From lf-lists at mattcorallo.com Wed Sep 1 12:44:35 2021 From: lf-lists at mattcorallo.com (Matt Corallo) Date: Wed, 1 Sep 2021 05:44:35 -0700 Subject: [Lightning-dev] Do we really want users to solve an NP-hard problem when they wish to find a cheap way of paying each other on the Lightning Network? In-Reply-To: <5s7jAga4AiiK4zp-WlEFm5rm611A6HgHPmzJR3AbziWv_qwLHbGms7Ot0VCCrCyq4hIrFFTR0ead7eM-e0dEr7TztdJXCoMtWs7iJHsJYvs=@protonmail.com> References: <5s7jAga4AiiK4zp-WlEFm5rm611A6HgHPmzJR3AbziWv_qwLHbGms7Ot0VCCrCyq4hIrFFTR0ead7eM-e0dEr7TztdJXCoMtWs7iJHsJYvs=@protonmail.com> Message-ID: > On Sep 1, 2021, at 00:07, ZmnSCPxj wrote: > > ?Good morning Matt and all, > >> Please be careful accepting the faulty premise that the proposed algorithm is ?optimal?. It is optimal under a specific heuristic used to approximate what the user wants. In reality, there are a ton of different things to balance, from CLTV to feed to estimated failure probability calculated from node online percentages at-open liquidity, and even fees. > > It may be possible to translate all these "things to balance" to a single unit, the millisatoshi. Indeed, in practice this is what we all do today. My point is less that you cannot create a single unit out of all the various things you consider and more that doing so involves some heuristics on the part of the application developer. There is no ?correct? or ?optimal? answer to how to do this, only various designs different folks have. How you balance competing costs may lead to different score units (eg instead of msat, probability of success) and that?s fine, neither is provably better than the other. Matt