15:00:53 <asn_> #startmeeting SponsorR
15:00:53 <MeetBot> Meeting started Tue Jun 23 15:00:53 2015 UTC.  The chair is asn_. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:53 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:01:05 <asn_> hello
15:01:10 <asn_> who is here today?
15:01:17 <karsten> but rather keep an ec2 instance busy for a few hours and give us results.
15:01:19 <karsten> hello!
15:01:23 * syverson is here
15:01:45 <asn_> no sign of ohmygodel yet
15:01:48 <asn_> let's start i'd say
15:01:56 <dgoulet> karsten: ec2 instance to crunch old data?
15:02:11 <asn_> during past week, on SponsorR/HS matters
15:02:17 <asn_> i worked with DonnchaC on his SoP project
15:02:18 <karsten> dgoulet: that, or your cluster in the living room. ;)
15:02:27 <dgoulet> karsten: data that hs health comes up with makes sense over a long period of time, I can't go beyond 5 HS without failing more heavily on fetch :S
15:02:28 <asn_> i reviewed dgoulet's #4862  some more
15:02:41 <asn_> and I also wrote a first analysis of the stats from #15513 and #15744
15:02:48 <asn_> this can be found here: https://etherpad.mozilla.org/F8P18C7wYp-hs-stats
15:02:53 <karsten> dgoulet: let's discuss more later!
15:03:13 <amjohn> hello
15:03:18 <asn_> i also spend an hour or so trying to understand why we are seeing intro circuits on the wild with lifetime 25 hours +
15:03:27 <asn_> when intro circuits should last maximum 24 hours
15:03:41 <asn_> i still don't have the answer to this. although it would be nice to come up with an answer before publishing our results.
15:03:52 <asn_> in any case, that's it from me this week.
15:03:53 <asn_> next?
15:04:22 * dgoulet can go
15:04:25 <asn_> please
15:05:31 <dgoulet> first big thanks to asn_ on reviewing/testing #4862, it's still running strong on an HS I have, I worked on #16399 and #16389 for which the latter should help improve #16381
15:06:19 <dgoulet> I wrote the analysis for #13209, found in pad above and finally I re-read 224 in order to prepare to meeting in July :)
15:06:21 * dgoulet done
15:06:30 <asn_> thx
15:06:33 <asn_> next?
15:06:38 * karsten can go next
15:06:41 <asn_> karsten: go
15:06:58 <karsten> so, I was planning to evaluate the #1944 data, but then saw that rob sent very similar results from onionperf.
15:07:23 <karsten> but today I looked at rob's time-to-first-byte graph and saw how that differs from the torperf data
15:07:38 * robgjansen waves
15:07:43 <karsten> and figured that a closer comparison of the two tools might be worthwhile.
15:07:46 * karsten waves back
15:08:03 <asn_> aha
15:08:06 <asn_> in what sense comparison?
15:08:37 <karsten> in what exactly they measure, whether there are different setups (no cache clearing, etc.).
15:08:38 <asn_> the results were pretty similar right?
15:08:39 <asn_> aha
15:08:42 <robgjansen> i think the difference has to do with when we start the timer
15:08:55 <karsten> quite possible!
15:08:57 <robgjansen> we can discuss more now, or in discussion?
15:09:03 <karsten> in discussion, I'd say.
15:09:05 <karsten> but yes, we should.
15:09:08 <robgjansen> ok, great :)
15:09:08 <asn_> sounds good
15:09:11 <asn_> #topic onionperf torperf
15:09:13 <karsten> nothing more from me.
15:09:14 <asn_> thx
15:09:15 <asn_> next?
15:09:36 * syverson can go
15:09:41 <asn_> please
15:09:50 <syverson> Wrote and gave talk to 18F/US Digital Services about Tor in general and also ways in which I thought they could support/encourage use of Tor, and in particular use onion services to better protect users of US Government service sites.
15:09:51 <syverson> Met with Peter Eckersley, Seth Schoen, and Mike Perry to discuss one-ion services (aka Direct aka �) onion services that would play well with Let's Encrypt, require minimal changes for Tor, etc. I haven't written up our results yet, but will share them soon. (I'll write them on the plane from SF today if not before.)
15:09:51 <syverson> Done.
15:10:16 <asn_> interesting!
15:10:25 <dgoulet> yeah cool
15:10:25 <asn_> thanks
15:10:32 <asn_> next? robgjansen
15:10:45 * karsten likes the one-ion idea. (without having many further ideas on that, but it's a funny idea.)
