problem@conference.ietf.jabber.com - 2003/03/21
[09:47] %% avri has arrived.
[10:11] %% mlshore has arrived.
[10:16] %% randy has arrived.
[10:17] %% azinin has arrived.
[10:17] <azinin> r we going to have a jabber scribe?
[10:19] %% lisaDusseault has arrived.
[10:19] %% jhutz has arrived.
[10:19] <lisaDusseault> I'm here to scribe
[10:19] <azinin> great!
[10:19] <lisaDusseault> Melinda is talking about the agenda
[10:20] %% dcrocker has arrived.
[10:20] %% agall has arrived.
[10:20] <lisaDusseault> Original agenda here: http://www.ietf.org/ietf/03mar/problem.txt
[10:21] <lisaDusseault> There were a couple additions, e.g. Eric Rescorla talking about IESG Review
[10:21] <lisaDusseault> First topic: Charter (http://www.ietf.org/html.charters/problem-charter.html)
[10:21] %% jis (at sfo) has arrived.
[10:21] <lisaDusseault> <Mda> The work of this wg is to be descriptive
[10:21] %% Olaf has arrived.
[10:22] <jis (at sfo)> t-mobile hotspot at SFO (United) works
[10:22] %% andy has arrived.
[10:22] <lisaDusseault> <Mda> Our two deliverables are the problem statement document and the process plan
[10:22] %% hta has arrived.
[10:23] <lisaDusseault> <Avri> Administrative stuff. Mailing list = problem-statement@alvestrand.no
[10:24] %% avshalom has arrived.
[10:24] %% duerst has arrived.
[10:25] %% avri has left.
[10:25] %% falk has arrived.
[10:25] <lisaDusseault> Somebody at mike: asks to add to charter a list of existing effective rules to preserve.
[10:27] <lisaDusseault> Next topic: Henning Schulzrinne, IETF "Goodput" measurements: packet size, delay and throughput
[10:27] <lisaDusseault> <Henning> Intend to measure delay from -00 to RFC
[10:29] <lisaDusseault> <Hng> It's not an easy thing to measure given extremely poor data requring title matching.
[10:29] <jhutz> I believe "somebody at mike" as Chris Allen
[10:29] %% mrose has arrived.
[10:30] <lisaDusseault> <Hng> We have lost our announcement history previous to the last five yrs.
[10:30] %% mallman has arrived.
[10:30] %% resnick has arrived.
[10:31] <lisaDusseault> <Aaron at mike> The announcement list does contain the file name in it.
[10:31] <lisaDusseault> <Hng> That's not a public archive.
[10:32] <lisaDusseault> Henning shows graph of rfcs/yr
[10:32] %% mrose has left.
[10:33] <lisaDusseault> <Hng> Average number of pages per RFC was expected to shoot up with all the req'd sections
[10:33] <lisaDusseault> <Hng> Roughly 20-30 pgs/yr
[10:33] <lisaDusseault> <Hng> After 1980 rfc lenght hasn't changed much.
[10:34] %% leslie has arrived.
[10:34] <lisaDusseault> Henning shows RFC avg delay per year
[10:34] %% mallman has left.
[10:34] <lisaDusseault> <Hng> The delay is not special in the last year
[10:36] <lisaDusseault> <Harald> Did you try doing this graph by the year started as opposed to the year ended
[10:36] <lisaDusseault> <Hng> no
[10:36] %% Bill Fenner has arrived.
[10:37] <lisaDusseault> <Hng> All the statistics seem to have the characteristic that around 2000 rapid growth ceases.
[10:37] <falk> surprise, surprise :/
[10:37] <lisaDusseault> <?> Where did you get the 00 draft publication date? <Hng> RFC announcement list with mid-month assumption.
[10:37] %% ole has arrived.
[10:38] <Bill Fenner> <?> was Pekka Savola
[10:38] <lisaDusseault> <Hng> This is id's per year, not only -00 ids but any number.
[10:38] <lisaDusseault> <Hng> I didn't do statistic on just -00 drafts.
[10:39] <lisaDusseault> Hng shows histogram of delays -- how many rfcs appear in each 3-month delay length bucket
[10:39] %% jis (at sfo) has left.
[10:39] <lisaDusseault> <Hng>Clustering up to two years followed by decay and signficant tail.
[10:40] <lisaDusseault> <falk at mike> You've made a dangerous mistake by showing a roomful of engineers some data.
