16:00:22 #startmeeting tor anti-censorship meeting 16:00:22 Meeting started Thu Aug 13 16:00:22 2020 UTC. The chair is cohosh. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:22 Useful Commands: #action #agreed #help #info #idea #link #topic. 16:00:40 i guess we can get started 16:00:43 Heyy o/ 16:00:45 hi! 16:01:27 here is our meeting pad: https://pad.riseup.net/p/tor-anti-censorship-keep 16:02:05 feel free to add things to the agenda 16:02:06 hey! i am in the background (if one meeting is good, two is better, right) if you need anything from me :) 16:02:33 the first announcement is that we have a temporary bug reporting pad at https://pad.riseup.net/p/tor-anti-censorship-bugs-keep 16:02:44 I am checking the public bug pad https://pad.riseup.net/p/tor-anti-censorship-bugs-keep and there is nothing new there. 16:02:51 thanks dcf1! 16:03:38 moving on to discussion 16:03:58 we had someone post a mail to the anti-censorship-team mailing list asking about the metrics timeline 16:04:04 Regarding the metrics timeline, this is my fault 16:04:10 which used to be here: https://trac.torproject.org/projects/tor/wiki/doc/MetricsTimeline 16:04:15 I meant to start a discussion on metrics-team earlier about finding a new home for the timeline 16:04:26 https://lists.torproject.org/pipermail/metrics-team/2020-August/001153.html 16:04:34 sorry, forgot about the time 16:04:44 hey phw! 16:04:50 hi all o/ 16:04:58 dcf1: no worries! we all forgot >.< the trac change left a lot of loose ends 16:05:02 I see the new wiki page at https://gitlab.torproject.org/tpo/anti-censorship/team/-/wikis/Censorship-Events, but I have some reservations 16:05:26 I don't like the name "Censorship Events" because that's not all the timeline is. It's supposed to be anything that might affect metrics graphs. 16:05:49 That includes censorship, but also things like new directory authorities, sometimes new Tor Browser releases, things like that. 16:05:49 dcf1: yup, that makes sense. i realized that after posting and why it was on the metrics wiki in the first place 16:06:08 cohosh: i added an item to the agenda, about thinking about things that our twitter crowd would be interested to hear. i put it in 'discussion', feel free to move it to a smarter place in the pad 16:06:25 another consideration about where to put the events wiki page has to do with gitlab permissions 16:06:42 But another important aspect is that the metrics team needs to be able to parse the plaintext of the page so they can put the information on graphs. I'm guessing there is no parser for the new wiki markup yet. 16:06:55 yeah 16:07:11 I've had a few items to add to the timeline, but I have just been keeping them in a local repo. But again, I should have started this conversation earlier. 16:07:28 For example, maybe it's easier to keep a CSV file in a repo somewhere. 16:07:38 ah that's a really good point 16:07:48 and then anyone can file a merge request 16:07:51 gitlab wikis are git repos, right? 16:07:56 The idea behind having it on Trac wiki was to permit non-insiders to add events, but that's not so easy with the gitlab permissions, as you say. 16:08:29 arma2: afaik no, you cant submit a merge request to change the wiki 16:08:45 or if you can it's not obvious to me how this is done 16:08:59 I also was really surprised that kuroneko tried to edit the page; I thought that I was practically the only one doing it. I was going to suggest even a private repo until I saw that. 16:09:34 i'm glad more people are interacting with it :D 16:09:56 * phw didn't even know that the "related events" section on metrics.tp.o was automatically generated 16:09:57 So I'm planning to start a thread on metrics-team asking what works best for them. 16:10:20 Yeah "related events" comes straight from the Metrics Timeline. 16:10:25 cool, sorry about jumping in and responding >.< i guess we'll have to damage control that 16:10:43 and tell the mailing list the updated way of interacting once we sort that out 16:10:47 No problem. If it gives people somewhere to contribute events in the meantime, that's fine. 16:11:50 okay so it seems we have an action item there owned by dcf1 16:12:04 thanks for doing that 16:12:50 we can move on to the next discussion item 16:13:09 Cool stuff happens in anti-censorship-team-land, and we have a twitter audience who would love to learn about them. So one of the items to consider periodically should be: is there something compelling to tweet about? This period includes Belarus, ESNI, Turbotunnel video, more. 16:13:19 arma2: do you have more you want to say on that here? 16:14:09 i wonder if one of us should get access to our twitter account. i dislike having to coordinate simple things, so i end up not doing them. the biggest reason i've been consistent with blogging our monthly report is that i don't have to coordinate with anyone 16:15:00 twitter has some feature that facilitates account sharing without having to share passwords iirc 16:15:19 I have barely been keeping up with the Belarus situation. I was meaning to start a thread on BBS just with a few links that have appeared on #ooni. 16:15:31 gus and alsmith are the ones to coordinate on the twitter side. usually i just get them a couple of sentences and then they take it from there. 16:15:32 what if coordinating is just sending comms@ an email that says "Please tweet this: "? 16:15:37 cohosh: yes, exactly thta 16:16:00 Under "Interesting links", it's nice to see so many recommendations of snowflake in the reddit thread: https://www.reddit.com/r/europe/comments/i82v45/surge_in_tor_activity_from_belarus/ 16:16:07 cohosh: i think i could live with that. it's the back-and-forth that typically bothers me 16:16:41 phw: my guess is that isa will be unexcited to have people from every team just twittering as @torproject in an uncoordinated way 16:16:48 dcf1: I've been seeing many Snowflake links lately. 