15:11:09 <asn_> one-ion?
15:11:15 <asn_> first time i hear of this term
15:11:20 <syverson> Current services are two-nion
15:11:20 <asn_> i also like it maybe
15:11:23 <robgjansen> um, i'm not sure i have much for status for this meeting
15:11:34 <karsten> https://trac.torproject.org/projects/tor/wiki/org/sponsors/SponsorR/Terminology?action=diff&version=26
15:11:35 <asn_> robgjansen: ok no worries
15:11:36 <syverson> Require one onions (client) or two (client and server).
15:11:38 <robgjansen> i mostly want to sync on the upcoming HS meeting and with karsten on onioniperf
15:11:48 <robgjansen> relatedly, i've been working on kist
15:12:01 <robgjansen> and hoping to be able to show how it can imporve onion service latency
15:12:18 <robgjansen> hold for ticket number...
15:12:37 <armadev> seems like the guard would be a bottleneck that kist can't help too much with. looking forward to learning more. :)
15:12:52 <robgjansen> #12541
15:13:27 <asn_> thx robgjansen
15:13:28 <armadev> maybe kist still helps because tor gets to make decisions about what goes out next, in a way that's much closer to the metal
15:13:41 <robgjansen> perhaps the guard could be a problem, depending on the popularity of the service
15:13:48 <robgjansen> i think so, i think kist helps end-to-end
15:14:01 <robgjansen> load balancing at the guard->HS link is another problem
15:14:29 <robgjansen> or, load balancing in general (it may not happen at the guard->HS link...)
15:15:06 <asn_> ack
15:15:12 <robgjansen> where by load balancing here i am thinking in the more traditional server load balancing sense, and not in the bandwidth measurement sense
15:15:15 <asn_> armadev: you have anything to report?
15:15:36 <armadev> i might, but nothing comes to mind. i've been doing too many things. i guess i report: my apartment is now cleaner than it was on thursday.
15:15:55 <asn_> that's nice
15:15:56 <asn_> ok thx
15:15:57 <armadev> also i asked david to sit in on the google hangout this afternoon. i don't think i have the hardware/software for it. he doesn't either.
15:15:57 * ohmygodel can go next
15:16:01 <armadev> go for it
15:16:04 <asn_> ohmygodel: !
15:16:08 <asn_> ohmygodel: go
15:16:20 <dgoulet> armadev: (still on my list of things to try to fix this afternoon... :S)
15:16:21 <ohmygodel> we're now running two exit relays
15:16:24 <ohmygodel> on a pretty fast host
15:16:34 <ohmygodel> i'm running PrivEx on one of them
15:16:39 <asn_> !
15:16:40 <ohmygodel> contributing to Tariq's deployment
15:16:53 <armadev> neat. where are they deployed?
15:16:56 <robgjansen> i'm starting a third today on a separate host
15:16:59 <ohmygodel> i hope to add all of the HS statistics we had talked about wanting to such a deployment
15:17:08 <asn_> how is privex working for you?
15:17:19 <ohmygodel> fine no problem
15:17:48 <asn_> ohmygodel: ack
15:17:53 <ohmygodel> it only collects the thing they talk about in the paper: a count of visits (aka DNS resolutions) to a fixed list of censored sites
15:17:59 <robgjansen> providers: arvixe.com and pulseservers.com
15:18:03 <ohmygodel> i also met again with dov gordon
15:18:05 <asn_> ohmygodel: check out our branch in #15744 for how to collect some of those statistics
15:18:14 <ohmygodel> asn_: cool thx
15:18:31 <ohmygodel> dov gordon is starting as an assistant CS prof at george mason in the fall
15:18:42 <ohmygodel> his expertise is secure computation
15:18:49 <asn_> oh nice\
15:19:01 <ohmygodel> he's interested in working on applying recent protocols to statistics collection in Tor
15:19:23 <ohmygodel> there are a bunch of recent papers on the threat model most appropriate to tor
15:19:35 <ohmygodel> namely, dishonest majority, malicious adversary
15:19:47 <asn_> aha
15:20:08 <ohmygodel> so i'll be continuing to look into that in the coming months
15:20:14 <asn_> very interesting
15:20:19 <asn_> is dov gordon coming to pets?
