00:01:54 Ooh 00:02:09 Hooray, ring member extraction is done! 00:02:26 Now to run iterations for up-to-date chain reactions 00:02:47 The preprint suggested a nontrivial number of post-CT inputs, but indexed them in a way that wasn't helpful 00:02:55 So I'm getting a fresh dataset that's indexed better 00:03:06 This way we can extract meaningful data from any similarities 00:03:18 I'll post my code when complete 00:03:27 (not the datasets... they're too big) 02:36:12 I finished an initial analysis of all deducible transactions up to block 2M 02:37:02 All deducible transactions after the confidential protocol changeover spend only pre-confidential funds, and are therefore "transition" transactions or (in limited cases) non-confidential 02:37:32 This explains the traceability preprint's claim that a small but nonzero number of post-changeover transactions are traceable: they didn't account for transition transactions 02:37:56 If you remove these from analysis in order to count only confidential transactions, the number of traceable (partially or fully) transactions drops to zero 02:38:41 Additionally, this means that the preprint's claim that guess-newest is 30% effective post-changeover is probably competely wrong, if it's only using transition transactions that can't use the gamma selection 02:39:05 The authors did update their preprint on May 29 to make some corrections, but I can email them again with this new information 02:46:11 hot dog dude 02:52:13 So to clarify, the preprint's numbers on traceable transactions after the changeover isn't wrong, but I think it should mention that the identified transactions aren't confidential transactions, which is an important distinction 02:52:39 The conclusions about guess-newest are almost certainly wrong 02:53:23 s/isn't/aren't 02:53:23 sarang meant to say: So to clarify, the preprint's numbers on traceable transactions after the changeover aren't wrong, but I think it should mention that the identified transactions aren't confidential transactions, which is an important distinction 02:53:36 good bot 02:54:22 Tomorrow I'll run an analysis on effective anonymity sets 02:55:16 Many thanks to gingeropolous for the use of a very fast machine to run this analysis! 02:55:29 sounds good