13:59:41 <karsten> #startmeeting Measurement Team meeting
13:59:41 <MeetBot> Meeting started Thu Oct 22 13:59:41 2015 UTC.  The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:59:41 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
13:59:49 <karsten> hi. who's here for the measurement team meeting?
13:59:58 <tomlurge> hi!
14:00:02 <karsten> hi tomlurge!
14:00:13 <karsten> here's the agenda: https://pad.riseup.net/p/zUNzEIFRq5S4
14:00:26 <karsten> please feel free to add topics.
14:02:00 <Letty|Katja> hi
14:02:12 <karsten> hi Letty|Katja!
14:03:56 <karsten> Letty|Katja: while we wait for the agenda to be finalized, would you like to say a sentence or two about what you're doing that is related to tor measurements/metrics?
14:05:11 <Letty|Katja> sure.
14:05:21 <karsten> cool. :)
14:08:37 <karsten> Letty|Katja: I hope you don't mind me utilizing this fine feature of IRC where we can all talk at once and still read what we were writing afterwards. I'd start with the team announcements in that case.
14:09:18 * karsten goes ahead and hopes not to come over as rude..
14:09:44 <karsten> team renaming: tomlurge suggested to call the team metrics team rather than measurement team.
14:09:55 <karsten> I asked around and heard a few pros and one con.
14:10:19 <karsten> con: the team is not just about Tor metrics but also about OONI, exitmap, etc. therefore a different name.
14:10:29 <karsten> pros: shorter name, better captures the role of the team.
14:10:54 <karsten> in general, people were in favor of the change. so I'd suggest if I don't hear strong objections by the end of this meeting, I'll rename it.
14:11:10 <karsten> does that make sense?
14:11:25 <phw> i'm also here now, sorry for being late.
14:11:30 <karsten> hi phw!
14:12:25 <karsten> I think there might be others joining us late, so I'll leave this question out here. strong objections to metrics team? speak up!
14:12:34 <karsten> team mailing list:
14:12:47 <karsten> there was a suggestion to create one at the berlin dev meeting.
14:13:22 <karsten> it would be used as "staging area" for tor-dev@ discussions where people would solicit feedback by the team before writing to the huge list with all the smart people on it.
14:13:35 <karsten> it would also be used for team-only discussions. though it would be a public list.
14:14:04 <karsten> I suggest to create this list once we have settled on a name. it would be metrics-team@lists.torproject.org, assuming we change the name.
14:14:25 <virgil> +1
14:14:26 <tomlurge> question: do the other 3 teams have their own mailing lists?
14:14:29 <karsten> similarly, I'd say please speak up if you have concerns against this idea.
14:15:08 <karsten> tomlurge: fine question. I don't think so, but they were not opposed to the idea of us being the first to create one.
14:15:41 <karsten> hi virgil, btw. :)
14:15:52 <virgil> oh hai
14:16:25 <karsten> okay, cool, moving on, because we need to end this meeting at 15:00 utc, because I scheduled another meeting for exactly that time. ;)
14:16:39 <karsten> team landing page: we should have one, and I'd like to start write one.
14:17:07 <karsten> it would be on the wiki and contain things like: role of the team inside tor, next meeting, link to mailing list, link to roadmap document, upcoming roadmap items, etc.
14:17:28 <karsten> but I wanted to ask: what else would people expect to read on such a team landing page?
14:17:39 <karsten> maybe we can brainstorm a bit about this here?
14:18:07 <tomlurge> project activity
14:18:33 <karsten> as in, which projects are most active at the moment?
14:18:46 <tomlurge> yes, like github :-)
14:18:46 <karsten> if so, we have something like that on the volunteer page. let me find it.
14:19:10 <karsten> https://www.torproject.org/getinvolved/volunteer.html.en#Projects
14:19:51 <karsten> just asking to be clear what you mean. didn't mean to discuss now.
14:20:24 <tomlurge> yes, taht’s the direction, although github has nice graphs and icons too
14:21:09 <Letty|Katja> tomlurge: do you mean the profil or startpage of github?
14:22:36 <tomlurge> letty|katja: i mean the part where there are graphs and names of people
14:22:50 <karsten> I think we're rather limited by trac here. but we could do something manual and try to update that every now and then.
14:22:50 <Letty|Katja> a combination of both could be useful (how active the team is with commits and maybe what kind of bugs / todos exist) and a links with additional info's (onionoo, ooni, metrics..)
14:23:23 <karsten> what we could do is add a table of related tickets.
14:23:38 <clv> hi there, sorry for being so late, i've already read the backlog
14:23:43 <karsten> hi clv!
14:24:28 <karsten> okay, I think that's a good start. I'll start writing something and then ask you all to review that.
14:24:38 <karsten> let's move on.
14:24:46 <karsten> - analytics server (tomlurge)
14:24:53 <karsten> tomlurge: want to give an update on that?
