A blockchain has a surprisingly difficult job: hundreds or thousands of independent computers need to maintain the same record without a bank, company or administrator deciding which version is correct.
It is easy to think of this as a matter of checking whether transactions are valid. But that is only part of the problem. A network also needs a way to decide which transactions belong in the chain, which block comes next and what should happen when different participants temporarily see different versions of the network.
That is the job of a consensus mechanism.
The term can sound more complicated than it is. At its core, a consensus mechanism is the set of protocols, rules and incentives that allows a distributed network to converge on a shared version of its history.
The important point is that “consensus” does not mean every computer literally agrees with every other computer. It means the network has a defined process for determining which state should be recognised as valid.
A valid transaction is not the same as an agreed blockchain
Suppose Alice tries to spend the same Bitcoin in two different transactions.
A node can inspect each transaction independently. It can check the digital signature, confirm that the inputs are valid and apply Bitcoin’s protocol rules. But eventually the network still needs a shared answer to a bigger question: which transaction gets included in the accepted history?
This distinction between validation and consensus is fundamental.
Validation asks whether a transaction or block follows the rules. Consensus deals with what happens when multiple valid possibilities compete, or when participants receive information at slightly different times.
Bitcoin’s design illustrates the distinction particularly clearly. Full nodes independently validate blocks and maintain their own view of the chain. When their valid views converge on the same chain, the network is in consensus.
So the consensus problem is not simply “Can this transaction be checked?” It is closer to: how can independent participants that do not necessarily trust one another settle on the same history?
That is where the mechanism itself matters.
Consensus turns competing versions of history into one
The process varies between blockchains, but there is a basic pattern.
Transactions are broadcast across the peer-to-peer network. Nodes check whether they follow the protocol’s rules. Some participant is then selected, through the blockchain’s particular consensus model, to propose a new block or otherwise help advance the chain.
Other participants inspect what was proposed. They do not have to trust the proposer personally. They use the protocol to determine whether the block is valid and which chain should be treated as the canonical one.
This is an important feature of blockchain design: trust in an individual is replaced by predictable rules.
The network does not need a central referee watching every transaction. Instead, participants run software that applies the same underlying rules, while the consensus mechanism creates a way for them to converge despite network delays, competing blocks or potentially dishonest actors.
Bitcoin and Ethereum reach that result differently.
Bitcoin makes changing the past computationally expensive
Bitcoin uses Proof of Work (PoW). In simple terms, miners compete to produce a valid block by performing large numbers of calculations.
The calculation itself is not useful in the conventional sense. Its purpose is to make block production costly and difficult to fake. A valid Proof of Work provides evidence that a significant amount of computation was performed before the block could be added to the chain.
Because each block is linked to the previous one, changing an older transaction would require rebuilding the subsequent chain of Proof of Work as well. Bitcoin nodes consequently favour the chain with the greatest accumulated Proof of Work when competing valid chains arise.
That creates an economic barrier.
An attacker is not merely trying to convince other computers that a false history is true. They need to perform enough computational work to compete with the honest network.
This is the key insight behind Bitcoin’s consensus mechanism: the network uses a real-world resource to make dishonest history expensive to construct.
Ethereum puts economic value at risk instead
Ethereum uses Proof of Stake (PoS), which reaches consensus through a different set of incentives.
Instead of competing through mining, validators deposit ETH as a form of economic commitment. Validators are selected to propose blocks and to attest to blocks they consider valid. Their behaviour affects whether they receive rewards or face penalties.
That changes the security model.
In Proof of Work, the cost of participating in consensus is closely tied to hardware and electricity. In Proof of Stake, validators put capital at risk. Certain forms of dishonest behaviour can lead to slashing, meaning part of the staked ETH can be destroyed.
Ethereum also combines Proof of Stake with a fork-choice algorithm and a finality mechanism. Validators’ attestations help determine which chain should be followed, while checkpoints can become final when sufficient stake supports them. Once a block is finalised, reverting it would require a very large economic cost.
The contrast with Bitcoin is useful, but it should not be reduced to “energy versus coins”.
Both systems are trying to solve the same underlying problem: how do you make it unattractive or difficult for participants to manipulate the shared history?
They simply use different resources and incentives to get there.
The harder question is who gets to influence consensus
This is where consensus becomes more interesting than a simple explanation of PoW versus PoS.
Every blockchain has to answer a difficult question: what gives a participant influence over the network’s decision?
In Bitcoin, Proof of Work gives influence to computational power. In Ethereum, Proof of Stake gives weight to staked ETH within the protocol’s rules.
Neither means that miners or validators can simply write whatever they want into the blockchain.
A Bitcoin miner can propose a block containing an invalid transaction, but other nodes can reject that block. Bitcoin’s consensus rules exist independently of the miner’s wishes.
Likewise, an Ethereum validator does not have unlimited freedom to define the chain. Consensus clients apply protocol rules, process attestations and use fork-choice logic to determine the chain’s head.
This is why decentralisation does not mean the absence of power. It means power is constrained and distributed according to the rules of the system.
And those rules matter enormously.
What happens when the blockchain cannot immediately agree?
Sometimes different participants temporarily see different blocks at the same height.
This can happen because blocks arrive at different times, because of network latency or because competing proposals are produced before the network has converged.
The result is a fork: two possible branches of the chain temporarily coexist.
A consensus mechanism needs a way to deal with this rather than simply assuming that every computer will always see exactly the same thing.
Bitcoin nodes follow the chain with the greatest accumulated Proof of Work under its consensus rules. Ethereum uses its own fork-choice mechanism, LMD-GHOST, which weighs validator attestations when determining which branch should be preferred.
This is one reason finality matters.
A transaction being included in a block does not necessarily mean that every possible version of the blockchain has immediately disappeared. As the network adds more blocks or reaches protocol-defined finality, the cost of reversing that history can become much higher.
Consensus therefore is not just about choosing a winner once. It is also about making that decision increasingly difficult to undo.
Why the average user rarely notices any of this
When someone sends cryptocurrency, they usually see a simple interface: enter an address, choose an amount and press send.
Behind that transaction, however, a distributed system has to do considerably more work.
Nodes need to check the transaction. The network has to incorporate it into a valid block. Other participants need to assess that block. The consensus mechanism then helps determine how that block fits into the accepted history.
The user does not need to know how fork-choice algorithms work to make a payment. But they rely on the result every time they trust that their balance, transaction and previous history will be interpreted consistently by the network.
That is the practical meaning of blockchain consensus.
It is the machinery that allows a shared ledger to function without requiring every participant to know, trust or even communicate directly with every other participant.
The paradox at the heart of blockchain consensus
The more a blockchain removes the need for a central authority, the more important its underlying rules become.
A bank can resolve a dispute by making a decision internally. A decentralised blockchain cannot simply call an administrator and ask which transaction should count. It needs a protocol that defines what participants should accept and an economic structure that makes following those rules preferable to attacking them.
That makes a consensus mechanism much more than a technical component used to “approve blocks”.
It is the framework through which a network turns disagreement into a common history.
Proof of Work does this by making competing history expensive in computational terms. Proof of Stake does it by placing economic value behind validator behaviour. Both illustrate the same broader principle: a blockchain can remove the central referee, but it cannot remove the need for a system that decides what the network should recognise as true.
That system is the consensus mechanism.