17:59:06 #startmeeting anti-censorship meeting 17:59:06 Meeting started Thu Mar 19 17:59:06 2020 UTC. The chair is phw. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:59:06 Useful Commands: #action #agreed #help #info #idea #link #topic. 17:59:10 hi everyone, let's get started 17:59:18 here's our meeting pad: https://pad.riseup.net/p/tor-anti-censorship-keep 17:59:28 hi 17:59:30 hi 17:59:43 hi 18:00:10 hi 18:00:26 a brief semi-announcement: we now have a riseup pad that contains tickets that we want other teams to work on: https://pad.riseup.net/p/tor-anti-censorship-tickets-keep 18:00:46 oh nice 18:00:47 this is to help us keep track of work that's blocked on other teams 18:01:03 the link is in our meeting pad, fwiw 18:01:31 let's jump straight into our discussion section! 18:02:21 . 18:02:29 first, snowflake is in increasingly good shape and it's becoming a bit harder to figure out what tickets should be a priority 18:02:58 i think we all agree that we quickly want it to get to a point where it can be actually useful to people 18:03:35 so the question is: what tickets or issues do we have to solve to get there? 18:04:12 i had a chat about this with cohosh the other day and we believe that it would be helpful to write a blog post about snowflake's current state 18:04:18 "there" can be maybe "moved out of alpha TB releases" 18:04:51 ...with the goal of giving our users an idea of where we are, and encourage feedback. hopefully, this feedback will provide some guidance in terms of what we should prioritise 18:05:13 but we may want to do this after turbotunnel got merged. 18:06:08 cohosh: does the above make sense? anything to add? 18:06:33 yep that makes sense. basically, if we could enumerate all the tickets we want done before we consider snowflake out of alpha, that would be best 18:06:56 turbotunnel is pretty much the big thing in terms of usability 18:07:13 and a bunch of other tickets that depend on it 18:07:16 like #25723 18:07:38 and #25429 18:07:47 enumerating the tickets that are necessary to reach a post-alpha world sounds like a great plan 18:08:45 so from what I'm working on it's #33336 #33519 and whatever ticket I file soon about the turbotunnel merge 18:09:25 Hi, I'm a student interested in the Android Snowflake project on google Summer of Code, and I was wondering about the core difference between the Salmon bridge implementation project and the Snowflake project. 18:09:32 that makes sense. Par of this meeting can be use as a release one to figure out what to get in. 18:09:41 one game plan could be to wrap up these tickets, then blog about snowflake and encourage people to give it a shot in the alpha. we could then gather a round of feedback which will result in new tickets and once these are addressed, we could transition to tor browser stable. 18:10:05 More specifically, which one has a greater practical impact? 18:10:10 juggy: hi and welcome :) let's tackle this towards the end of the meeting 18:10:34 ok! 18:11:01 dcf1: phw: that sounds good to me. imo the other tickets are nice to have but not major game-changers from a baseline usability perspective 18:13:05 I'm still having troubles getting a working snowflake sometimes. I'm testing the #33519 packages from last night and it took about 20 tries to get a working 'flake 18:13:19 hm 20 tries seems way too high 18:13:29 I don't know if it's my NAT or what, sometimes it's faster, but not always 18:13:34 phw: maybe the one about "snowflake needs to stop using my network when tor has shut down all connections". maybe having some guidelines for what we want in a stable snowflake will help us put tickets on one side of the line or the other. 18:14:05 #21314 18:14:12 yep, that one 18:14:14 dcf1: i went through some old pcaps looking for clues as to why this happens and it looks like a NAT problem 18:14:27 but maybe a NAT problem for the proxies 18:14:40 it might even be related to the recent "tunnel this in that" design work dcf is doing. like, maybe when there is no socks request, we don't need to maintain some of the things. 18:14:57 STUN seems to work and sdp's are exchanged but the probes from one peer to another don't go through 18:15:10 I think #33519 will make #21314 possible. check the sessionManager -- we could add something that notifies it when a SOCKS connection begins and ends, and it could kill the session when there are zero. 18:15:23 dcf1: yep, i was hoping this 18:15:33 s/was/am/ 18:15:44 +1 on that ^ 18:16:03 dcf1: i'll make a ticket for investigating proxy failures 18:17:21 cohosh: should we... have something that rendezvouss with each snowflake and tells the broker not to give out the ones that it couldn't connect through? 