14:25:01 <tomlurge> yep
14:25:13 <tomlurge> we decided on what software to use
14:25:22 <tomlurge> have set up a server and installed that software
14:25:40 <tomlurge> are now working on transforming collector data to JSON
14:25:47 * karsten throws in a link: https://github.com/tomlurge/mteam
14:26:04 <tomlurge> will then ingest the JSON data into the Hadoop machinery
14:26:20 <tomlurge> and then announce it’s availability on tor-dev
14:26:29 <karsten> yay!
14:26:30 <tomlurge> hopefully within the next 2 weeks
14:26:35 <tomlurge> end
14:26:48 <karsten> I started working on the collector data transformation.
14:27:15 <karsten> I ran into a few issues where I don't know yet how to represent data in json in a way that they will be easy to analyze later.
14:27:38 <karsten> but I guess I'll just give you some samples and you tell me whether that works for you or not.
14:28:08 <tomlurge> yep.
14:28:17 <karsten> do you think we'll have first results in two weeks from now?
14:28:22 <Letty|Katja> sometimes is csv better. some software is not allowing json
14:28:42 <tomlurge> letty|katja: i checked that
14:29:02 <tomlurge> karsten: 2 weeks is ambitious but doable i hope
14:29:13 <karsten> great! :)
14:29:33 <tomlurge> karsten: let’s talk about modelling issues afterwards
14:29:47 <tomlurge> (after the meeting)
14:29:56 <karsten> tomlurge: let's try to find a time tomorrow. there's another meeting right after this one.
14:30:02 <karsten> but yes, happy to talk more.
14:30:24 <tomlurge> karsten: tomorrow is fine
14:30:58 <karsten> Letty|Katja: regarding csv, I think that would limit us quite a bit in what data we can provide, right?
14:32:12 <Letty|Katja> karsten: true. csv grows too much by entries
14:33:12 <tomlurge> letty|katja: JSON is accepted by all Hadoop softwares I chose, also by MongoDB, PostgreSQL and Tabulator. What tools would be nice to have that don’t support JSON?
14:33:36 <virgil> with tomlurge.  JSON is the new csv.
14:34:41 <Letty|Katja> tomlurge: there was tableau mentioned in the proposal. the desktop software just allows csv
14:34:54 <tomlurge> we can generate csv from JSON. csv would have to be generated anyway since its flat and most collecTor data is hierarchically organized
14:35:32 <tomlurge> letty|katja: oh, then I misread that. but Tableau can be used on top of some Hadoop tools, so that would be covered
14:36:10 <Letty|Katja> tomlurge: ok great
14:36:47 <karsten> okay, let's stick with json for now, unless we find out that that would get us into trouble later.
14:37:06 <karsten> should we move on to the next topic?
14:37:19 <karsten> sorry for rushing, but we have many interesting topics today. :)
14:37:40 <karsten> - metrics data visualization (Letty)
14:37:49 <karsten> Letty|Katja: want to tell us a bit about that?
14:37:58 <Letty|Katja> i'm currently analysing the data about the performance of downloaded files via tor with tableau. sketched some idea's and next step would be programming a interactive version.
14:38:09 <Letty|Katja> i can add some links to the pad
14:38:18 <karsten> please do!
14:40:38 <phw> Letty|Katja: (i've been experimenting with visualising Sybils in tor.  it would be useful to hear your opinion on that if you are good at visualisations.)
14:40:41 <virgil> I do love tableau
14:41:23 <Letty|Katja> first pic: downloads that failed on different sources ( you can see a total number)
14:41:24 <Letty|Katja> second: same with multiple axis (easier to see individual source)
14:41:27 <anonym> how long, usually, does it take for tor -rc .deb:s to be built from the source release?
14:41:40 <anonym> i.e. when can I expect a .deb for tor 0.2.7.4-rc?
14:42:16 <weasel> when it's ready.
14:42:50 <Letty|Katja> third pic: could be a possible detail chart for one source
14:42:50 <Letty|Katja> fourth: all servers by the time (as an information, maybe it correlates failures and relays)
14:43:24 <Letty|Katja> phw: i don't know sybils, can you tell me more?
14:43:46 <Letty|Katja> virgil: i hate it.. programming is sometimes easier than click the stuff together..
14:44:27 <Letty|Katja> my idea for the download failures is having multiple charts that are connected for discovering the data
14:44:28 <virgil> Letty|Katja: lulz.  I'm just n00b.
14:44:52 <karsten> Letty|Katja: you used already-aggregated data for these visualizations. would you also be interested in aggregating data yourself or using aggregated data from the analytics server experiment?
14:45:16 <karsten> (assuming that tomlurge wouldn't defend results to make his own visualizations. ;) )
14:45:52 <karsten> in particular I'm thinking of interesting data about bridges. we have almost zero statistics about them.
14:46:22 <tomlurge> tomlurge would probably give away some of his preciuos aggregations
14:46:27 <karsten> heh
14:46:36 <Letty|Katja> karsten: i can also use data from the server, i was just exploring in the easy way
14:46:45 <karsten> makes sense.
