Track 1 · Mechanics · lesson 9

Forks and reorgs

7 min

Two validators produce a valid block for the same slot, moments apart. Both are legitimate. Half the network sees one first, half sees the other.

Nobody has cheated, and the chain has two tips.

Two valid blocks, one surviving chain.

The rule: follow the heaviest chain

Nodes do not argue about this. Each follows the branch with the most accumulated work — or, under proof of stake, the most attestation weight.

The moment one branch gets extended first, nodes on the other branch switch. The abandoned block is orphaned, and the transactions inside it that did not make it into the winner return to the mempool to be included later.

This is why "one confirmation" is not certainty, and why the finality ladder from lesson 5 exists. A transaction one block deep can be reorganised out through no fault of anyone. It is rare, brief, and completely normal.

Deeper transactions are safer because reorganising them means out-building everything that came after.

Predict

Your payment was in a block that just got orphaned. What happened to it?

It is lost and must be signed againIt goes back to the mempool and will likely be included in the next blockIt is executed twice, once on each branch

Two words that look alike

Reorg — routine. The chain reshuffles its most recent block or two while nodes converge. Happens regularly. No rules were broken.

Hard fork — deliberate. Humans change the rules, and nodes must upgrade to follow. If enough refuse, the chain permanently splits into two, each with its own history from that point.

Ethereum has hard-forked many times to ship upgrades, and almost always everyone upgrades together, so nothing splits.

The famous exception was 2016. After the DAO hack drained $60 million, the community forked to reverse it. Most followed the fork — that is the Ethereum you use today. A minority refused, arguing the ledger must not be editable at any price, and kept the original chain as Ethereum Classic.

Both chains still exist. That disagreement is the clearest example of what these systems are actually for.

Check

What is the difference between a reorg and a hard fork?

A reorg is an attack; a hard fork is an upgradeA reorg reshuffles blocks under unchanged rules; a hard fork changes the rules themselvesA reorg affects one node; a hard fork affects all of them

Worth remembering

Two valid blocks for one slot is normal; nodes follow the heaviest branch.
Orphaned blocks lose their claim, and their transactions return to the mempool.
Shallow reorgs are routine and are why one confirmation is not finality.
A hard fork changes the rules on purpose; a permanent split happens only if some nodes refuse.
The DAO fork in 2016 split Ethereum from Ethereum Classic over whether a ledger should ever be edited.