In the documentation for optimism (which is transparent about being somewhat incomplete at this experimental stage in their development), it is stated that

"It's important to note that a successful challenge does not roll back Optimism itself, only the published commitments about the state of the chain. The ordering of transactions and the state of Optimism is unchanged by a fault proof challenge."

But how is this possible? If a transaction gets executed and someone balances are not correct (the state commitments in the StateCommitmentChain are not a valid result of the transactions in the CannonicalTransactionChain, and then other transactions may happen on top of this that also have an incorrect state commitment. Many transactions that depended on the correctness of that faulty transaction will have to be reverted.

How can they possibly claim that the state of optimism is unchanged by a fault proof challenge? If the published commitments about the state of the change are changing (their words), how does this not imply a change in the state of optimism? And more generally, how can you hop to have finality of your transactions on optimism if you are building on top of a state that could be challenged for an entire week?

If someone could please explain this to me, that would be amazing. Maybe I'm missing something, but it kind of seems like a big lie...



2 Answers 2


the thing with optimistic rollups is that this is a layer 2 scaling solution for blockchains. Rollups and challenging the published state root in a rollup is not a part of the main layer 1 chain (which is the state of Optimism as such). As long as the challenge period is not over, the rollup is not considered as part of the blockchain and the transactions as not truly final. If no fraud proof is provided, the rollup is being publish to the main chain (layer 1) and this is the state Optimism refers to. If a fraud proof is provided, the rollup is not being accepted at the main chain and as such does not affect the state of Optimism. Best, Diba (see https://research.paradigm.xyz/rollups for more)

  • ok, so by "does not affect the state of optimism", you don't mean the optimism layer 2 chain (which IMO is the more obvious interpretation of that statement) -- you mean the state of optimism from the perspective of the layer 1. that makes sense. but it should be clearer that the state of optimism on the actual optimism chain absolutely does change with a fraud proof, and that building on top of it in real time is virtually impossible under the assumption of frauds happening.
    – Paul
    Mar 7, 2023 at 20:48

I hope this helps - https://twitter.com/bkiepuszewski/status/1471116288261038088.

Ok, let's say there is 1 transaction "10 ETH Alice -> Bob" mined in the L2. Rollup operator packs it into the "calldata" in the a bit compressed form, and then pushes this data to L1. Together with the calldata it should also push a Merkle Tree root (state commitment) of the new state.

How I understood it... Rollup operator can push wrong state commitment. Say, it can also include a transaction "5 ETH Bob -> Operator".

Offtopic I can not understand why TXs are without signatures here... In Plasma all TXs were signed so Operator can not cheat. Probably it is because Plasma used simple UTXO model with "asset owners", but Rollup can not. See my question here

If Operator cheats, then L2 state is still OK (tx is still valid there! and he was unable to forge a "Bob -> Operator" tx), but L1 chain state is wrong until Operator pushes correct data. So the only problem here is if Bob wants to withdraw his money...and he will need to wait until everything is settled.

p.s. Personally I don't yet understand what if L2 gets attacked with the 51% "double spend" attack... And it can be done by the User + bribed Operator. It looks to me like you are doomed in this case, because L2 is in bad state.

If anyone knows the answer, please help.

  • 2
    I'm very curious about the 51% double spend attack on a layer 2 network like optimism as well, maybe you should post this as its own question!
    – Bruce
    Mar 13, 2023 at 19:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.