Notes taken by Elliot Lear
Meeting Chaired by Avri Doria

Jim Galvin discussed the differences between 
draft-ietf-nomcom-rfc2727bis-01.txt and -02.txt

Several Issues were brought up during this discussion:

 One issue that was discussed was whether the definitions of the roles for 
advisors and liaisons reflected current practice or not.  There was a 
disagreement on this point between 2 previous NomCom chairs: Don 
Eastlake and Avri Doria.  Jim maintained that the current text reflects the 
intent of RFC2727.  The old version indicated explicitly who votes on 
what, while new version specifies that only voting volunteers vote on 
candidate issues and everything else is open to other members. The -02 
version lists explicit responsibilities of each role.  This is new text.  
The issue needs to be reviewed by the WG list.

 Phil Roberts brought up an issue about filling replacements.  The 
current random selection process is renewable, but needs 
clarification about how to reinitiate the system.  Right now someone can 
figure out who the next choices would be.  Jim pointed out that this is 
covered in the -02 draft with an additional week to reset the random 
process.

 WG participants were asked to review the latest draft as it contains new 
language in many areas.

After the review of the -02 draft, open issues that had not yet been 
resolved were discussed.

The first issue involved the qualification for volunteering with the basic 
question being was attendance at 2 out 3 of the last IETF meetings a 
sufficient criteria.

Comments:

 Joel Halpern observed that we seem to have a problem of not getting 
enough experienced IETF participants on the nomcom.  We need to 
encourage more participation.  He  didn't like the idea of 2 pools.  Maybe 4 
out of 6 IETF meeting would be better then 2 out of 3?

 Don Eastlake observed that a jury pool model would not work because the 
commitment involved was too substantial.

 Dave Allen commented in favor of allowing people who had attended many 
IETFs to be included, the big N proposal, even they did not meet the 
recent attendance requirements.  He further commented that the process 
could arbitrarily select people from one area in the same way it has it has 
selected people from one company.

 Mark Allman agreed that the 2 out of the last 3 rule is 
insufficient.  Would like to see some weighting of rules for including 
management experience.  Would like to see something that has a little bit 
longer time scale.

 George Swallow commented that after 2 IETF meetings he didn't have a clue as 
to how the IETF was organized.

 Michael Richardson, currently on the NomCom, commented that the current 
NomCom doesn't suffer from lack of experience.  But if you've been to one, 
missed one, and then went to a third, then that's a different problem.  
Things aren't all that broken.  He believes that self selection already 
occurs, based on whether people believe they know what's going on.

 Don Eastlake stated that as long as the nomcom has a finite number of 
volunteers there will always be some bias in some way or another.  It's 
okay for it to be imperfect.  Not sure he'd be happy with an IETF that ran 
purely smoothly.

 Harald suggested two changes:
  - add a statement to the oral tradition 'if you don't know how the IETF 
works don't volunteer for NomCom.
  - change from 2 out of 3 IETFs to 3 out of 5 IETFs

 A hum was requested on the 3 out of 5 versus of 2 out of 3.  The 3 out of 5 
was the unanimous view of the room.

 John Klensin commented that there were 2 reasons to volunteer.
 1 is obligation.
 2 is that one has an ax to grind.
 He's worried about the "don't volunteer" folk self selecting out in favor of 
the ax grinders.

 A hum was requested on adding a statement to the oral tradition of 
"don't volunteer if you don't know what's going on."
 The group seemed to lean toward adding the statement.

 The chair reminded the meeting that a hum is just an indication.  This all 
goes out to the mailing list.

 Jim Galvin asked what goal is the group trying to achieve?

 Harald responded that since he made the suggestion he should say 
something.  One property of mailing list discussions is that they fly off in 
all directions.  It's easier to discuss when you have a specific 
proposal.  So he made a specific proposal.  The problems that is to be 
solved is that we want people who have more experience just than 6 months 
participation volunteering for the NomCom.

 Eliot asked why 3 out of 5?  Why not 4 out of 6 or 5 out of 7, etc?

 Jim mentioned pool of people increases as you go from 2/3 to 3/5, etc.

 Avri responded that we don't really know that.

 George Swallow mentioned that though he hummed in favor of 3/5 he would 
also support requiring a minimum of having attended 5 meeting ever, AND 
attendance at 2 out of the last 3 meetings.

 Avri pointed out that this that would bias the selection procedure in 
favor of long time participants.

 Harald said we should ask the secretariat for numbers.

 Avri asked if the room favored adding a requirements for increased 
number of meeting attended in the past as George recommended.  The room was 
not in favor of the proposal.

 Ron Da Silva had a last comment regarding language stating that it's 
important to encourage people to understand the process.  He suggested that 
the oral tradition indicate where people could learn about the process and 
recommend that they then reconsider volunteering.