[10:40] <lisaDusseault> <falk> Does it occur with some freqency that docs published as individual submissions become wg docs
[10:40] <lisaDusseault> <Hng> slides upcoming, but to that point, this data is title matching
[10:41] <lisaDusseault> <Hng> Hopeless to try to distinguish.
[10:41] <lisaDusseault> <Fred B> A comment... What's more interesting is the frequency of updates and the number of rfc's obsoleted.
[10:42] <lisaDusseault> <Fred> we seem to now stay at ID stage for a while.
[10:42] <lisaDusseault> <Hng> It's amazing the # of ways you can format ID announcements.
[10:43] %% mrose has arrived.
[10:43] <lisaDusseault> <Hng> Some IDs last very long for being blocked on somebody else's ID
[10:43] %% sleinen has arrived.
[10:43] <lisaDusseault> <Hng> If we care that this matters a sampling aprpoach allows us to get biography of sample and much better impression whether thiere is 1 id iter'n per IETF mtg.
[10:45] <lisaDusseault> <Hng> I'm not sure this data helps but my conclusions are that statistics have not changed dramatically in the last few years.
[10:45] <lisaDusseault> <Hng> We had a problem but it took us a while to be sure we had it.
[10:46] <lisaDusseault> New Topic: Rescorla, analysis of IESG review process
[10:47] <lisaDusseault> <EKR> Apparently there's something in the water that makes us want to use statistical analyses.
[10:47] <lisaDusseault> <ekr> My data more focused on IESG review to address complaints that we have no visibility what is going on in there.
[10:48] %% ole has left.
[10:48] <lisaDusseault> <ekr> Henning's data confirmed that I have the right data.
[10:48] <lisaDusseault> <ekr>People ask why my draft isn't published.
[10:48] <lisaDusseault> <ekr> It takes quite a while. Mean is 147 days from last call to approval.
[10:49] <lisaDusseault> <ekr> It's really variant - 130 days standard variation.
[10:49] <lisaDusseault> <ekr> Not unusual for drafts to take 300 days from last call to approval.
[10:49] <lisaDusseault> <ekr> One might ask whether docs approved as is go through faster. Unchanged documents still have a mean of 78 days.
[10:50] <lisaDusseault> <ekr> The areas vary quite a bit in terms of how long it takes to approve. Fast areas are internet, ops and transport.
[10:50] <lisaDusseault> Slow areas are apps and security.
[10:51] <lisaDusseault> <ekr> Huge difference in output volume between areas.
[10:52] <lisaDusseault> <ekr> Length of document *doesn't* matter in time to approval.
[10:52] <lisaDusseault> <ekr> Volume of changes doesn't matter either.
[10:52] <jhutz> hm... neither change volume nor document length has an effect. that really is surprising
[10:53] <lisaDusseault> <Crocker> You have firm conclusions. I need to comment on methodology. There's a problem in variance when the data is not standard distrib'n.
[10:53] <lisaDusseault> <ekr> It's log normal.
[10:53] <mrose> dave is referring to http://www.ietf.org/internet-drafts/draft-etal-ietf-analysis-02.txt
[10:53] <jhutz> I wonder if that means the issue is delay in getting to reviewing documents, rather than how long it takes to actually evaluate them
[10:53] <mrose> and also http://xml.resource.org/ietf-analysis/current/
[10:53] <lisaDusseault> <crocker> Use of std deviation and variation in non-normal distrib'n may be mathematically valid but not useful.
[10:54] <lisaDusseault> <ekr> I said this, and it's log normal. Survival analysis techniques do not need standard distribution.
[10:54] <azinin> jhutz: the turn-around delay is much bigger than the processing delay
[10:54] <lisaDusseault> <crocker> Use of those statistics in social environment is risky because community use legitimately normal data and don't know how to process non-normal data.
[10:55] <jhutz> wesommer++ for ending the argument between crocker and ekr about statistics theory
[10:55] <lisaDusseault> <Bill Sommerfeld> I'm a wg chair... One of the frustrating bits is that it seems you submit docs, you get 1 bug report, you fix it, submit, get another bug report
[10:55] <lisaDusseault> <Bill> Do you have info'n about what the corrections were?
[10:56] <lisaDusseault> <ekr> The # of revisions also has log normal shape and the bulk of the curve lies between 0 and 4.
[10:56] %% avri has arrived.
