See more at Roast this Concept: One big off chain shard
Generalizing the proof of publication concept here, what if the inclusion in the main chain requires n-number of random validators to reconstruct a root.
- Main chain signs a root and commits to it.
- Main chain distributes a list of included transactions.
- Randomly selected validators reconstruct the root.
- Validators sign root commit to it.
- Main chain calls commit with hash, all unwound hashes must match.
Result: proof that tx list / collation is available across the network.
Attack; main chain cheats and distributes hash in network so validators don’t have to retrieve data and do work.
Mitigation: Some kind of slashing game where validators can be slashed if the main chain lies to them and they believe the main chain. Ie. Main chain gets more than commit reward if it can trick validators therefore no validators trust artificial hashes.