Consensus: How Blockchains Agree

Decentralization sounds simple until you realise every node acts on its own. In a blockchain each node holds its own copy of the ledger and processes data independently. Nodes see transactions at different times, some go offline, some behave incorrectly and some may try to cheat. Without a way to agree the network would split into a thousand truths. Consensus is what keeps all those independent nodes and all those independent ledgers marching in the same direction.

With this, let’s dive in what actually is consensus?

What is Consensus?

Consensus is a way for decentralized blockchain to decide what is the correct state of blockchain should be. Every node has its own view of the world, so the network needs a shared method to agree on one chain, one block, one history. At it’s core consensus solves two major problems -

• If everyone could add blocks whenever they want, who gets to propose the next block which is added to the final chain.
• How do we know which block is the real one? Nodes receive data in different orders, so they need a consistent way to choose the valid chain even when the network is noisy or under attack.

Blockchains also need to deal with Byzantine faults, which is a simple way of saying some nodes can behave unpredictably. They might be offline, buggy, or even dishonest. Consensus has to keep the honest part of the network aligned even when some nodes act weird or go silent.

This is why good consensus mechanisms aim for two core properties: Safety (honest nodes should never accept two different histories) and Liveliness (the network should keep making progress even if some nodes fail or misbehave).

If a blockchain gets both right, the network behaves like one system rather than thousands of independent machines trying their luck.

Now, that we are clear on the goal of consensus - let’s explore how different consensus algorithms actually achieve this.

Consensus has two sides to it -

1. How does the network decide who gets to propose the next block?
2. How does the network decide which chain is the real one when multiple valid chains appear?

So, let’s tackle them one by one!

How are blocks created? (PoS, PoW, PoA)

Proof of Work (PoW)

PoW uses computational effort to decide who gets to propose the next block. Miners race to solve a cryptographic puzzle. The first miner who finds a valid solution wins the right to produce the next block.

How it works?

On a high level -

• Miners collect pending transactions.
• They try to find a hash below a target value.
• Finding it takes real energy and hardware.
• The winner broadcasts their block to the network.
• Nodes verify the block and add it to their chain.

Doing the work is expensive. Verifying the work is cheap, and all of this happens pretty quickly. For example - Bitcoin (PoW) targets to create block in around 10 minutes . Verification takes around milliseconds and propogation take around 1-2 seconds.

It has it’s strength and weaknesses -

• Strength - Permissionless (anyone can join the network, participate in block production, send transactions, or run a node without needing approval from any authority) and highly secure as long as the majority of mining power is honest.
• Weakness - Energy intensive. Slow. Vulnerable to 51 percent attacks if mining becomes centralized (in a developed Blockchain like Bitcoin - this is not feasible because of the cost associated with getting the majority).

Example of Proof of Work Blockchains - Bitcoin, Litecoin/Dogecoin, Monero, etc.

Proof of Stake (PoS)

PoS uses economic weight (some stake) instead of energy to decide who gets to propose the next block. Validators lock up some of their tokens as stake, and the protocol selects a proposer based on this staked amount. The larger the stake, the higher the chance of being chosen.

How it works?

On a high level -

• Validators lock up tokens as stake.
• The protocol randomly selects one validator to propose the next block, usually weighted by stake.
• The selected validator creates the block and broadcasts it to the network.
• Other validators check the block and attach their signatures.
• Once enough signatures are collected, the block is accepted and added to the chain.

The entire process is fast and efficient because nodes are only verifying signatures and basic block checks. Most modern PoS chains produce blocks within a few seconds. For example - Ethereum PoS creates a block every 12 seconds and Solana targets around 400 milliseconds per block.

It has it’s strength and weaknesses -

• Strength - Energy efficient, fast, and scalable. Anyone with stake can become a validator, making the system permissionless while reducing hardware requirements. Strong economic incentives encourage validators to behave honestly because stake can be lost if they misbehave.
• Weakness - Requires careful design to avoid issues like long-range scenarios or misaligned incentives. Also depends on a healthy distribution of stake so that no small group controls too much voting power.

Example of Proof of Work Blockchains - Ethereum, Solana, Cardano, etc.

Proof of Authority (PoA)

PoA is used in permissioned or enterprise blockchains. Instead of open participation, a fixed and trusted set of validators is responsible for producing blocks. Their authority comes from their identity or reputation rather than tokens or computational power.

How it works?

On a high level —

• The network has a predefined set of approved validators.
• Validators take turns creating blocks based on a schedule or voting process.
• The selected validator produces the next block and broadcasts it to the network.
• Other validators verify the block and sign it.
• Once enough validators confirm it, the block is added to the chain.

Because the validator set is small and known, block production is predictable and extremely fast. Private and enterprise PoA networks typically produce blocks within 1 to 5 seconds, since they do not need mining or stake-weighted selection.

It has its strength and weaknesses —

• Strength — Very fast, low energy, and easy to maintain. Predictable block times and instant finality in many cases. Ideal for enterprise, consortium, and internal blockchain deployments where participants are known and trusted.
• Weakness — Not truly decentralized. Trust shifts from open participation to validator identity and governance. Requires strong organizational trust to avoid misuse or centralization bottlenecks.

Examples of Proof of Authority Blockchains — Ethereum Clique networks (private chains), Aura-based Substrate chains, VeChain, and various enterprise Ethereum deployments.

Summary for PoW, PoS, PoA

Now that we know how different blockchains create blocks, we need to address the second half of the puzzle. Block production alone does not guarantee that all nodes will agree on the same history. Even with PoW, PoS, or PoA, the network can still see multiple valid blocks at the same height. This creates temporary forks where different parts of the network follow different versions of the chain.