15:20:28 <ohmygodel> i don't think so
15:20:37 <ohmygodel> his phd advisor (jonathan katz) is giving the PETS keynote though
15:20:44 <ohmygodel> on the topic of secure computation
15:21:04 <asn_> ah very interesting
15:21:20 <armadev> or, will he be in dc the week after? asn could meet him then too maybe.
15:21:25 <asn_> read some katz papers during my msc thesis. lots of papers on group key exchanges.
15:21:26 <robgjansen> also...
15:21:55 <ohmygodel> armadev: dov gordon lives in DC, and i bet a meeting could be arranged
15:22:09 <dgoulet> !
15:22:11 <armadev> i'll leave it to other people what their priorities are
15:22:18 <robgjansen> we have been hoping to get teor to help with some direct onion services development
15:22:25 <ohmygodel> in fact, dov and i are already meeting in DC at 5PM on July 13th, if that works for anybody else
15:22:55 <asn_> ohmygodel: i have a flight at 6PM off DC that same day.
15:23:00 <asn_> robgjansen: wow interesting!
15:23:02 <robgjansen> we are trying to figure out how to arrange a contract with him
15:23:20 <armadev> great.
15:23:25 <robgjansen> i will be working on this, and on this project in the coming months
15:23:44 <asn_> i wish we could have teor in PETS or in the HS meeting. or at least in the Tor meeting on september.
15:23:47 <ohmygodel> asn_: ok, perhaps the meeting could be moved
15:24:03 <teor> you rang?
15:24:04 <robgjansen> i hope we can get him to the dev meeting in berlin
15:24:11 <robgjansen> oh! :)
15:24:28 <asn_> ohmygodel: well, i know nothing about secure compuation so my presense is not that important
15:24:35 <asn_> ohmygodel: let's speak about secure compuattion during PETS though
15:24:35 <armadev> (hastur hastur teor)
15:24:46 <asn_> ohmygodel: maybe i will find something to inject myself into
15:25:09 <ohmygodel> ohmygodel: ack
15:25:10 <asn_> ok
15:25:11 <asn_> great
15:25:20 <asn_> i guess this sums up our briefing phase?
15:25:22 <asn_> let's move to discussion?
15:25:29 <asn_> what topics would you like to see addressed?
15:25:31 <robgjansen> teor: look forward to talking with you as planned
15:25:35 <syverson> Impending Memex madness.
15:25:44 <asn_> - we have the onionperf vs torperf thing
15:25:57 <robgjansen> ok, cool
15:26:02 <robgjansen> so as i said in previous emails
15:26:09 <asn_> - we have the upcoming HS meeting that maybe we should think of agenda or something.
15:26:37 <robgjansen> oh sorry did i jump the gun?
15:26:48 <asn_> robgjansen: maybe or maybe not.
15:26:55 <asn_> not sure if anyone wants to suggest other discussion topics.
15:27:02 <asn_> if not, we can go ahead with yours.
15:27:11 <robgjansen> asn_: ok tell me when to go
15:27:17 <dgoulet> I would love to clarify this also: 11:27 < syverson> Impending Memex madness.
15:27:18 <asn_> - i'd also like to discuss dgoulet's results from the HS health tool
15:27:31 <syverson> It seems like Phil has had a bunch of Memex meetings, but more focused specifically on
15:27:31 <syverson> the stuff most of the rest of the Memex program is doing rather than things that, e.g., we
15:27:31 <syverson> usually talk about in Sponsor R IRC meetings. Likewise there are some planning documents
15:27:31 <syverson> about the next quarter that basically don't seem to fit at all with most of what NRL or TPI
15:27:31 <syverson> do for Sponsor R. I hope people attending this hangout this PM (I'll be headed towards
15:27:32 <syverson> the airport at that time) can find out how/if we should be involved with this. Unless we get
15:27:34 <syverson> much better clarification (not expected) I'm going to ask Wade for guidance.
15:28:17 <asn_> oh
15:28:27 <armadev> syverson: can you point me at the planning documents?
15:28:40 <armadev> rob, aaron: i hear you'll be at the hangout? can you take notes and report-back?
15:28:42 <dgoulet> that damn google hangout...
15:29:00 <ohmygodel> armadev: i'll be at the hangout
15:29:01 <syverson> Basically, they don't exist yet. There's a template available on the memex wiki that I got.
15:29:08 <syverson> I'll mail them to you.
