The Diem Blockchain

The five-minute read

One action, walked through

1. A governed set of validators is configured, with voting power and stated identities.
2. A client submits a payment transaction authenticated by an account key.
3. Validators order it with a Byzantine protocol derived from the HotStuff line and execute the Move module that defines the currency.
4. A quorum signature on the ledger state lets a client accept the payment without asking one operator.
5. A governance action can upgrade a module or change membership. That action is a transaction with its own authority, not an invisible edit.

The argument, unpacked

What has to be true

• The validator list is the one users think it is. A quiet change to the list is a quiet change to the trust model.
• More than two thirds of voting power follows the protocol and stays online.
• Currency modules do not grant a hidden admin a right the user did not notice.
• Clients actually verify quorum signatures rather than trusting a single endpoint.

What happened after the paper

The Diem association wound the project down. Move, the BFT lineage and several engineers continued into Aptos, Sui and other efforts. Those networks are not Diem. The 2020 paper is the permissioned payment design, including its compliance posture.

What to check before you use the idea

• Who can change the validator set, and how is that change published?
• Can an asset be frozen, and which module states that?
• Do clients check a quorum certificate or a company's API?
• Is this paper being cited for a public chain that no longer has a permissioned set?

Terms

Permissioned validators

A known set of operators whose identities and voting power are part of the protocol configuration.

Move module

The code that defines an asset and the operations on it, subject to the language's resource rules.

Quorum certificate

A set of validator signatures large enough to establish a ledger state under the BFT assumption.

Governance action

A privileged transaction that upgrades code or membership, itself supposed to be visible on the ledger.

The problem the paper names

A global payment system that puts every transfer on an anonymous proof-of-work chain inherits that chain's fee market, finality profile and open membership. Diem's paper proposes the opposite defaults: identified validators, a BFT commit, and a programming model where a coin cannot be copied.

What the design proposes

• Validators are members. Safety is a BFT assumption about that set.
• Move treats assets as resources. The type system is supposed to stop a module from duplicating them.
• The currency section describes a basket reserve. That is a legal and custody design, not a consensus theorem.

How the mechanism is specified

• The consensus chapter specifies a HotStuff-style commit pipeline among the members.
• Transactions are scripts that call published modules. The paper separates the language from the ledger.
• On-chain accounts, gas and events are specified as part of one system.

What this page does not treat as proven

• The association wound the project down in 2022. There is no Diem network to 'use' on the back of this paper.
• A reserve and a validator set are institutions. The paper does not create the licences they would have needed.
• Aptos and Sui are not Diem with a new logo. Read their papers for what they changed, including open membership.

Why a venture studio still reads it

Still the right primary source when a venture says 'Move-based' or 'reserve-backed settlement' and cannot name the document those phrases come from. Read it as a design that did not ship, which is information, not a reason to ignore the specification.