Published on: 2018-03-05T13:50:10+00:00
The email discusses the complexities of the key revocation system in the Lightning Network and compares it to Christian Decker's solution for duplex micropayment channels. The writer acknowledges that the revocation system was already outlined in the whitepaper but finds it challenging to handle, potentially requiring a third-party observing service to detect cheaters. Rusty Russell also mentions this complexity in his article on the matter.Under the Decker-Wattenhofer approach, a potential thief could attempt to reuse old invalid state, which necessitates being online after the kickoff transaction is confirmed to broadcast the latest commitment transaction. If offline for an extended period, a watchtower-like service is also needed. However, under the Poon-Dryja approach, watchtowers only need to store the shachain and the funding txid, while those under Decker-Wattenhofer would have to store entire relative-timelocked transactions, potentially leaking economic information at each update. Additionally, channels drop on-chain, allowing economic information to be publicly readable.The email delves into the differences between Poon-Dryja channels and Decker-Wattenhofer duplex micropayment channels. The maximum lockup period for funds in a channel is determined by the number of updates the channel can have, with fewer updates resulting in shorter worst-case lockup periods before the channel can only be closed. Poon-Dryja channels have no limit on the number of updates possible and utilize O(1) lockup time and transactions for an n update limit. On the other hand, Decker-Wattenhofer employs O(log n) lockup time and transactions for n update limits. Each payment in Decker-Wattenhofer requires two updates: one to transfer payer funds to an HTLC and another to transfer the HTLC funds to the payee.While Decker-Wattenhofer offers the advantage of being extendable to any number of participants per channel, the construction of Poon-Dryja channels cannot be easily expanded beyond two participants. To address this limitation, Burchert-Decker-Wattenhofer channel factories combine multiple two-participant Poon-Dryja channels, allowing for any number of updates. The Decker-Wattenhofer part is only updated if all participants agree to redistribute their funds. This approach enables a potentially infinite number of possible channel updates for individual Poon-Dryja channels while minimizing the O(log n) cost for the maximum number of updates for fund redistribution. The email also mentions Rusty Russell's "shachain" concept, which reduces storage requirements for a sequence of revocation keys to just 64 bytes.Despite the writer's initial confusion about why the lightning spec went with the revocation key system instead of Decker's solution, they express respect for the creators of the Lightning Network and believe it will go down in history as a significant development in the field.
Updated on: 2023-07-31T19:49:18.961542+00:00