Rene Pickhardt lately kicked off a thread discussing the variations between two social gathering and multiparty (greater than two members) cost channels because it pertains to his analysis work round cost reliability on the Lightning Community. He voices a rising skepticism of the viability of that course for improvement.
The excessive stage concept of why channel factories enhance the reliability of funds comes all the way down to liquidity allocation. In a community of solely two social gathering channels, customers should make zero sum selections on the place to allocate their liquidity. This has a systemic impact on the general success charge of funds throughout the community, if individuals put their liquidity someplace it isn’t wanted to course of funds as a substitute of the place it’s, funds will fail because the liquidity in locations individuals want is used up (till it’s rebalanced). This dynamic is solely one of many design constraints of the Lightning Community identified from the very starting, and why analysis like Rene’s is extremely essential for making the protocol/community work in the long term.
In a mannequin of multiparty channels, customers can allocate liquidity into giant teams and easily “sub-allocate” it off-chain wherever it is sensible to within the second. Which means that even when a node operator has made a poor resolution during which particular person to allocate liquidity to, so long as that particular person is in the identical multiparty channel with individuals that will be peer, they’ll reallocate that poorly positioned liquidity from one to the opposite off-chain with out incurring on-chain prices.
This works as a result of the idea of a multiparty channel is basically simply everybody within the group stacking typical two social gathering channels on prime of the multiparty one. By updating the multiparty channel on the root, the 2 social gathering channels on prime will be modified, opened, closed, and so on. whereas staying off-chain. The issue Rene is elevating is the price of going on-chain when individuals don’t cooperate.
Your entire logic of Lightning is predicated round the concept that in case your single channel counterparty stops cooperating or responding, you may merely submit transactions on chain to implement management over your funds. When you’ve gotten a multiparty channel, every “stage” within the stack of channels provides extra transactions that must be submitted to the blockchain with the intention to implement the present state, that means that in a excessive charge atmosphere multiparty channels will probably be dearer than two social gathering channels to implement on-chain.
These are core trade-offs to think about when taking a look at these methods in contrast to one another, however I feel focusing completely on the on-chain footprint ignores the extra essential level relating to off-chain methods: they’re all about incentivizing members to not go on-chain.
Correctly structuring a multiparty channel, i.e. the way you set up the channels stacked on prime, can will let you pack teams of individuals into subsections which have a repute for prime reliability, or who belief one another. This may permit individuals in these subgroups to nonetheless reorganize liquidity inside that subgroup even when individuals outdoors of it are usually not responsive quickly, or go offline because of technical points. The on-chain value of implementing issues, whereas essential, is sort of tangential to the core design purpose of an off-chain system: giving individuals a motive to remain off-chain and cooperate, and eradicating causes for individuals to not cooperate and pressure issues onc-chain.
It’s essential to not lose sight of that core design facet of those methods when contemplating what their future will appear like.