-- 
The next issue involved limiting the number of participants from a single 
group or company

In the WG list discussions to date, there was no consensus on 
definition of company, group, or organization, but but there seem to be 
consensus that the maximum number should be 2.

Was a strict definition of control group necessary and should other 
diversity or lack there of be considered?

The meeting reviewed some very legalistic language that had been 
proposed on the list.

Comments:

 Michael Richardson thought that it seems like a good idea.  
Understanding that this would be done only at selection time, and that 
subsequent acquisitions or change of employment would not impact 
membership.

 Ross Callon asked if all the various wings of the U.S. or other 
governments would count as one organization.

 John Klensin commented that when ANSI went down this path for 
balloting, their first cut labeled the U.S. as a single entity, but that 
would have cut attendance to nothing.  Same issue involving large 
companies with separate divisions.

 Ross Callon likes the rule.  Doesn't mind U.S. government agencies being 
counted separately.

 Don Eastlake doesn't expect perfection.  It is unfortunate that we have to 
do something like this.  Where there appears to be substantial bias or the 
appearance such that people bend over backwards to avoid it, then public 
oversight is important.  If people try to game the system then people will 
notice.  When the pools were small, this was impossible.  If you have 100 
volunteers where 10 or more are from the same company, the 
probability of 4 or more from the same company is 1.3%.  With multiple 
large companies, that probably increases to 5% or so.

 John Klensin is suspicious of measuring probability with reality.  We 
should say that packing behavior is inappropriate.  We should consider a 
mechanism for challenging a list of selected members, and a way to 
challenge the company to indicate whether it was packing or not.

 Jim Galvin responded there is in fact a one week challenge period after the 
announcement, but there is no process for actually challenging.

 Don Eastlake believes the criteria needs to be public and 
challengeable, but only for a brief period of time so that you have a 
stable nomcom.  Not sure there is a corporate conspiracy, it's more word of 
mouth.  E.g. he works for a company that gives bonuses based on 
standards activity.  Perhaps there are companies who would remunerate 
based on their involvement.  Wants some sort of rule like what is 
proposed.

 John Klensin would prefer to rely on integrity of individuals.

 Steve Silverman asked "as a consultant, how does he fit in"?

 Avri responded one would state affiliation, so it's up to the 
individual, and is subject to challenge.

 Lou Berger commented we have an existence proof that community 
pressure does not work.  He seconds the rule.  No subjectivity to the 
proposal.  No bias to particular organizations.

 Joel Halpern stated that there are other dimensions as to why we need such a 
rule.  If companies believe it is possible to pack the nomcom, then they 
will try in order to protect themselves.  And we need to make it clear that 
they can't.

 John Klensin completely agrees with Joel.  But he doesn't think this is the 
right solution.  He disagrees that public pressure doesn't work.  We're 
creating lawyer bate through the limitation as it is worded.  If we 
introduce a formal procedure then the courts will decide.

 Harald said that at the time of volunteering a person states their 
affiliation, the person doing the picking rejects people who have the same 
string in their affiliation.  Put the onus on the individual, not on the 
process.  Joel says this proposal has too much detail, but agrees it 
should be made mechanical.

 Marshall Rose claimed that we've already crossed the lawyer bate 
threshold.

 Joel Halpern argued that John's point is valid, but that we can fix it by 
stating that this is a best effort and not perfect.

 John Klensin responded that if a company decides that it's in its 
corporate advantage to game the procedure, they will do it.  The rule is too 
fuzzy.  Get people to self identify, again putting the onus on the 
individual, and then use a challenge process.

 Eliot Lear was concerned about both the perception and the potential 
reality of a government packing the nomcom.  Depending on the 
administration of a government this could be the case.

 Jeff Hultzman stated it doesn't hurt us if more than 2 folk from the same 
government can't serve.  We're not ANSI.

 Avri stated that seems to be a consensus for some sort of a rule, 
choosing the middle path of having people self identify and remove 
themselves.  Asked if that was correct?

 Marshall Rose pointed out that there are 2 issues.
  - Level of formality.
  - Voluntary or mandatory removal.

 Don Eastlake said we need a rule that can be executed by a person, not a 
computer.

 Lou Berger said he was confused by the summary.  Informal or formal?  If 
formal, what's the enforcement mechanism?

 Avri said she believed legalistic language was more a problem than an 
advantage.

 Lou Berger said we ask the ISOC counsel to help us here.

 Avri asked if we want legal language?

 Lou Berger said we still need to ask the ISOC to review the language to 
keep us from getting into trouble.

 Harald informed the group that the language of the present RFC has been 
reviewed and that the language of the next RFC will be reviewed prior to 
adoption. 

 Eliot added that once you have legalistic language, then you need 
lawyers to interpret it, and then you have blood in the water.

-- 

The next issue involved the question of whether there should  be a 
process for resolving challenges over volunteer selection?

