Pierre Rochard from BitcoinAdvisory.com wrote a great series of articles where he covers in details his participation at the ChainCode Labs Lightning residency event.
Margherita Favaretto, a student working on remediation protocol for Lightning Network double-spend attacks asked for feedback for a proposed solution to double spend attacks using a “trusted remediation” gossip protocol.
ZmnSCPxj pointed out that double spend attacks are not possible on the Lightning Network unless both parties involved in the channel agree to it, which is not likely, first because the man at the other end of the channel will lose money. Secondly even if the other end of the channel is irrational enough to help the other guy double spend, they will still ask for an invoice and give the money using “existing invoice-payment mechanisms.” ZmnSCPxj added:
If the problem you are trying to solve, is the inadvertent publication of revoked commitment transactions, then the correct solution is not to have revocable transactions in the first place, i.e. eltoo. While it can be argued that it would take time for needed features of eltoo to appear on the blockchain layer (SIGHASH_NOINPUT_UNSAFE), it would also take time to implement “trusted remediation”, by which time the problem could be solved by switching over to eltoo.
Commitment transactions are huge part of the penalty system Lightning enforces to make sure everyone plays nicely with one another, it make sure that if someone in the channel broadcasts an older transaction, thus trying to scam the other party, the commitment transaction will allow the first party access to the scamming party’s coins.
There were recently a few proposals to edit the commitment transaction, mainly the edits are about the format. Making the CLTV timeout symmetrical to avoid trying to pressure the peer into closing, making the remotepubkey BIP-32 styled and using the OP-TRUE style output to allow Child Pays For Parent fee dependancy.