[10:56] <lisaDusseault> <falk> There is AD review between last call and full last call and this is hard to account for .
[10:57] <lisaDusseault> <ekr> Yes, some ADs do much review to clear problems up front and some take right to IESG.
[10:57] %% mark.ellison has arrived.
[10:59] <lisaDusseault> New Topic: James Kempf, Helping the internet to age gracefully
[10:59] %% avshalom has left.
[10:59] <lisaDusseault> <kempf> I don't have as much data as Eric and Henning -t his is observ'ns on what problems are.
[10:59] %% avshalom has arrived.
[10:59] <falk> perhaps the iesg should create a liaison with AARP
[11:01] <lisaDusseault> <kempf> Internet is getting older. We don't do much architecture but we do some philosophy e.g. the end to end principal.
[11:02] <lisaDusseault> <kempf> Interdependencies make it difficult to get a consistent design.
[11:02] <avshalom>
[11:02] %% resnick has left.
[11:02] %% randy has left.
[11:02] <lisaDusseault> <kempf> These sometimes extend across areas.
[11:03] <lisaDusseault> <kempf> We try to minimize dependencies by taking a slice but sometimes it's not possible or desirable.
[11:03] %% ole has arrived.
[11:03] %% ole has left.
[11:04] %% ole has arrived.
[11:04] %% ole has left.
[11:04] <lisaDusseault> <kempf> Mabye system design is a better word than architecture.
[11:04] <lisaDusseault> <kempf> We've always bee successful with the stovepipe approach.
[11:05] %% ole has arrived.
[11:05] %% ole has left.
[11:05] <lisaDusseault> <kempf> Sometimes problems don't turn up until deployment or use.
[11:06] <lisaDusseault> <kempf> Ignoring our design problem is at least resulting in lack of transparency and understandability of internet design.
[11:06] <lisaDusseault> <bob hinden> It seems that's what proposed std means - deployment not known. This has gotten changed but that's my understainding.
[11:07] <lisaDusseault> <bob> There's very few things that make it ot DS.
[11:08] <lisaDusseault> <bill> I underline what bob said that it is so difficult to get to DS people burn out.
[11:09] <lisaDusseault> <hng> It's easier to understand your points if you pont to concrete examples of these problems.
[11:09] %% avshalom has left.
[11:09] <lisaDusseault> <hng> It sounds plausible but how big is it ... is it "i don't like radius" or do people disagree
[11:11] <dcrocker> I need to make clear that calling my concern "statistics theory" exactly highlights the problem. There is a basic difference between the mathematics of statistics and the procedures of methodology. Methodology is a separate set of issues that USE statistics.
[11:11] <hta> demonstrating that all presentations should be on one PC when you have a crowded agenda.
[11:11] <dcrocker> So the statistics are fine, but the methodology is flawed. This calls to question any conclusions that are asserted.
[11:11] <lisaDusseault> Next topic: Ted Hardie, "working group particip'n"
[11:12] <dcrocker> And, by the way, I did not bother to raise the fact that Eric actually did analysis of multiple variables, without using multivariate algorithms. This is simply not kosher.
[11:12] <lisaDusseault> <ted> I've written a draft called WGs and their stuckees.. what is the problem
[11:13] <lisaDusseault> <ted> WGs are historically defined by a DL where consensus is really measured on DL.
[11:13] <hta> dave: you also need to listen to the comment about "you talk to different social scientists than I do" - it's relatively clear that you two have information enough to discuss intelligently what your preconceptions are, but that the WG doesn't have enough information to evaluate your positions-at-present.
[11:14] <lisaDusseault> <ted> THe process is very open - anybody can make a comment at any time.
[11:14] %% smb has arrived.
[11:14] <lisaDusseault> <ted> However making a comment does not mean you're taking on commitment or responsibility.
[11:14] %% avshalom has arrived.
[11:14] <lisaDusseault> <ted> That makes it hard to estimate how much attention something will receive.
[11:15] <lisaDusseault> <ted> It's hard to make anybody but WG chairs and doc editors accountable.
[11:16] <lisaDusseault> <ted> How do we retain openness but increase commitment?
[11:16] <lisaDusseault> <ted> You get to figure out how.
[11:16] <lisaDusseault> <crocker> Let's see if I have this right.. we have wg members consuming bandwidth who don't help make progress.