15:29:23 <robgjansen> i plan to let aaron take notes and avoid the meeting in favor of productivity
15:29:26 <ohmygodel> armadev: i will be taking notes, and i'll send them to memex@csl.sri.com
15:29:38 <armadev> great.
15:29:43 <asn_> ohmygodel: very useful. thanks.
15:29:53 <syverson> It is all about domain specific crawling indexing etc. with the domains mentioned in the recent email exchange w/ Phil.
15:30:23 <syverson> We're on the hidden persona domain. Nobody seems to know what that means.
15:30:28 <armadev> yeah. ideal for us is to keep doing useful things for the world while also having them think we're useful.
15:31:04 <syverson> I'm hoping Wade recognizes why Chris included us and still thinks it's a good idea.
15:31:31 <armadev> i guess i hope this too.
15:32:35 <asn_> ok
15:32:37 <syverson> But we are kind of a side thing, and when I see planning for the summercamp (their latest term) I worry that we
15:32:52 <syverson> 're a side that they're happy to set aside.
15:34:07 <syverson> Planning that is all about big integrated tools in support of domain specific searches for target consumers.
15:34:45 <ohmygodel> Well SRI is a big help here
15:34:53 <ohmygodel> they are doing crawling
15:35:04 <ohmygodel> and have developed some tools for specific domains
15:35:27 <ohmygodel> and have worked with hyperion grey on some pseudonym (aka "hidden persona") tracking
15:35:43 <syverson> Yes. That helps the overall LIGHTS contribution for sure. But I think that has been progressing as a straight SRI effort, not as a LIGHTS effort.
15:35:47 <armadev> yes. hoping that sri keeps them happy so we can keep writing tor is an..exciting gambit.
15:36:08 <armadev> a smart one if it works. a foolish one if it doesn't. :)
15:36:17 <dgoulet> lately, seems they need Tor for their PR also :P
15:36:54 <syverson> Is that what "hidden persona" is? That is much clearer than any of the long recent exchange I had w/ Phil.
15:37:13 <syverson> I never heard any reference to pseudonymous tracking.
15:37:33 <ohmygodel> syverson: yes, i believe that term was used in that sense in earlier memex meetings
15:38:37 <syverson> ohmygodel: Yes I now recall that, but it would have been nice if the domain leader who has been working with Hyperion Gray had responded with that.
15:38:59 <ohmygodel> ¯\_(ツ)_/¯
15:39:07 <armadev> i am, alas, expecting to get a sense of what the phrase means right about the time that the presentations about it have ended
15:39:46 <syverson> I wasn't planning to gambit per se. I was hoping to be explicit with Wade about what we are doing, how it has helped (e.g. in some of the public media about Memex) etc.
15:40:44 <syverson> We have done things *with* SRI and helped improve crawling. I don't want to downplay that to him, but I want to play up the other contributions.
15:40:50 * robgjansen goes afk momentarily
15:41:07 <armadev> plausible. maybe at pets we should chat a bit about how to position that. a fine hike topic.
15:41:23 <syverson> A good walk spoiled.
15:41:32 <syverson> ;>)
15:41:34 <asn_> :)
15:41:41 <asn_> so i guess we will learn more soon when the upcoming hangout happens
15:41:54 <asn_> syverson: should we talk more about this topic or move to next?
15:41:58 <syverson> You dream big.
15:42:01 <armadev> i encourage anybody with working hardware, software, and internet at the time to attend and tell us about it :)
15:42:19 <syverson> Next is fine I think. I just wanted to make sure all were aware.
15:42:50 <asn_> ok
15:43:01 <asn_> let's talk about onionperf and torperf next? karsten / robgjansen ?
15:43:13 <karsten> is robgjansen back?
15:43:14 * robgjansen is back
15:43:19 <karsten> ah cool
15:43:20 <robgjansen> hi, yes
15:43:34 <robgjansen> so, onionperf
15:43:53 <robgjansen> the main reason i did it was to have a baseline for onion service performance
15:44:26 <robgjansen> but karsten did something similar in torperf and it makes much more sense for him to keep running that since it is integrated with metrics@
15:44:38 <karsten> it's not yet, I just have a plan for that.
15:44:39 <robgjansen> (or can be made to be)
15:44:44 <robgjansen> right :)
15:45:07 <karsten> but you mentioned that we might be able to re-use parts of onionperf.
15:45:09 <robgjansen> the main advantage of onionperf is that it uses my shadow traffic generator
15:45:20 <karsten> yes, that one.
