Pure Proof of Stake (PPoS)

Pure proof of stake (PPoS) is the consensus mechanism used by the Algorand blockchain. In each round, a small group of participants is selected at random to propose a block and then to vote on it, with each account's chance of selection proportional to how many ALGO tokens it holds. Selection is done privately with cryptography, so nobody knows who has been chosen until they reveal it by taking part, which makes targeted attacks on the chosen participants difficult.

The design was developed by the cryptographer Silvio Micali, a professor at MIT and winner of the 2012 Turing Award, who founded Algorand; the network launched its main chain in 2019. PPoS is a variant of proof of stake that its designers present as an alternative to delegated proof of stake, where holders elect a fixed set of block producers. The token itself is covered under Algorand (ALGO).

Imagine a large club that needs to approve a new page in its shared record book every few seconds. Instead of asking every member, or letting a few elected officers decide, the club holds a secret lottery each time. Members with more shares get more lottery tickets. The winners step forward, check the proposed page and vote on it, then go back into the crowd.

Because the winners change every round and are only revealed when they speak, an outsider cannot easily bribe or attack them in advance. And because the lottery is weighted by shares, anyone hoping to control the outcome would need to own a large part of the club.

Algorand uses a technique called cryptographic sortition. Each participant runs a verifiable random function, a calculation that takes their secret key and public data from the chain and outputs a random-looking number together with a proof that the number was computed honestly. If the number falls within a range set by the participant's stake, they have been selected for that step. Each round then proceeds in stages:

1. Proposal. A few randomly chosen accounts propose candidate blocks, and the network favours the proposal with the best priority.
2. Soft vote. A freshly chosen committee votes to settle on a single candidate.
3. Certify vote. Another committee checks the candidate's transactions and, if enough weight agrees, certifies it.

Committee members are replaced after each step, so a participant's influence lasts only for a single message. The protocol is a form of Byzantine fault tolerance and assumes that more than two-thirds of the stake is held by honest participants.

A key claim of PPoS is immediate finality. Once a block is certified, it is final, and the protocol is designed so that two different blocks cannot be certified for the same round under its honesty assumption. This contrasts with proof of work chains such as Bitcoin, whose probabilistic finality means a payment becomes more secure as further blocks are added. For payments and exchanges, the practical result is that a transfer can be treated as settled within a few seconds, without waiting for several confirmations. If the network is split or heavily disrupted, the protocol is designed to pause rather than produce conflicting histories, then recover once communication improves.

In PPoS, holders do not lock tokens with a third party or hand them to a validator to take part. Any account can register as online by running a participation node, and its full balance counts towards its selection chances. Original Algorand PPoS also did not use slashing, the confiscation of stake for misbehaviour that is common on other proof-of-stake chains; the security argument instead relies on honest holders having a stake in the network's value. Algorand has revised how participation is rewarded several times since launch, moving from rewards paid to all holders, to a governance programme and later to incentives aimed at accounts that actually run consensus nodes. Readers should check the Algorand Foundation's current documentation for the rules in force.

Supporters see PPoS as a fast, low-energy design that avoids the concentration of power in a few elected producers. Critics raise several points:

• Participation in practice. Security depends on enough stake being online, and in the network's early years a large share of tokens sat with the foundation, early investors or exchanges rather than with independent node runners.
• Wealth weighting. As with all staking-based systems, larger holders have more influence, which some see as a structural bias towards concentration.
• Incentives. Without slashing or meaningful rewards, ordinary holders had little reason to run nodes, one reason the reward model was changed.
• Liveness over safety. Choosing to halt rather than fork under severe disruption protects consistency but means the chain can stop producing blocks, as other consensus mechanisms with similar priorities also can.

• Algorand Developer Portal
• Algorand Foundation