18:18:06 arma2: i'm working on something like that... the problem is, it requires the proxy to be honest at the moment 18:18:19 and moving beyond this trust model requires a lot of work 18:18:32 and turns snowflakes into something more similar to tor relays 18:18:50 yep. that might be where we need to go, given the uncertainty for each snowflake whether it will actually help 18:18:54 we should consider doing this but i wouldn't consider it for MVP snowflake 18:19:02 that or if we have probabilistic snowflakes we teach the client to fetch k of them 18:19:17 but that requires us to know the expected success rate, so we're back to the original need 18:19:18 i'd rather right now see if there's an issue that's more easily solved and fix that 18:19:24 ok 18:19:30 arma2: yep, multiplexing will help with this 18:19:48 which is why #25723 should go on the mvp list 18:20:03 okay i made #33666 which should also go on the list 18:21:33 anything else? i'm pretty happy with this list 18:21:45 cohosh: how about we create a new ticket (like "create a snowflake mvp") and make all mvp tickets a child of this new ticket? 18:21:48 I suspect it may be something with my network, lik maybe I can only use IPv6 snowflakes, or something weird like that. I haven't investigated it much. 18:22:20 ah interesting 18:22:36 i didn't consider that ipv6/ipv4 only proxies could cause problems 18:22:37 how about #19001 for the meta ticket? 18:23:00 I suppose there's also mobile Tor Browser to think about, what side of the mvp line that falls one 18:23:01 oh, we already have one, nice 18:23:03 *on 18:23:07 nice 18:23:17 yeah I was so mad someone sniped #19000 from me 18:23:30 lol 18:23:39 we will reach 190000 eventually 18:24:49 we can put android tentatively on the mvp list 18:25:16 okay i can go through and parent those tickets after the meeting 18:25:31 thanks for the discussion, that was really helpful! 18:25:49 so, to make sure we're on the same page: #33336, #33519, #25723, #33666, and #21314 are on the list. is this correct? 18:26:06 ...and we will make them children of #19001 18:26:08 and #25429 18:26:13 thanks cohosh 18:27:01 and once we are done with these tickets, we will encourage users to give it a shot and provide feedback? 18:27:23 yup! 18:27:28 with a blog post 18:27:36 excellent 18:27:46 anything else that we want to discuss here? 18:28:05 not from me 18:29:28 so, next item is some sort of reading group 18:29:29 dcf1: you know the bug where snowflake-client crashed, and tor quietly logged a warning, and then things got really bad? did we ever learn anything more there? 18:29:38 it seems like we should file a tor ticket for this 18:29:49 (oh whoops, there is more on the agenda, i'll wait. i thought 'anything else' was the end) 18:29:57 +1 to reading group 18:30:26 +1 18:30:27 cohosh had the great suggestion that it would be useful for us to discuss papers, projects, events, in the anti-censorship world 18:31:04 i thought i was your suggestion lol 18:31:15 cohosh: i took it from your self-feedback :) 18:32:09 for example we could decide to read a paper over the next, say, two weeks and then discuss it. specifically, we could talk about the paper's assumptions, how we can benefit from it, how it could be improved etc. 18:32:30 same goes for software projects or censorship events. the goal is for us to stay up-to-date on what's going on outside of our team 18:32:38 fwiw, there already are reading groups at https://github.com/net4people/bbs/issues 18:33:58 sounds cool 18:33:58 how about we have such a reading group every two weeks? one week may be too little time and more than two weeks is probably too much 18:34:07 phw: it would be swell if somebody would write a several-paragraph (or more) summary of the paper and our thoughts on it, and get that posted somewhere. so there is something that comes out of the reading groups. 18:34:07 I would welcome other people staring threads for papers on bbs. There are about 4 more NDSS papers I hope to do eventually, but each one is a lot of work. It's about as much work as doing a review. 18:34:39 nice 18:34:46 and every two weeks somebody else gets to decide what we're going to read next 18:34:52 arma2: yes, good idea, and agreed 18:34:59 and whoever decides can write up the summary? 18:35:08 cohosh: +1 18:35:10 phw: you should also let mikeperry know it's happening, since i've wanted one of these for general tor research papers for a while too. where 'one of these' is a periodic blog post teaching people about a paper, and highlighting what we wish people would do next with it. 18:35:23 are you all thinking about an irc reading group? 18:35:34 we should let know internal that this is happening 18:36:06 gaba: good question. i'm not sure what's best. we could try irc and if it doesn't work use https://github.com/net4people/bbs/issues instead? or the other way around? 