pull down to refresh
213 sats \ 11 replies \ @optimism 15 Aug \ on: Lightning swaps *are* the connective tissue - btc++ Insider lightning
I don't really like that lightning is nowadays advertised to be a settlement layer, as it feels like a goal-post shift. From where I'm sitting, all txs through an LN channel are unsettled until the channel is closed? What am I missing?
They're are technically unsettled but they are game-theoretically settled given the penalty (assuming it's economically and otherwise possible for parties to get transactions confirmed).
Are all but the oldest bitcoin transactions settled if reorgs are theoretically possible? Meaning, even what we consider "settled," an onchain transaction with N confirmations, can be unsettled. Settled is a matter of degree, and a matter of feasibility, cost, and consequences to the party trying to unsettle transactions, right?
It's kind of like the difference between a gold bar in my hand and a gold bar in a bank vault. You can steal either, perhaps one easier than the other, one with greater consequences than the other, so is the gold bar that's harder to steal with greater consequences the only gold bar that's mine? Or are both mine if it's improbable that someone would steal either gold bar?
Perhaps there's a distinction between what we mean by settled and confirmed. Both mean improbable to reverse, but confirmed is the kind of settled that only depends on the network continuing to behave honestly and self-interested long enough. But there are other kinds of settled.
reply
On L1, you have a network-wide game theory implementation where with each confirmation, new transactions are economically important to not get reorg'ed away, and every txo in a block helps every other historical txo to stay confirmed. There's network incentive.
On LN, ignoring flaws, all game theory is afaict between 2 parties, and heads-up changes things, because there is no longer the safety of the group. How is there any network incentive on LN grounded in game theory like on L1? Asking because you know better than I.
PS: apologies for the late reply - jetlag is being nasty to me atm.
reply
While I'm rambling: this is the coolest thing about lightning and I'm surprised no one has done something as creative with game-theoretic settlement (settlement without confirmation). I haven't tried, and I'm probably wrong, but I'd bet you could make a decent side chain that solves the mass exit problem by designing incentives for the mass exit-ers to fork the side chain together (and avoid going onchain).
reply
try 0-conf channels. Are really fun and interesting. Definitely an underrated thing by many people as I pointed in this old post: #909079
reply
If you are routing, it’s more desirable to have the extra sats in the channel vs onchain. True settlement is always onchain, but if you want to put those sats to work, you’re going to have to pay a miner to open another channel.
So in a sense that if that wallet/nodes purpose is routing, “settlement” in lightning sats —- increased channel balance is more desirable.
reply
reply
There also is the ecash component, true, final settlement is onchain. But moving from ecash to channel balance is a l3->l2 settlement too! Settlement is any layer transition, not just onchain.
reply
Well, in the examples here neither the payer nor the payee has a lightning channel. For instance, Fedi to Ark via Lightning. In this case who is actually "settling" on LN? Maybe the fedimint and the the swap provider?
As far as the mint user and the ark user, their trust models haven't changed no matter what is the state of the lightning channel that was used to link them.
So the mint has an open channel with someone and boltz has a channel with someone else (or maybe even they have a channel with each other). They are responsible for managing that channel just like any other lightning channel.
The people who want settlement have the same level of settlement (or trust) as they had when they started, and which, presumably they accept.
reply
Well, in the examples here neither the payer nor the payee has a lightning channel. [..] As far as the mint user and the ark user, their trust models haven't changed
Haven't they? Afaik if it's not your lightning channel (or you are a co-signer on it) then it's not your coin, even if just for a second; i.e. custodial?
reply
I would agree. I'm saying they were holding ecash tokens or ark vutxos to start and at the end they are holding ecash tokens or vutxos, not bitcoin.
So it doesn't seem like they are actually using lightning as a settlement layer. They are using it as a transport layer?
reply
holding ecash tokens or ark vutxos to start and at the end they are holding ecash tokens or vutxos, not bitcoin.
I think that as long as we recognize the "not bitcoin" part, this is fine, and that's my primary point: if it's not on L1, it's not bitcoin. The nasty thing I will add, and I truly believe this: also not on LN (sorry). This means that settlement can only be on L1 (if you're a purist) and everything other than L1 is at best
bitcoin * (1-<risk factor>)
.They are using it as a transport layer?
Yes. Between unsettled bitcoin-derivatives.
reply