To keep the blockchain unified, the network needs a way to decide which chain becomes the real one. This decision is made by something called the fork choice rule. It is the logic that every node follows when there is more than one valid branch.

Let’s explore why forks happen and how different blockchains resolve them.

Fork Choice Rules

This is the answer to the question - “How does the network decide which chain is the real one?"

A blockchain can temporarily split into multiple valid chains when nodes receive different blocks at the same height. Everyone has a copy of the blockchain and anyone can build a block and try to push it to the chain, so collisions are natural. If two miners publish block 100 at the same moment, half the network may see one version while the other half sees the other.

There are multiple ways such forks happen -

• Block propagation takes time
• Two valid blocks can be created at the same height
• Network latency varies
• Validators or miners may compete for the same slot
• Some nodes may temporarily disagree on state

The fork choice rule is the protocol that resolves forks in a blockchain and decides which chain becomes the canonical history. When multiple valid branches appear, the fork choice rule tells every node which one to follow so the entire network agrees on a single timeline.

How does different blockchains choose the final chain?

Different blockchains use different fork choice rules based on their consensus design. The goal is the same everywhere: when multiple chains appear, pick one timeline and move forward. The method, however, depends on whether the blockchain is using PoW, PoS, or PoA.

Proof of Work

PoW chains follow the chain with the most accumulated proof of work. This is often described as the “longest chain rule,” but what nodes really track is the total work, not just the block count. Each block contains a proof of work tied to its difficulty then nodes measure the total difficulty across the entire chain and the chain with the highest accumulated work becomes the canonical chain.

You will often see Bitcoin described as using the “longest chain rule,” but the protocol actually selects the chain with the most accumulated proof of work. Each block contains a difficulty target encoded in its header, and nodes compute the total work across the entire chain by summing the difficulty of every block. The chain with the highest cumulative work becomes the canonical chain, not necessarily the one with the most blocks. The two look similar only because Bitcoin’s difficulty changes slowly, so in practice a longer chain almost always represents more total work.

Proof of Stake

PoS chains use validator votes and stake weight instead of energy. The fork choice rule picks the chain that has the most active validator support behind it.

PoS chain like Ethereum uses something called Latest Message Driven – Greediest Heaviest Observed SubTree or LMD-GHOST.

• Latest Message Driven - Each validator’s most recent vote (their “latest message”) is what counts. Older votes are ignored.
• Greediest Heaviest Observed SubTree - Nodes walk down the chain choosing the branch that has the most validator weight behind it. “Heaviest” refers to stake weight, not work. “Greediest” means the algorithm always picks the branch with the maximum weight at each step.

In simple words, LMD-GHOST tells Ethereum nodes to follow the branch with the most recent, most weighted validator support. Stake represents economic commitment, so the chain supported by the most stake is considered the correct one.

Other PoS networks use different/hybrid fork choice rules. Example -

• Tendermint - Uses a BFT (Byzantine Fault Tolerance) voting mechanism - Validators vote in rounds (pre-vote, pre-commit, commit). Once 2/3 majority signs a block, it is final. No reorgs possible.
• Polkadot - Uses two separate components - BABE (block production) & GRANDPA (finality gadget). Follow the longest chain until a GRANDPA finality vote happens. After finality, the chain does not fork.
• Solana - Uses Proof of History for ordering, Tower BFT for consensus and finality. Follow the chain with the highest “lockout” (latest voted state). It has a completely different design.

Proof of Authority

PoA chains rely on a small set of approved validators with known identities. Forks are rare because block production is predictable, but fork choice rules still exist.

PoA isn’t a single algorithm. Different networks use different engines under the PoA umbrella:

• Clique (Ethereum PoA): Follow the chain with the highest block number. If tied, follow the chain with the most validator signatures.
• Aura (Substrate PoA): Follow the chain that matches the expected validator rotation and time slots.
• BFT-style PoA: Once a block is finalized, it cannot be reverted. Deterministic finality.

Even though PoW, PoS, and PoA create blocks in very different ways, they all rely on a fork choice rule to keep the network aligned. Whether it is accumulated work, stake-weighted votes, or validator identity, the fork choice rule ensures the blockchain converges back to one shared history. Without it, decentralized networks would drift into competing timelines instead of behaving like a single coherent system.

With all this, there is one tiny component called Finality, which answers the questions - When can we trust that a block will never be reversed?

Finality describes the point at which a block becomes permanent and can no longer be reversed. Different consensus models achieve this differently.

• (Probabilistic Finality) In PoW systems like Bitcoin, finality is probabilistic and strengthens as more blocks are added on top.
• (Economic Finality) In PoS systems like Ethereum, finality is economic and reached once validators vote across checkpoints, making it extremely costly to revert.
• (Deterministic Finality) In BFT-style PoS and PoA networks, finality is deterministic, meaning a block is instantly final once enough validators sign it. No matter the method, finality gives users confidence that a transaction is truly irreversible.

Now that we’ve covered all the pieces, here’s the final note. Consensus is the invisible engine that keeps a blockchain coherent. From block production to fork choice to finality, every part ensures that thousands of independent nodes behave like one system with a single shared history. Different blockchains may choose different designs, but the goal remains the same: trust the data without trusting the actors.

End Note - Consent is important!

We didn’t get into the security side of consensus here, and that was on purpose. I really wanted to keep this one clean and focused on understanding how everything works before we start breaking things apart. I’ll cover all the security issues, attacks, and weird edge cases in a separate blog. Until then, if you’ve got questions, thoughts, or anything about Blockchain and Security you can always reach me at @0xCardinal .