[11:16] <lisaDusseault> <ted> that may be a problem but it's not the problem I'm trying to get at.
[11:17] %% resnick has arrived.
[11:18] <lisaDusseault> <crocker> The wg isn't real if people don't show up and do work.
[11:20] %% azinin has left.
[11:20] <lisaDusseault> <ted> We have very few ways to put people on the hook. A small number of people get very overloaded.
[11:21] <lisaDusseault> <crocker> I belieev what you said is reasonable but it represents a different philosophical or strategic difference from what the IETF ought to be.
[11:22] <lisaDusseault> <bob> I don't see this as a big problem.. if there aren't people to do the work, then forget it.
[11:22] <lisaDusseault> <bob> Much bigger problems are in getting consensus, getting quality docs.
[11:23] %% randy has arrived.
[11:23] <mrose> just a little something that keith moore sent to the mailing list (keith doesn't appear to be jabbering right now): I just realized that the problem-statement WG is currently providing a
very good example of one of our biggest problems with the way we do
work:
we have precious little face-to-face meeting time. despite this, we
spend the vast majority of our meetings in presentations of material
that could (in most cases) easily be published as internet-drafts and
read by participants at other times.
the one thing we can do in meetings that we can't do online is discuss
things face-to-face, and take advantage of the increased fidelity and
bandwidth of communication in meatspace. this is often incredibly
useful for reducing dissent and promoting closure. but when we try to
do this in meetings, we are told that the agenda is full with speakers
and that we are already behind schedule.
Keith
[11:24] <mrose> aaron responded via email, you may want to do so here too
[11:24] <lisaDusseault> <?> I know there's a difference between a wg wanting work to get done and a wg wanting to do work.
[11:24] %% azinin has arrived.
[11:24] <falk> I agree with you frusteration but I also think that a short
presentation is useful for focusing the group on the topic at hand to
get a more useful discussion. If you want to have disussion on a
bunch of topics...
[11:25] <resnick> Reply by me: It's only useful if people haven't read the documents beforehand. I have. Seems like a waste of time to me.
[11:25] <lisaDusseault> New Topic: Margaret Wasserman, Process Document
[11:25] <resnick> Not that this is something I haven't said a million times before.
[11:26] <lisaDusseault> <mw> The doc is in progress..
[11:27] <lisaDusseault> <mw> When the industry is doing we we tend to rest on laurels. Bad times we tend to be critical and focus on failures.
[11:27] <mlshore> I agree with the comments, but we chose to take a different approach this time because we know that a lot of people haven't read the documents, in some cases there are no documents, and as a brand-new working group we need to do a level-set (there was no BOF). We'll be more conscious of this issue in future meetings. Please raise stuff on the mailing list in the interim
[11:27] <lisaDusseault> <mw> We could accidentally change some good things. We could be overreacting now. To avoid this, build a culture with consistent focus on improvement rather than wait for emergency.
[11:28] <lisaDusseault> <mw> I listed the core values from the plenary. These need to remain consistent.
[11:29] <lisaDusseault> <mw> Also there are some things that aren't core values: e.g. division of work into certain areas.
[11:29] <lisaDusseault> <mw> Current tension includes "us vs them" and lack of trust
[11:29] <lisaDusseault> <mw> But we hvae really high commitment to improve.
[11:31] <lisaDusseault> <mw> What we haven't figured out his whether to evolve or revolt.. make granular change or ...
[11:31] <lisaDusseault> <mw> Let's get some feel for this.
[11:31] %% avshalom has left.
[11:31] <lisaDusseault> <mw> Evolution is iterative improvement.
[11:32] %% avshalom has arrived.
[11:33] <lisaDusseault> <mw> Granular means we could address separate problems separately, rather than monolithic.
[11:33] <lisaDusseault> <mw> We can do these deliberately step by step or treat this as an emergency.
[11:34] <lisaDusseault> Immediate change is actually good to address frustration or perception problems.
[11:34] <lisaDusseault> <mw> It aslso avoids analysis paralysis.
[11:36] <lisaDusseault> <mw> A coordinated approach involves establishing teams
[11:36] <lisaDusseault> <mw> We could just ask the leadership to correct problems.
[11:36] <lisaDusseault> <mw> At the other extreme problems could be assigned to indivduals. In the middel, various team orgs.
[11:38] <lisaDusseault> <kempf> People seem to assume there will be one sol'n for entire IETF. Another way is to try multiple experimentally.
