Author: Christian Decker 2018-04-06 12:23:45
Published on: 2018-04-06T12:23:45+00:00
The issue of retaliatory constructions in multi-party cases has been raised by ZmnSCPxj on the Lightning-dev forum. Specifically, the challenge is how to extend the two-party model to three-party cases where A cheats. In this scenario, C detects the cheat and should be able to claim some money from A. However, it is unclear how much C can claim without disrupting the correct state. This problem becomes more complex as the number of participants increases, making the scalability of offchain "purse" a concern.Pre-committed retaliation transactions have been proposed as a way to identify and punish the misbehaving party. However, this solution has its own challenges, including the large number of pre-committed transactions that need to be carried around and the possibility of colluding parties punishing each other, reducing the cost of the attack. Retaliatory constructions, on the other hand, do not impose limits on the number of updates allowed to the offchain "purse". Rusty shachains require only O(1) storage on the number of updates, making retaliatory constructions more suitable for channels.Channel factories can use the Duplex construction near the root of the initial onchain anchor transaction but be terminated in Poon-Dryja retaliatory channels. This balance allows for restructuring channels, which is expected to be rarer, using the Decker-Wattenhofer construction while using the Poon-Dryja construction for "day-to-day" normal operation. The DMC construction has parameters that allow tweaking the number of invalidations, enabling years of operation without re-anchoring. Penaltyless invalidation has several advantages over LN, including no state asymmetry and no toxic state information leakage.
Updated on: 2023-05-24T19:01:57.186543+00:00