--- layout: post title: Logs for the Monero Research Lab Meeting Held on 2019-10-14 summary: Sarang work, Surae work, and miscellaneous tags: [dev diaries, community, crypto, research] author: el00ruobuob / sarang --- # Logs **\** GREETINGS **\** howdy! **\** Hello **\** Small crowd today, apparently **\** Even so, we carry on **\** Let's move to ROUNDTABLE **\** I've been working on a few things this past week **\** First is getting caught up with the usual literature review **\** Second was finalizing things for World Crypto Conference and some background research associated to that **\** Third was getting balance proofs working in Triptych, which is now successful **\ \** hello **\** This means that Triptych now supports a single proof showing all spends, correct key image construction, and balance **\** nice! **\** How about you, suraeNoether? **\** i've been furiously debugging my matching code as my primary task. there are some persistent problems. i wanted to finish this weekend but it didn't happen **\** Earlier you had indicated some known bugs... are these the same? **\** no... every problem i solve reveals like... a small handful of new bugs, but the newer and newer bugs are becoming less frequent and less severe **\** it \*feels\* like there's a single problem lurking that will cause the house of cards to stop falling down **\** i'm very close. **\** i really wanted it to be today **\** i'm taking a break later today to read sarang's WCC talk (sorry for the delay on that) and I am taking a break later today to work on \*literally anything else\* **\** i'm very frustrated with this project **\** Are the known bugs documented anywhere, so others might assist you? **\** i'm sure a lot of community members are also frustrated, but i this is nearing completion **\** no **\** "test X not working for unknown reason" is not a helpful document to write **\** Hmm ok **\** Well, I selfishly hope you will take time off that project today and review my talk :D **\** Perhaps it will also help you clear your head **\** Does anyone else have interesting research to share as well? **\** In that case, let's go ahead and discuss ACTION ITEMS first, and then any lingering questions **\** First, I have an efficient verifier for the inner-product argument in IACR/944 that I've been meaning to implement in kenshamir[m]'s Rust code, which will be useful for benchmarking... that's in progress but with some algebra problems that I'm working out **\** Second, Triptych needs plenty more work: key aggregation, better Fiat-Shamir challenges, and some questions on proof elements and efficiency **\** Third, I want to see if it's possible to backport some of the new RCT3 changes to the older version without using spend aggregation, to check the resulting efficiency **\** and that's about it for now **\** suraeNoether: ? **\** pushing this commit once my code is flowing. reading your WCC talk. catching up on tryptychychch **\** It definitely remains to be seen how efficient we can make Triptych... but as I mentioned last week, the underlying changes to the Groth proving system are very interesting regardless **\** and, as before, there is no security model for it yet **\** All righty, are there other questions on research? **\** This meeting has gone quite quickly **\** Oh, one note about what Isthmus brought up last week regarding transaction keys and subaddresses **\** It is apparently still the case that transactions to only standard addresses retain a single transaction key **\** Mandating separate transaction keys for all outputs would add 32 bytes to each additional output **\** Standard = 4? **\** but we're already saving > 32 bytes per output after the last change to the Pedersen mask format anyway **\** Could there be a way to deterministically generate keypairs in such a way that the sender generates the secret keys from a seed, the recipients generate the pubkeys ? I think Bitcoin has such a scheme for generating addresses. **\** And hopefully the seed is \<= 32 bytes :) **\** Well, a big selling point of subaddresses is the efficient scanning across all addresses at once **\** Isthmus: only need to read up a few lines **\** Would such a scheme invalidate the efficient scanning ? It seems doubtful since the tx keys are currently arbitrary. **\** How much effort is it to scan and see what proportion of transactions are only to standard addresses? **\** sgp\_: to get a distribution of how common subaddresses are? **\** @sgp\_ I think that @n3ptune accidentally did that recently **\** Lemme see if the plots are on GitHub anywhere **\** sarang: essentially yes **\** Presumably this would be affected by which large players (like exchanges) support them **\** Thanks Isthmus **\** https://github.com/noncesense-research-lab/tx\_extra\_analysis/blob/master/tx\_extra\_viz.ipynb **\** 404 **\** Oh, private repo. Lemme grab the juicy parts **\** This might be the relevant one **\** https://usercontent.irccloud-cdn.com/file/LgrrzOIS/image.png **\** I suspect the diagonal is transactions that include a subaddress, while the horizontal bands are primary-only **\** Though I'm open to alternate interpretations **\** Oh I get it. The fast lookup would still exist, but verifiers would have to generate pubkeys, and \*that\* might be slow. **\** Thanks **\** If that is the case, then I can slide a window over time and calculate fraction of transactions that appear to include no subaddresses **\** I'm not the one who can say yes or no to that :/ **\** Probably worth bringing up at the next dev meeting to see what others think of it **\** It is trivial to know whether >= 1 subaddress was used as an output in a tx. **\** If that was the question... **\** Oh wait. Maybe not, there's some funky going on with change being treated differently... **\** A more meta question: how did this happen? What could have been done differently to help prevent this from happening? **\** That's probably a question for someone like stoffu who was more directly involved in the code **\** I suspect space saving was one consideration **\** knaccc too? **\** but it's quite minor for the most part **\** @sgp\_ meta answer: we rolled out a new feature that: **\** 1) you could tell use from blockchain as external observer **\** 2) was optional **\** Either one of those alone is ok, but together we end up in this situation. **\** I always assumed 1 wasnt the case. I was very misinformed and thus misinformed others **\** Yeah, I think we're all just putting 2+2 together on that now **\** OK, something to discuss at next dev meeting, then **\** Are there any other topics to discuss for this meeting? **\** Oh yea, lemme grab a link **\** The CryptoEconSec paper by hasu and all is very interesting, and parts are relevant to both Monero and our lock time conversation **\** \*et al **\** I definitely recommend reading it. Very approachable. **\** Here's the writeup: https://uncommoncore.co/research-paper-a-model-for-bitcoins-security-and-the-declining-block-subsidy/ **\** And here is my analysis: https://twitter.com/Mitchellpkt0/status/1183581226357014528 **\** I won't rehash it all here. Just take a pass through on your next commute. :- ) **\** Thanks Isthmus **\** Any last questions before we adjourn and continue discussions? **\** Righto, thanks to everyone for attending!