Non-attributable censorship attack on fraud-proof-based Layer2 protocols

So we have two potential results here:

  1. This attack is free to pull off, as spamming transactions isn’t actually needed. Conclusion: a majority of stake can censor any transaction (not just a fraud proof transaction) for free without attribution. The version of PoS chosen for Eth 2 is fundamentally broken and we can pack up our bags and go home, the whole thing is a failure.
  2. The attack isn’t free to pull off, spamming transactions is needed. Conclusion: users can coordinate a soft fork to force inclusion of the fraud proof. They don’t need to identify the attacker because the attacker identified themselves and penalized themselves through the absurd amount of transaction fees already paid.

Case 1 is an interesting result, as it would require drastic changes to Eth 2’s consensus protocol. Case 2 is a non-result.