[11:39] <lisaDusseault> <mw> Figure out what you're going to measure -- yes -- this is captured as one possibility
[11:39] <lisaDusseault> <hta at mike> I am going to make a paradoxical statement: fixing our short-term problems is harmful to the internet.
[11:40] <lisaDusseault> <hta> Where we stand, fundamental changes to IETF are needed. If we fix only short term problems to live with current structure, for five more years, we're dead.
[11:40] %% hildjj has arrived.
[11:41] <lisaDusseault> <mw> Do you think there's community consensus that this is all broken?
[11:41] <lisaDusseault> <crocker> The suggestion about experimental -- what will be the experimenatl group?
[11:41] <lisaDusseault> <crocker> Also we've been doing an experiment for 15 yrs.
[11:41] %% leslie has left.
[11:41] %% avshalom has left.
[11:42] <lisaDusseault> <crocker> If we make fundamental changes, the only thing we know is we have no idea what the outcome will be.
[11:43] <lisaDusseault> <bob> I think harald is right - fine tuning will not work, we need to make fundametnal changes.
[11:43] <lisaDusseault> <hta> On reaching consensus: If I have to declare and the IETF doesn't agree with me I'll go home. This is an open process and I want the community to find the conclusions.
[11:45] <lisaDusseault> New topic: Elwyn, problem statement
[11:46] %% avri has left.
[11:46] <lisaDusseault> <elw> The problems are not new. The aim is improvment, not fignre-pointing.
[11:46] <lisaDusseault> <elw> I didn't attempt to link problems to root causes.
[11:48] <lisaDusseault> <elw> Read the draft.. I will not spend time reading everything
[11:48] <lisaDusseault> <elw> First, we do not appear to have a common understanding of our mission.
[11:48] <mrose> and people are already starting to line-up at the mic...
[11:50] <lisaDusseault> <falk> You should be commended for taking this role. However I felt that I basicalloy did not agree with the rpoblem statements in the draft.
[11:50] <lisaDusseault> <elw> It's difficult to do that, let's try.
[11:51] %% hardie has arrived.
[11:51] <lisaDusseault> <keith moore> Please let's discuss this rather than have it presented.
[11:51] <lisaDusseault> <avri> The presentations are to be quick and then discuss. Did people read the draft?
[11:51] <lisaDusseault> Fairly large show of hands
[11:52] <hardie> Dave thanks Elwyn, but agrees that it comes across as too much the facts
[11:52] <hardie> Believes that the rank ordering and priorities are so valuable worth any flaw
[11:53] <lisaDusseault> <crocker> The draft is actually really useful to list hese althgouh I also don't necessarily agree with them.
[11:53] <lisaDusseault> Ahh, ted do you want to take over? My wrists would appreciate it :)
[11:53] <hardie> I just came in; hadn't really someone was already doing it.
[11:53] <hardie> Happy to take over it if you need it.
[11:53] <lisaDusseault> Thx
[11:54] <hardie> Chairs say that the presentation is only to structure discussion
[11:54] <lisaDusseault> I'll at least take a break; let me know if you want me to take over later.
[11:54] <hardie> Speaker notes that perceptions are not facts, but that the perceptions existence is a fact, and if it is wrong, it is a problem.
[11:55] <Bill Fenner> we learned in the IESG plenary... perception *is* reality
[11:55] <hardie> Elwyn continues through the draft, focused on "common understanding of its Mission slide"
[11:55] <hardie> Harald believes we have a common understanding, which we don't know how to express, so we can't tell we agree
[11:56] <hardie> Christopher Allen believes there is a cultural element here that is important to capture. He misses a line between mission and visioin. Hard to connect them to encourage contribution
[11:56] %% andy has left.
[11:56] <hardie> Next slide:IETF does not use Effective Engineering Practices.
[11:57] <hardie> He believes the draft has stimulated discussion of new practices (drafts) etc.
[11:57] %% kmoore has arrived.
[11:57] %% avri has arrived.
[11:57] <hardie> Bob Hinden comments that this mixes two things. "How IETF works" and "what's inside drafts and whether the protocols work"
[11:57] <hardie> Break it out?
[11:58] <hardie> Elwyn agrees that project management and quality management are different
[11:58] <hardie> Hennning says that it goes back to lack of effective contracting with the document authors and editors. Expectations are not set.