14:47:34 <karsten> Letty|Katja: how about we brainstorm possible visualizations on the yet-to-be-created mailing list?
14:47:55 <karsten> by going through the available data and thinking of good visualizations that we don't have yet?
14:48:05 <karsten> btw, do you know https://collector.torproject.org/index3.html ?
14:48:14 <Letty|Katja> karsten: +1 would be great, i'm not that familiar with all the data like you
14:48:37 <karsten> that page has a lot of stuff, maybe too much, but maybe it's a start.
14:48:47 <Letty|Katja> yes and great there is a table of contents :)
14:48:50 <sstx31> Hi i am receiving an error from a server... gpg --keyserver keys.gnupg.net --recv 886DDD89
14:49:01 <karsten> Letty|Katja: heh. tomlurge's suggestion.
14:49:12 <sstx31> gpgkeys: HTTP fetch error 7: couldn't connect: Connection timed out
14:49:28 <karsten> okay, I think we should move on to the next topic. let's talk more on the mailing list or tomorrow if you want!
14:49:35 <Letty|Katja> karsten: do you need some help with the css for that side?
14:50:01 <karsten> oh, I totally do. tomlurge already sent suggestions, but more help wouldn't hurt.
14:50:29 <karsten> thanks for the offer!
14:50:32 <Letty|Katja> is there a repo?
14:51:02 <karsten> yes, but this page is not in it yet. I'll create a branch tonight.
14:51:06 <karsten> and send you the link.
14:51:11 <Letty|Katja> ok thx
14:51:15 <tomlurge> repo would be nice!
14:51:23 <karsten> - Roster update (virgil, Sean)
14:51:28 <virgil> we exist.
14:51:33 <karsten> cool!
14:51:36 <karsten> how's it going?
14:51:48 <sean> Hi karsten, will make this quick given the time!
14:51:55 <karsten> hi sean!
14:51:56 <virgil> (1) public repo! https://github.com/seansaito/Roster
14:52:20 <virgil> (2) Trying to get OnionMap to display a single family. If you know opi would appreciate an intro and nudge. https://opi.github.io/onionmap/
14:52:26 <karsten> great. I didn't have the chance to look, but thanks for making this available.
14:52:46 <virgil> (3) Tinkering with RouteViews's AS-peering graph to make a rudimentary AS-diversity metric. This metric is intended to later be replaced by something more sophisticated, but it gives something to whet the palette.
14:53:06 <virgil> (4) definitive listing of all badges is coming!
14:53:11 <virgil> (5) Proposal: Roster will supersede the (unmaintained) Tor Weather.
14:53:52 <karsten> re 2), I'm not sure. where was that announced?
14:54:32 <karsten> re 3), curious to learn more, possibly on the new mailing list.
14:55:15 <karsten> re 4), curious about this, too. speaking of, when I looked at a version of roster I wondered if you can add more explanations to the site why somebody received a badge.
14:55:34 <virgil> (3) current plan is just to use a graph-theory metric that makes sense to me.  Probably something like minimum mode betweenness centrality between all pairs of nodes
14:55:59 <virgil> no wait: just the mode betweenness centrality between all pairs of ASs with Tor-nodes
14:56:05 <karsten> re 5), sean and I talked about that in berlin, and I think it makes sense. just don't spam poor relay operators like weather did/does.
14:56:23 <virgil> (5) can do.
14:56:42 <virgil> (2) Just saw OnionMap on tor-talk.  It's better than our current GoogleMaps thingy
14:57:15 <virgil> (4) Yeah we don't really know what kinds of badges we want.  This will be a sort birds-eye view of possible badges
14:57:35 <karsten> okay, I don't think I know this opi person. I'd suggest to say hi anyway.
14:58:01 <karsten> sounds good, re 4).
14:58:15 <karsten> (1 minute until next meeting.)
14:58:21 <virgil> okay I'm done
14:58:40 <virgil> CAN ANYONE MAKE USE OF LOADS OF ATLAS CREDITS!?
14:58:41 <karsten> okay, cool. lots of stuff going on there.
14:59:04 <virgil> IF SO EMAIL ME AND WILL HOOK YOU UP
14:59:07 <karsten> maybe explain that more on the mailing list. :)
14:59:10 <karsten> or that.
14:59:15 <Letty|Katja> virgil: ripe?
14:59:20 <virgil> Letty|Katja: yes
14:59:21 <karsten> thanks everyone for coming! next meeting in two weeks from now.
14:59:26 <clv> karsten: one last thing.. a quick note.. here is a wiki landing page i like, maybe you could get ideas from it:
14:59:31 <clv> https://trac.torproject.org/projects/tor/wiki/org/roadmaps/GetTor
14:59:42 <karsten> and sorry for rushing, but I think it's a feature to keep meetings at 60 minutes. (let me know if you disagree.)
14:59:49 <karsten> clv: cool, will take a look!
14:59:50 <Letty|Katja> virgil: can check how much i have
15:00:02 <karsten> #endmeeting