Reposting this Twitter thread here, where I don't have to break it into a dozen pieces. ;) https://x.com/szarka/status/1894048207396639169

This Peter K. Todd post is interesting, & provides a nice jumping-off point to discuss/ask about something I've been thinking a lot about lately: https://petertodd.org/2025/fake-channels-and-rgb-lightning

I fully expect someone to tell me that it already exists, & this post points to a similar idea, but I still haven't found exactly what I'm thinking about: https://x.com/OrangeSurfBTC/status/1883280652524761139

Lightning is a great protocol that gets us part way to fully scaling Bitcoin. One big drawback, however, is that it requires locking up a lot of liquidity. Worse, this results in some pressure to centralize.

Lightning is also sometimes *too* good of a solution! One way it's too good is that final settlement is virtually instantaneous. Necessary for some use cases, but definitely not all. Another way that Lightning is "too good" is that it requires no* trust. (* Except in edge cases where value is small, so close enough.) Sometimes a middle ground between trusting a custodian and full trustlessness is OK. In business, you often have trusted partners.

What if you want to route a payment via Lightning that doesn't need instant settlement and/or where you're dealing with a trusted partner? ("Trusted" is relative, of course. I'd trust most of my channel partners with 100k sats. Some with 1mm sats. Etc.) Possibly repeatedly? In this case, tying up enough liquidity to handle the maximum balance of funds between us is horribly inefficient! I assume large centralized exchanges already have ad hoc methods of handling this, but there seems to be (?) a lacuna in the Lightning protocol space.

Traditional payments systems often handle this use case via net settlement: instead of exchanging the full amount of each payment, the balance of payments is periodically (daily or even intraday) netted and only the difference has to be exchanged.

So why doesn't Lightning have something like a Net Settlement Channel? As @PeterKTodd's post observes, such a channel doesn't even have to correspond to a UTXO. The two "peers" wouldn't even have to have *any* "real" Lightning channel between them to route payments.

Imagine Alice & Bob are LSPs whose customers frequently exchange payments, & sometimes customers would be willing to wait 5 minutes or an hour for settlement, if the fee were lower. Maybe they typically route 100mm sats back & forth every hour, but payments roughly balance. Over a shorter time frame, though, the channel balance can move to one side or the other. So, with today's approach they either need a big channel or a way to rebalance (either on Lightning or off-chain) frequently.

With a net settlement channel, the partners could set up (& even advertise) a large virtual channel supported by a smaller private channel (or even no channel) to make payments periodically of the net balance due after netting accumulated invoices. The partners might even agree to settle small invoices immediately whenever the net balance isn't greater than some amount they agree to trust one another for, saving liquidity in exchange for a small amount of risk.

So, I'm wondering, is there some reason we can't have nice things like this? And are instead tying up too much capital and/or doing weird crap like rebalancing out-of-band over Liquid?