
- State certification
- A quorum signature on the result of execution, so a client can trust the new state root.
- Pipeline
- Overlapping the stages of a block so ordering does not wait for execution and execution does not wait for the next proposal.
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.