[11:59] <hardie> Elwyn replies that this is a very difficult job, and this also tends to mean things are not delegated (since they might not get done)
[12:00] <hardie> Margaret Wasserman made this wording, but now wants to adjust it.
[12:00] <hardie> This is a problem with this as a problem--we need to identify what problem effective engineering processes would solve (we're not happy with quantity? we're not happy with quality?)
[12:00] <jhutz> name of current speaker?
[12:01] <hardie> Eric sees a problem not with quality, but with effort, as contention in working group discourages folks. Inclusiveness would help; fighting over who drives does not.
[12:02] <hardie> Elwyn says this harkens back to the possibility of hijacking.
[12:03] <hardie> Keith says this slide has good points, but doesn't hit the fundamental issue. We need to be better at describing the problems, so people can do better at determining whether something solves a problem.
[12:03] <hardie> next slide: IETF contributors appeart to be less engaged than in earlier days.
[12:03] <hardie> Elwyn asks why?
[12:04] %% sleinen has left.
[12:04] <hardie> Keith suggests that we need to express this in terms that do not harken back to "good old days"
[12:05] <hardie> Christian Huitema notes that is variable.
[12:05] <hardie> Randy notes that absolute number committed participants is up, just more tourists
[12:05] <hta> nathan lutchansky (I'm sitting where I can see the badges - never saw him before)
[12:05] <hardie> Nathan notes that we have a lot more "audience members"
[12:06] <hardie> Steve Trowbridge worries that some folks have been frightened away from the process or have so much trouble getting traction that they don't see it worth it.
[12:06] <hardie> Eric notes that rudeness is so bad that participation suffers.
[12:07] <hardie> Harald notes that working group chairs have the right to remove folks from mailing lists; the bar is high, but use.
[12:07] <hardie> James notes that it is not used; he knows of a case where physical violence was threatened but nothing happened.
[12:07] <hardie> Steve Trowbridge notes that stepping needs to happen much ealier
[12:08] * mrose has changed the subject to: logfile at http://www.jabber.com/chatbot/logs/conference.ietf.com/problem/2003-03-21.html
[12:08] <hardie> Jonas adds that it is not abuse always, but intimidation, because they are told they aren't familiar with 20 years of history and are idiots as a result
[12:08] * mrose has changed the subject to: logfile at http://www.jabber.com/chatbot/logs/conference.ietf.jabber.com/problem/2003-03-21.html
[12:09] <Bill Fenner> Jonne
[12:09] <hardie> Spenser thinks things are actually better, because we don't have as many new folks. but nothing stops a return to vituperation, but we need something that does.
[12:10] <hardie> Keith notes document doesn't directly address how we conduct meetings or run mailing lists; it should do.
[12:10] <hardie> Elwyn agrees, and notes it hurts our reputation.
[12:10] <hardie> new slide: authority and influence in the ietf are concentrated in too few hands.
[12:10] <hardie> Maybe Lisa should take over again here....
[12:11] <hardie> :)
[12:11] <lisaDusseault> OK
[12:11] <hardie> Thx.
[12:11] <lisaDusseault> <keith moore> I don't believe there's any grounds for this statement.
[12:11] <Bill Fenner> since you are an untrusted insufficiently accountable authority figure?
[12:11] <hardie> Yes.
[12:11] <lisaDusseault> <km> There is a perception of this and perception differences are part of our problem.
[12:11] <hardie> I am the enemy, and I need (B)
[12:11] <hardie> (Later, naturally, it's a bit early)
[12:12] <falk> try (C)
[12:12] <lisaDusseault> <km> Observation might get agreement.
[12:12] <lisaDusseault> <bob> I'm pretty comfortable with this slide/issue. Accountability is a problem although its getting better with doc tracker.
[12:13] <lisaDusseault> <steve trowbridge> There are design teams that don't have transparent processes and have backroom deals.
[12:14] %% randy has left.
[12:14] %% randy has arrived.
[12:14] <lisaDusseault> <henning> The quantative problem: the people that matter at the upper levels have not increased proportionally to the overall popul'n
[12:15] <lisaDusseault> <hng> However I have noticed a shift from wg level to AD level.
[12:15] <lisaDusseault> <hng> AD used to be present in background but not always mentioned. Now any discussion needs to have AD approval, or at least, that's the perception.
[12:15] <hta> comment: on this slide, perception is SHARPLY divided between those who have not been IESG/IAB and those who haven't been.
