00:48 <kanzure> "Bitcoin coin selection with leverage" https://arxiv.org/pdf/1911.01330.pdf

00:48 <kanzure> https://engineering.cmu.edu/thailand/_files/documents/tendrilstaller-block-delay-attack-in-bitcoin.pdf

01:40 <harding> I don't quite get that "coin selection with leverage" paper. IIUC, the basic idea seems to be that you have two payment requests; you attempt to generate a changeless transaction to fulfill the first request and store whichever attempt produces a change output nearest the value of the second payment request. That way, if you fail to generate a changeless transaction for the first request, you can at least ensure that you generate

01:40 <harding> a changeless transaction for the second request by sending the saved near-match. That seems smart, but I would think the spender would save money by just batching the two payment requests into the same transaction, since that would eliminate the overhead of a second transaction and since they already found a set of inputs whose amounts equal the total desired payout amounts (including fees).

12:47 <waxwing> i found this interesting: https://gist.github.com/SmeltingBitcoin/de48d1994f2117757475855c06eb22ba

12:47 <waxwing> i would write comments here but i already did at the end of the gist

12:49 <waxwing> 'do a coinswap but one side's payment is miner fees so history is destroyed' is i think the TLDR