18:36:14 gaba: (a) why make it a sekrit, (b) careful, we sound like we're getting ambitious with scope here :) 18:36:36 yes, i'd rather not have our reading group on the nytimes front page tomorrow 18:36:39 too much responsibility 18:37:51 the benefit of irc is that it's more of an actual discussion, which is a bit harder on a forum. i'm inclined to try irc first. 18:38:01 thoughts? 18:38:21 i'm down to try it 18:39:26 then i'd suggest we just get started and adjust our process as we're doing it. any volunteers for our first reading group lead? :) 18:39:50 actually, i can volunteer 18:40:23 there are several ndss'20 papers that seem highly relevant to our work 18:41:03 how about we start reading sergey's paper? https://censorbib.nymity.ch/#Frolov2020a 18:41:11 and discuss it two weeks from now? 18:41:43 sounds good to me! 18:41:50 https://ntc.party/t/paper-summary-detecting-probe-resistant-proxies-ndss-20/405 18:41:50 there's a summary of that ^ 18:42:00 spoiler alert: there's already a thread on that on net4pps 18:42:03 links at https://lists.torproject.org/pipermail/tor-project/2020-February/002740.html 18:42:17 for NDSS papers that are also on censorbib 18:42:53 i suggest having goals to the discussion. for example, not just 'talk about what you found cool' and 'ask questions' but also: 'what future work does it make us want to highlight for the world?' and 'what specific actions should we do based on this paper'? 18:43:36 good point. i'll try to work out a few key points that we should work out for each paper 18:43:50 (or whatever goals we want. basically, yes we should use this chance to all get smarter about stuff, but let's not lose the opportunity to help this drive our roadmaps and goals.) 18:44:23 anything else wrt reading group? 18:44:56 was the idea to have the discussion during this meeting or in lieu of it? 18:45:24 arlolra: i would say during the meeting. 18:45:44 might that run a little long? 18:46:22 true. i think it's sensible to have a time limit, for things to not drag on forever. 18:46:38 we could maybe book an extra 15 minutes for meetings that involve a reading group? 18:47:09 or maybe we'll decide that irc isn't very useful for this, and switch to net4people 18:48:00 does that sound reasonable? 18:48:10 (another potential homework question while reading: 'which things in the paper did you find questionable?") 18:48:37 this sounds good to me, and we can always try and see what works 18:49:10 ok, then let's talk about sergey's paper two weeks from now. i'm excited to see how it goes :) 18:49:40 ten minutes left. let's move on to our 'needs help with' sections 18:49:57 i see #33637, #23226, and #33593 for cohosh 18:50:31 i'll grab #23226 18:50:32 okay uh 18:50:46 for #33637, idk i don't actually know much about copyright 18:51:00 ip stuff annoys me so i never really learned what to do in these situations 18:51:03 i've been realizing that you're not alone, and many people in tor don't understand copyright much 18:51:09 and we should..solve this somehow 18:51:16 like making a task force resource group that you can go to 18:51:27 but before that, we need more of tor to get on the same page about what exactly we want 18:51:37 which is an ongoing topic in some other circles 18:51:48 i feel like we've talked about this before and written it down, but i don't remember where 18:52:03 catalyst: if you find it, let us know 18:52:16 my understanding is that it's up to arlolra and serene about any updates/changes? since they're the current license holders? 18:52:21 cohosh: i can help you think through it after this meeting 18:52:27 arma2: ok 18:52:41 fwiw i'm good with leaving it but i think debian will want us to update the year 18:52:57 i think debian cares most about the license, not the copyright holders or year 18:53:07 and 'all rights reserved' is a crappy license. hopefully that's not actually the license. :) 18:53:21 no it's bsd-3-clause 18:53:27 great 18:53:53 okay then for splitting the snowflake repo #33593, we can continue discussion on the ticket 18:54:00 but i'd like consensus 18:54:16 i think we still need to hear from dcf1 on that 18:54:35 I'll comment on the split ticket 18:54:40 dcf1: thanks! 18:54:46 ok that's it from me 18:54:53 phw: thanks for taking that review 18:55:02 https://en.wikipedia.org/wiki/All_rights_reserved#Obsolescence 18:55:18 ok, any other reviews? 18:55:27 i already coordinated bridgedb reviews with agix 18:55:57 i think juggy had some questions 18:56:09 right, shall we continue in #tor-dev? we're almost out of time 18:56:10 ya haha 18:56:14 sure 18:56:16 sorry it took so long juggy :) 18:56:23 and thanks for your interest and patience 18:56:46 let's wrap it up for today. thanks everyone for attending! 18:56:53 #endmeeting