15:45:23 <robgjansen> a.k.a. tgen
15:45:26 <robgjansen> yes
15:45:27 <karsten> how does it work?
15:45:35 <asn_> oh
15:45:44 <robgjansen> you write a python script to define a behavior
15:45:55 <robgjansen> the python script spits out an xml file
15:45:59 <robgjansen> in graphml format
15:46:10 <robgjansen> you feed that as a parameter to tgen
15:46:34 <robgjansen> and tgen will create connections and transfer random data following that xml model
15:46:52 <robgjansen> the idea is you can model any behvaior with this
15:47:04 <armadev> how does the server side learn what the behavior is?
15:47:05 <karsten> and tgen is your tool or written by someone else?
15:47:10 <robgjansen> right now i am modeling torperf-style 50k, 1m, 5m single downloads
15:47:25 <robgjansen> i wrote tgen
15:47:29 <robgjansen> i use it in shadow
15:47:30 <karsten> ok
15:47:36 <karsten> also python?
15:47:37 <robgjansen> i maintain it in shadow and will continue to
15:47:48 <asn_> how do you define behavior for downloading a 5mb file?
15:47:56 <robgjansen> i made it so you can compile it outside of shadow and run it normally
15:48:01 <asn_> or do you define behavior on how much time to  wait between downloading  these files many times
15:48:04 <asn_> ?
15:48:09 <robgjansen> the client and server both run tgen
15:48:30 <robgjansen> so they both understand "tgen" protocol
15:49:18 <asn_> oh i se
15:49:19 <asn_> e
15:49:24 <robgjansen> an example of the python script used to create the behaviors is here
15:49:26 <robgjansen> https://github.com/shadow/shadow/wiki/3-Simulation-Customization#customizing-generator-behaviors
15:49:54 <robgjansen> tgen has several "actions"
15:50:17 <robgjansen> the main ones are 'start' 'pause' 'transfer' 'end'
15:50:28 <robgjansen> this is how you add a 5m file download
15:50:31 <robgjansen> G.add_node("transfer", type="get", protocol="tcp", size="5 MiB")
15:50:32 <karsten> how do we know that these profiles are realistic?
15:50:37 <robgjansen> remember this is just to bulid the xml file
15:51:05 <robgjansen> yes, so we will need to generate plausible models
15:51:06 <armadev> is it one client and one server that matches up to the client? or can it be one server that does the same tgen behavior in parallel to n clients?
15:51:17 <teor> nickm: ^^^ ideas for chutney 2
15:51:18 <robgjansen> the latter
15:51:27 <armadev> great
15:51:44 <robgjansen> that same github page has more documentation
15:51:45 <karsten> how crazy would it be to have the server run on port 80 and answer to the world?
15:51:51 <armadev> teor: yes indeed, something like tgen should be reusable in the chutney context
15:52:08 <robgjansen> not that crazy
15:52:11 <karsten> (just thinking if we could use it for torperf's default model of measuring non-HS performance)
15:52:24 <robgjansen> i think we should most definitely use it for torperf!
15:52:29 <Yawning> (shouldn't it need authentication)
15:52:35 <Yawning> (if it was world reachable)
15:52:36 <teor> armadev: I created a similar scheme in my draft patch to load-test using chutney. But it's a hack.
15:52:59 <robgjansen> true, it currently has no auth
15:53:18 <robgjansen> that because it now just runs either in shadow
15:53:30 <robgjansen> or on localhost in onionperf
15:53:39 <asn_> robgjansen: so torperf as it is measures "time to download a 5kb file over hidden services"
15:53:43 <asn_> what would torperf + tgen measure?
15:53:58 <robgjansen> (i.e., the onionperf client goes out to tor and comes back to the onionperf server running on the same machine)
15:53:59 <asn_> "time required for a client of type X to complete its jobs"?
15:54:22 <robgjansen> sure
15:54:33 <robgjansen> you can also add dependencies between transfers
15:54:41 <robgjansen> so say transfer1 goes first
15:54:56 <robgjansen> and then transfer[2-20] start after transfer one
15:55:15 <robgjansen> those could all be considered a "web page"
15:55:31 <robgjansen> right, get the top level html page, then get the embedded objects
15:55:38 <asn_> i see
15:55:47 <asn_> yes, that sounds more realistic than a plain 5kb file.
15:55:50 <robgjansen> so we could use real data from htmlarchive to generate plausible models
15:55:53 <karsten> we have wanted more realistic measurements in torperf for a while.