[12:16] <lisaDusseault> <mw> This issue mixes together a lot of things. It's not like 'those evil rules'.. we have a small number of them.
[12:17] <lisaDusseault> <?> This is a fact, no finger pointing: people do their work in good faith, but there are changes need to be made.
[12:17] <mlshore> Jonne Soinenen
[12:17] <lisaDusseault> <Jonne> To me, there is not much reason to discuss practice and perception, both are equally important.
[12:17] <lisaDusseault> <elw> Perhaps we need a bigger disclaimer that this is view of perception
[12:18] <lisaDusseault> <km> The solution to perception is different than the solution to root cause (?)
[12:18] <lisaDusseault> oops
[12:18] <lisaDusseault> <km> The bullet point about the IESG process allowing an IESG member to block a doc is just wrong.
[12:18] <lisaDusseault> <elw> Evidence was produced.
[12:19] <lisaDusseault> <kw> I spent 4 yrs on IESG. Even when a doc is really bad it's hard to block it.
[12:19] <lisaDusseault> <kw> I sweated that stuff every day.
[12:19] <lisaDusseault> <kw> Basically that's not an accurate statement.
[12:20] <lisaDusseault> <mrose> I knew Keith was going to argue that last bullet point.
[12:20] <lisaDusseault> <mrose> Some former ADs do think this is a serious problem. So contrary to what Keith said, we have existence proof this can happen.
[12:20] %% duerst has left.
[12:20] <lisaDusseault> <keith droge> The item about a 'ruling class' is a problem.
[12:21] <lisaDusseault> <kd> What people want to see is that when the ruling class makes a decision it's recorded
[12:21] <lisaDusseault> <Eric> If openness is our key value, it's interesting the IETF gets to decide what wgs can form, what chairs get selected etc.
[12:21] <falk> quality is our key value, imo
[12:22] <lisaDusseault> <eve> I've taken a reprieve... come back after a few years... I don't think the IETF is that much more broken than it used to be.
[12:22] <lisaDusseault> <eve> What I've seen is people were groomed for these jobs. If you can find effective participants in the WGs themesleves in a few years you're gong to have more IESG members .
[12:22] <lisaDusseault> <eve> How about more recruiting? I see not a ruling class but an informed class.
[12:23] <lisaDusseault> <eve> we need more worker bees
[12:24] <lisaDusseault> <mw> Where do we want the balance of power between ADs and WGs?
[12:24] <lisaDusseault> <mw> The other issue, whether or not our selection processes are fair.
[12:25] <lisaDusseault> <jonne> We should see documented decisions with reasons, and currently that's not the case.
[12:25] <lisaDusseault> <jonne> Without that people get alienated.
[12:25] <mrose> comment on mw's remark: that's very perceptive
[12:26] %% falk has left.
[12:26] %% falk has arrived.
[12:26] <lisaDusseault> <elw> Back to slides: our decision making processes are flawed.
[12:26] <lisaDusseault> <elw> Our participants and leaders are inadequately trained.
[12:27] <lisaDusseault> <?> 10 yrs ago there was a lot of research on this like DLs.
[12:27] <mlshore> Christopher Allen
[12:27] <lisaDusseault> <CA> We don't teach people things that are relatively well-understood like how to manage DLs.
[12:27] <lisaDusseault> <CA> Mail has specific advantages, disadvantages - need to know those to not get into ttrouble.
[12:28] <lisaDusseault> <CA> Changing our mediums e.g. to web-based thing has huge effect on the way we work.
[12:28] <lisaDusseault> <resnick> The last bullet may also be a perception point that people are perceived as not knowing enough.
[12:28] <lisaDusseault> <resnick> The more you distrust the other end, the less contact, a spiral increasing distrust.
[12:29] %% randy has left.
[12:29] %% randy has arrived.
[12:29] <lisaDusseault> <mw> I couldn't even define what wg chairs are and do because every time I state this somebody tells me that's now what it is.
[12:30] <lisaDusseault> <CA> There's a mentoring process that may be broken due to scale.
[12:30] %% ole has arrived.
[12:30] <lisaDusseault> <CA> There are brand-new chair swith no explicit mentor. Other orgs have people that become nominated in advance of getting the job and work under somebody.
[12:31] %% dcrocker has left.
