ALGORAND: The Efficient and Democratic Ledger

The five-minute read

One action, walked through

1. Each online account holds a participation key and a stake weight.
2. For a given round and step, the account runs cryptographic sortition against a shared seed and learns whether it has a vote, and of what weight.
3. If selected, it gossips a signed vote with the sortition proof. Others can verify the proof without a private channel.
4. A committee of proposers offers candidate blocks. A committee of voters reduces them until one block has a quorum.
5. The seed for the next round is updated from the round's output so the next committee cannot be predicted far in advance.

The argument, unpacked

What has to be true

• Honest stake exceeds the paper's threshold, often two thirds in the Byzantine setting it argues.
• The adversary cannot corrupt a newly revealed voter before the step completes.
• Enough stake is online and participating. Offline stake does not vote, and the paper's liveness depends on participation.
• The random seed is unpredictable and unbiasable within the round structure.

What happened after the paper

The Algorand network implemented the idea with participation keys, a relay topology, and later performance work. Those operational choices are not theorems in the 2017 paper. Committee size and timing parameters are where the proof meets the network.

What to check before you use the idea

• Is the committee known before it votes?
• What share of stake must be online for a round to complete?
• How is the seed produced, and can the last proposer withhold it?
• Are votes weighted by stake or by account?

Terms

Cryptographic sortition

A local lottery, proved to everyone else, that selects a voter with probability proportional to stake.

Player replaceability

The property that a voter matters only for the message they already sent. Later corruption cannot change that step.

Committee

The small stake-weighted sample whose votes decide a step, then is discarded.

Adaptive adversary

An attacker who can choose whom to corrupt as the protocol runs, not only at the start.

The problem the paper names

Classical Byzantine agreement is expensive if every holder talks to every other holder. Bitcoin-style longest chain is permissionless but slow to become confident. Algorand wants a committee that is large enough to represent stake and small enough to finish quickly, without publishing the committee in advance.

What the design proposes

• Cryptographic sortition lets a holder learn privately, from their own key and a public seed, whether they were selected.
• A verifier can check the selection proof after the message appears. Before that, the holder is not a target.
• Player replaceability means the next step draws a new committee. Bribing the last one does not carry forward.

How the mechanism is specified

• The ledger is a sequence of certified blocks, not a race of forks that the heaviest chain wins later.
• Safety is argued from the stake-weighted honesty of the selected sets.
• The 2017 systems paper by Gilad, Hemo, Micali, Vlachos and Zeldovich is the implementation companion, not a different project.

What this page does not treat as proven

• Sortition only helps if the underlying stake and keys are what the paper assumes.
• The paper is not a description of later Algorand product features or relay networks.
• Committee designs still need an account of who can censor inclusion. Certification is not the same as open membership of every transaction.

Why a venture studio still reads it

The idea worth stealing, carefully, is replaceability: do not leave a standing set of signers who can be found and pressured. A venture that publishes its validator set on a webpage has not adopted this property.