15:56:06 <robgjansen> sorry, httparchive
15:56:08 <robgjansen> http://httparchive.org/
15:56:41 <robgjansen> another cool thing is that reserchers or devs could just generate the xml
15:56:45 <robgjansen> using python
15:56:52 <robgjansen> and not ever need to worry about the code
15:57:11 <robgjansen> so you could get help generating realistic models
15:57:32 <asn_> ack
15:57:46 <asn_> btw, i think i have a Summer of Privacy meeting in 2 mins...
15:57:53 <asn_> so maybe we should start wrapping up?
15:57:53 <kernelcorn> I'm here
15:57:53 <robgjansen> so, if this sounds useful for torperf to move to this
15:58:08 <asn_> robgjansen: it does sound useful. but of course karsten knows best on how to do this.
15:58:10 <karsten> it does sound useful.
15:58:11 <Yawning> hi
15:58:15 <robgjansen> i am willing to visit karsten and help work on whatever needs to be done to integrate into torperf
15:58:19 <asn_> kernelcorn: give us a few mins please :)
15:58:27 <robgjansen> including adding auth if needed
15:58:49 <karsten> well, having the option to add auth would be great.
15:59:04 <karsten> I agree that it's not that important right now, with HS measurements being the first target.
15:59:04 <robgjansen> we can discuss missing features later
15:59:09 <asn_> ok
15:59:12 <asn_> maybe a trac ticket is in order?
15:59:20 <armadev> one of the challenges with longitudinal tests is that you really want to do a given test forever. so in the ideal world we would pick the best test first. in the real world we might find ourselves always adding to the set of tests we do, and never closing old ones down, because they still might be useful. so be it.
15:59:45 <robgjansen> yep, have to start somewhere :)
15:59:59 <robgjansen> so now, we both have datasets
16:00:00 <karsten> robgjansen: okay, let me read up the onionperf/tgen documentation.
16:00:03 <armadev> (for example, even though we might be able to make a much more realistic test, we probably will want to keep doing the 50/1m/5m ones)
16:00:13 <robgjansen> agreed
16:00:43 <karsten> robgjansen: and by that I mean: let me *try
16:00:44 <robgjansen> should we plan what to do with out datasets short term for the memex meeting
16:01:07 <karsten> to read up the onionperf/tget documentation as time permits, as the month end is getting closer and stuff is piling up here.
16:01:16 <asn_> we can use both graphs for the slides or whatever? and say we tried with two methodologies?
16:01:17 <robgjansen> karsten: ack, should we plan my visit off-irc?
16:01:26 <karsten> robgjansen: sure.
16:01:37 <karsten> robgjansen: so, why are our graphs different?
16:01:44 <karsten> (if we still have time)
16:01:55 <karsten> (if not, -> email)
16:02:02 <asn_> let me suggest email on this one? :)
16:02:07 <asn_> or trac ticket?
16:02:08 <karsten> ok :)
16:02:21 <robgjansen> this is a short term item for memex meeting ?
16:02:34 <robgjansen> how to combine or show our datasets?
16:02:34 <asn_> figuring why the graphs are different?
16:02:35 <asn_> maybe not.
16:02:45 <asn_> i think even one set of graphs is sufficient for the upcoming meeting?
16:02:50 <robgjansen> i assume we want to show this data during our 1 hour presentation
16:02:51 <asn_> or maybe both sets of graphs is OK?
16:03:08 <robgjansen> right, we need to decide this
16:03:11 <asn_> "we made this graph with this methodology." "and we also have this new technique of simulating traffic, that produced this other graph"
16:03:30 <robgjansen> i have over a month worth of data at this point
16:03:39 <robgjansen> i started it during the last april meeting
16:03:51 <robgjansen> or shortly after fixing some things
16:03:54 <karsten> same here, I think. not sure exactly how many weeks, but a few.
16:04:11 <armadev> robgjansen: ..we have a 1 hour presentation? when?
16:04:21 <syverson> Next week, most of us will be in a PETS talk on anonymous communication at this time Tuesday. Can we just agree that we'll figure out a time for the next Sponsor R meeting (which will be crucial given the PI meeting starts the following week) some time there when we're all (most) face-to-face?
16:04:35 <syverson> armadev: ?
16:04:36 <asn_> syverson: sounds reasoanble.
16:04:38 <robgjansen> that works for me
16:04:47 <robgjansen> am i mistaken about the hour?