[12:31] <lisaDusseault> <hta> Pedantic nit: In the last bullet on this slide, you say "ruling class". Two slides back that was IESG/IAB. Now it's applied to the "old farts network".
[12:31] <lisaDusseault> <hta> Bob Hinden's in one and not in other.
[12:32] %% resnick has left.
[12:32] <kmoore> fwiw, the process has changed a bit since mrose was on iesg, and some of the stuff that was fixed was specifically designed to address the 'one person can veto' problem
[12:32] <falk> wireless goes down at noon, right?
[12:33] <lisaDusseault> <bob> Getting new people into the ietf to become leaders... it's hard for new people to become leaders.
[12:34] <lisaDusseault> <bob> The last couple nomcoms have cycled people between IAB and IESG.
[12:35] <lisaDusseault> <?> An AD can block things but that's sometimes appropriate.
[12:35] <mlshore> Joel Halpern
[12:35] <lisaDusseault> <JH> I once wanted to undertake work but the AD quite properly said that's not something we ought to do.
[12:35] <lisaDusseault> <JH> We did a good reexamination of the problem.
[12:36] <lisaDusseault> <Elw> I don't have problem with saying no. It's when you see a black hole.
[12:36] <lisaDusseault> <bob> At the beginning of process, saying no is good. At end, that's bad.
[12:36] <lisaDusseault> <Elw> ADs need to take more notice at beginning stages.
[12:37] <lisaDusseault> <Ted> The doc is missing a critical thing - a series of metaphors built into our processes about consensus.
[12:38] <lisaDusseault> <Ted> Our goals are defined by consenus but in fact we use the mechanism of disputation.
[12:39] <lisaDusseault> <Ted>Disput'n is fine but leads to acrimony. Our mechanisms don't allow us to call the question between two sides but instead insist on consensus.
[12:39] <lisaDusseault> <Melinda> That didn't make it into the doc in that form but we did discuss "decision process" in general.
[12:39] <lisaDusseault> <Ted> Would be good to have your formulation in draft.
[12:39] %% ole has left.
[12:40] <lisaDusseault> <Crocker> The previous several folks have been insightful. However We have plenty of ability to call the question.
[12:41] <lisaDusseault> <Crocker>Ted's probelm is real but it's more on the training side than structural.
[12:41] <lisaDusseault> <?> The process was created in 92 as a way to express that decision should not come from authority.
[12:41] <falk> Christian Huitema
[12:42] <lisaDusseault> <CH> We have lost some of our focus on working code and rough consensus and become constipated.
[12:42] <lisaDusseault> <CH> You judge that a sol'n is good by how it is implemented
[12:42] <lisaDusseault> <CH> We try to control too much.
[12:42] <lisaDusseault> <CH> That prevents trying ideas, just becomes somebody believes it's a bad idea. The way to orot out bad ideas is the market.
[12:43] <lisaDusseault> <Eve> We've pointed to good examples of good IDs in the past. OTOH, we don't have documents how to traing WG chairs.
[12:43] <lisaDusseault> <Eve> Can we write down what makes a good WG chair?
[12:43] <lisaDusseault> My note on Eve's statement: I've tried to canvass people on what makes a good WG chair and there is not agreement.
[12:44] <lisaDusseault> <Ted> WGs are charged with making decisions good for the Internet as a whole.
[12:44] %% Olaf has left.
[12:44] %% randy has left.
[12:44] <lisaDusseault> <Ted> We don't empower WGs any more. We take responsibility for rejecting
[12:45] %% randy has arrived.
[12:45] <lisaDusseault> <Elw> Thank you for ripping me from limb to limb. Very good discussion.
[12:45] * falk thinking 'boy that's low expectations'
[12:45] <hardie> No, no. Thank you for not ripping me limb from limb!
[12:45] %% hardie has left.
[12:45] <lisaDusseault> sorry!
[12:46] %% hta has left.
[12:46] %% jhutz has left.
[12:46] %% jhutz has arrived.
[12:46] %% jhutz has left.
[12:47] %% Bill Fenner has left.
[12:47] %% agall has left.
[12:48] %% hildjj has left.
[12:49] %% smb has left.
[12:50] %% avri has left.
[12:51] %% mrose has left.
[12:54] %% kmoore has left.
[12:54] %% mlshore has left.
[12:57] %% falk has left.
[13:00] %% mark.ellison has left.
[13:02] %% lisaDusseault has left.
[14:24] %% azinin has left.