16:16:52 so either way, the coordinating will need to be happening 16:16:54 on Reddit* 16:18:13 "Now, boys, the network has a problem with some of your lyrics. Do you mind changing them for the show?" "Our lyrics are like our children, man. No way." 16:18:27 * dcf1 imagines coordinating tweets 16:19:01 lol 16:19:48 cohosh your suggestion of emailing comms w/ please tweet this is good. there is generally minimal or no back and forth when things like this are suggested. we could try it and if it’s not working, we can re-asses? 16:20:03 alsmith: sounds good to me! 16:20:51 great! ty 16:20:55 woo 16:21:03 i guess we can pivot into the next discussion point about figuring out what's going on in belarus 16:21:14 phw: do you want the mic for this one? 16:21:41 i only have a brief update from my side: i've been trying to reach out to folks from belarus who could help us with measurements 16:22:12 could we use twitter to make an open call for people to run measurements for us inside belarus? 16:22:16 i found two people on twitter. the time difference makes it a bit difficult but i'm going to talk more with one of them later today 16:22:30 nice! 16:23:14 is the plan to ask them to run emma? 16:23:28 that's an option, but i prefer a low key approach for now 16:23:47 yes, and to learn more about what bridge types work 16:24:33 a low key approach being reach out to people individually and not through an open call? 16:24:40 for example, all of our default bridges have high usage counts from belarus but arma2 pointed out that these people may not actually be able to use these default bridges because of the way the counting happens 16:25:28 correct. no need to rally the internet if we can work with a handful of people and avoid attention 16:25:57 (tor relays and bridges count a user when the relay/bridge finishes queueing the bytes toward the client. if the connection gets cut by dpi or something after that, it still counts as a user. plus, the user will then come back and try doing it again, because it didn't get its consensus. 16:25:58 ) 16:26:21 ah good point 16:27:42 (might be working great. might not be. needs more input from places other than the metrics data set) 16:27:44 that's it from my side. i just wanted to keep y'all in the loop 16:28:13 thanks phw! 16:30:05 any other discussion before we move on to our needs help with? 16:30:54 I lost track of when the GFW started blocking connections from Tor exits 16:31:21 I believe it was in 2016 or 2017, but I can't find any notes I may have made. Somehow I remember it happening during https://www.bamsoftware.com/proxy-probe/ research. 16:31:39 I mention it because today I can access China-based websites using Tor again. 16:31:59 If anyone has a more specific memory of when the blocking of inbound Tor traffic to China started, let me know. 16:32:49 woah 16:33:54 btw currently on the showflake bridge ("by=" is Belarus) 16:33:58 bridge-stats-end 2020-08-12 22:52:58 (86400 s) 16:33:58 bridge-ips us=24,ir=16,??=8,au=8,bo=8,by=8,ca=8,cn=8,fr=8,gb=8,gu=8,hu=8,in=8,jp=8,ma=8,nl=8,ru=8,tr=8 16:34:11 dirreq-v3-ips ??=8,au=8,by=8,ca=8,cn=8,fr=8,gu=8,hu=8,in=8,ir=8,jp=8,ma=8,nl=8,ru=8,tr=8,us=8 16:34:14 dirreq-v3-reqs ??=664,ca=208,ir=16,ru=16,au=8,by=8,cn=8,fr=8,gu=8,hu=8,in=8,jp=8,ma=8,nl=8,tr=8,us=8 16:34:29 So, nonzero but low 16:34:35 it's binned to 8 so we have at least 1 user 16:36:03 dcf1: i think china has been doing bidirectional blocking the whole time. the question is when they shifted from ip:port blocking to ip blackholing 16:36:30 arma2: no, this is something different. It was not bidirectional with respect to Tor before 2016. 16:36:45 We know within a month when the switch from ip:port to ip blocking happened, let me find it. 16:39:36 October 2016 https://gitlab.torproject.org/tpo/anti-censorship/bridgedb/-/issues/28655#note_2599830 16:39:54 By coincidence observed by two different research groups. 16:41:36 The recent ESNI block is IP:port 16:43:35 dcf1: ok. when i'm done with this other mtg, i will try to hunt down the "can we try to extend to china and thus learn that a bridge got blocked" ticket 16:44:29 * cohosh waits another minute to see if there is more discussion before assigning reviews 16:46:26 https://gitlab.torproject.org/tpo/core/tor/-/issues/1851 here it is 16:46:33 i need a review of snowflake#21314 16:47:30 arma2: it may not be strictly about bidirectionality, but also bridges vs. exits. Exits inbound to China were formerly not blocked, then for the past 3/4 years have been blocked, then today are apparently not blocked again. 16:48:05 interesting! 16:49:22 okay, let me review #21314 16:49:32 thanks! 16:50:12 cohosh: did you get the azure cdn email this time? 16:50:29 dcf1: nope :/ 16:50:36 did you resend it in the last few days? 16:50:45 I sent you one from my own account too 16:51:25 i commented on the ticket, that I followed the link you emailed me yourself and it prompted me to create an account i thought i'd already created 16:51:33 Oh, I see now you've already replied at https://gitlab.torproject.org/tpo/anti-censorship/pluggable-transports/snowflake/-/issues/30510#note_2705578 16:51:38 ah yeah 16:52:07 On 04 Aug I sent you a PGP mail from david@bamsoftware.com 16:52:33 Subject: Re: Snowflake | Share access to the Snowflake broker domain front CDN configuration (#30510) 16:52:43 yup that's the link i tried 16:52:55 Oh I understand now 16:53:10 I don't know what's going wrong then. 16:53:26 now that i've (re)created an account, can you try sending the invite link one more time? 16:53:32 okay 16:55:11 anything else before i end the meeting? 16:57:22 * phw shakes head 16:57:30 #endmeeting