16:05:04 <syverson> You mean at the PI meeting? I don't think so.
16:05:14 <robgjansen> yes i mean at the PI meeting
16:05:15 <syverson> I don't think we have a presentation at all at this point.
16:05:25 <robgjansen> afaik, SRI+Tor+NRL has 1 hour together
16:05:36 <robgjansen> oh
16:05:42 <syverson> No. Hidden personas have an hour.
16:05:42 <robgjansen> ok, well i have to defer to paul
16:05:56 <armadev> i alas haven't heard of such a thing either.
16:06:10 <armadev> it would be useful to have it, but i guess we'd find the other 14 hours of presentations boring, so.
16:06:11 <syverson> I say this all with a confidence in my lack of understanding WTF is supposed to happen there.
16:06:42 <armadev> it might be that we have a surprise 45 minutes for pitching everything to wade. might be nice to have some graphs for that.
16:07:01 <asn_> yeah, there will be graphs.
16:07:10 <syverson> I'm thinking at this point our time will be best in one-on-ones w/ Wade and working amongst ourselves. But that's just a guess.
16:07:40 <robgjansen> such a great feeling to know what is expected of us. wish i had it here.
16:07:49 <syverson> Being ready for an elevator speech with slides is a good idea.
16:07:52 <Yawning> ...
16:08:03 <asn_> Yawning: give me a few minutes plz
16:08:13 <asn_> ok
16:08:24 <asn_> so that's that on this.
16:08:27 <asn_> one last thing i wanted to ask
16:08:39 <asn_> do you think it would be worthwhile discussing agenda for the upcoming HS meeting in DC?
16:08:44 <asn_> like voting on topics we want to see discussed etc?
16:08:51 <asn_> or should we just do this process the first 2 hours of the meeting?
16:08:58 <syverson> OK let's figure out when to meet at PETS and have it also on IRC for those not in attendence (to be announced when).
16:09:07 <armadev> asn_: can't hurt to vote, to get people thinking early?
16:09:14 <asn_> armadev: ack
16:09:28 <armadev> asn_: also, did people sort out a plan for whether to have the meeting physically nearby memex, or not? there are tradeoffs each way.
16:09:30 <asn_> syverson: should we figure out meeting time beforehand? or just do it ad-hoc during PETS?
16:09:40 <asn_> armadev: i saw you saying that
16:09:50 <asn_> armadev: i'm fine with having the meeting nearby memex.
16:10:06 <asn_> maybe the people who go there the first days, can see what options there are for this?
16:10:12 <syverson> asn: I was thinking we need to wrap up now. So don't have time to figure that out.
16:10:16 * karsten won't be at PETS and would be fine reading a summary of that meeting after it has happened.
16:10:20 <asn_> karsten: ack
16:10:21 <asn_> syverson: ok
16:10:35 <armadev> asn: yes. we might even talk to phil about getting one of the little office rooms reserved for us for those days.
16:10:37 <asn_> syverson: we will be like 4-5 people, so I think figuring out a common time might not be super hard.
16:10:52 <asn_> ok
16:10:57 <ohmygodel> asn_: there should be more than 4-5
16:10:58 <asn_> all these things make sense for now.
16:11:05 <ohmygodel> more like 7-10
16:11:10 <asn_> during PETS?
16:11:13 <syverson> I think the copresent can sync. And we'll just announce for the others.
16:11:18 * teor is still trying to get head around PETS
16:11:20 <asn_> me, you, armadev, syverson, robjohnson, nick mathewson?
16:11:22 <ohmygodel> i meant at the HS meeting
16:11:27 <asn_> ohmygodel: ah yes for sure.
16:11:40 <asn_> ohmygodel: i meant figuring out a common time for the SponsorR meeting during PETS.
16:11:44 <asn_> teor: what do you mean?
16:12:19 <robgjansen> wait, do you mean robjohnson?
16:12:27 <asn_> hno i meant robgjansen
16:12:34 <teor> asn_: just catching up with all the meetings
16:12:34 <asn_> i just dont know how to spell
16:12:37 <robgjansen> oh, cause he could be there too... :)
16:12:42 <asn_> robgjansen: :P
16:12:51 <syverson> I'm guessing we'll do Monday or Tuesday evening some time.
16:12:52 <teor> asn_: in my head
16:12:56 <asn_> ack
16:12:57 <asn_> #endmeeting