The current draft allows time is allowed for challenge but does not 
indicate a specific process. Especially if control group notion is added, 
challenges might be more likely.

The question is whether resolving any challenges should be left to nomcom 
chair?

Comments:

 Jim Galvin responded that to be precise, there is no ADDITIONAL 
process.  Anyone can move for a challenge, and they use the ordinary IETF 
appeals process, with the exception is that the challenge is to be 
directed at the IETF chair.  It's also worth noting, there's no defined 
time period for the IETF appeals process.

 Eliot Lear commented that maybe there should be some time frame.

 Marshall asked Harald how long would it take for him to resolve a 
challenge?

 Harald that it depends.  If the challenge was specious, up to 2 weeks. If it 
was a serious issue, it would be harder to bound.  There has never been a 
challenge to deal with.

 Marshall responded that we have a serious problem, because the IETF 
appeals process wasn't designed for this purpose.

 Harald said we might want to add language saying that makes it 
possible for the nomcom to operate with those who were not 
challenged.

 Jim pointed out that the NomCom is permitted to continue forward, but they 
cannot submit a slate to a confirming body while a challenge is in 
process.  Dispute resolution process has a time limit of 2 weeks.

 Avri pointed out that using the IESG means the possible candidates could 
become the judges in this process.

 Steve Coya suggested going right to the IETF chair and not involving the 
entire IESG.  Harald commented that he objects.

 Lou Berger suggested it go from the NomCom chair to the ISOC 
President.

 Jim Galvin suggested that the NomCom chair runs the process, so it would 
have to be the ISOC president that handled a challenge.

 Don Eastlake pointed out that it would not necessarily be an 
adversarial process in every case.  The chair should be given 
opportunity to correct mistakes.

 A Hum was in favor of Lou Berger's proposals.

---

The next issue involved mid term vacancy procedures

A new procedures has already been documented but has not seen much 
discussion.  The new procedure involves a 3 month overlap for handling 
midterm vacancy.  There would be no overlap for handling regular 
process.

Comments:

 Scott Bradner noted that there is potential for a great deal of 
confusion. Perhaps it would be better for  a seat to sit idle.

 Harald mentioned that a possible vacancy of both members of an IESG area 
could be a serious problem.  In fact it would be disastrous to have 6 
month vacancies.

 Scott Bradner mentioned that the NomCom only acts when the IESG or IAB 
informs them of a vacancy.

 Jim responded that the IAB or IESG didn't have a choice as to which 
NomCom filled a vacancy.  But the IESG or IAB could choose to not to fill a 
vacancy by not informing the NomCom.

 Leslie Daigle pointed out that we must be mindful of what happened in the 
routing area last year.

 Jim Galvin agreed that we should have overlapping NomComs, but that is a 
slippery slope.

 Phil Roberts commented that if you have two NomComs going at the same 
time, one of which is filling a midterm vacancy and the other is filling the 
other vacancy in the same area, it seems strange, and this might lead to 
poor dynamics between the two people who are selected.  If you 
reconstitute a NomCom after they've gone dormant, there is more of a risk of 
having to find replacement volunteers.

 A hum was requested in favor of leaving wording of document as is. A hum 
was inconclusive about adding a restriction about serving in 
consecutive 15 month NomComs.

 Scott Bradner pointed out that question was perhaps confusing.

 Eliot Lear addressed Phil Roberts' point saying that hopefully the old 
NomCom would fill midterm vacancies prior to the new NomCom submits its 
slate.

---

The next topic concerned the recall procedure of IESG/IAB members, and how it 
gets initiated

The current procedure is susceptible to DOS attack.  A mechanism needed to 
protect against this is needed.  The proposal presented involved 
requiring that 2% of average attendance sign a petition in order to 
initiate the current recall procedure.

Comments:

 Don Eastlake thought the procedure should be 2% of number of 
QUALIFIED people according to the 2 out of 3 meetings rule, and that it 
should not be based only on attendance.

 Harald supported Don's change.

 John Klensin indicated that if we are going to have petition lists the 
list should be public.

 Steve Silverman asked if anyone had done this? [NO] He suggested that 3 
people instead of one, would be ok.  But questioned requiring 40?

 Marshall explained that this procedure was as added as a last minute 
afterthought.  Surprising no one has invoked it.

 Harald asked how many people are needed?  Thinking about the end game, is 
there a situation where someone should be removed from office without 40 
people being able to agree. Should you really want to have that person 
removed?
 
 David Putzolu supported the proposal and agreed that it was useful to only 
count nomcom eligible signatures, as otherwise one could have random 
people sign recall petitions. 

 Avri pointed out that the proposal called for the petition signers to be 
qualified to serve on the NomCom

 A hum was generally in favor of the proposal


The meeting was adjourned.