--On Monday, August 22, 2022 19:35 -0500 Pete Resnick
<resnick=40episteme.net@xxxxxxxxxxxxxx> wrote:
On 16 Aug 2022, at 22:41, Michael Richardson wrote:
Michael StJohns <mstjohns@xxxxxxxxxxx> wrote:
> On 8/16/2022 5:35 PM, Michael Richardson wrote:
>> Eric Rescorla <ekr@xxxxxxxx> wrote: >> qualified
>> volunteer.
They have
>> no control over who they could >> transfer to, their
>> only
choice is to
>> transfer or not.
>>
>> > Correct, but they get to look at the next person on
>> > the list.
>>
>> Given the two from one org limit, what's the attack
>> here?
> DId you already forget what happened last time
> around? E.g.
I'm the
> second member from a company, and the 11th person is
> the third
member.
> I decline to serve, the other guy picks it up and the
> company
doesn't
> lose a slot.
I'm still actually lost.
The company still has only two people, right?
It's just not the same two people. How is it a win?
Why would 11th person be more or less honest than you?
If the any of the two people picked from the company is an
inexperienced and non-influential member of the community and
#11 on the list is a very experienced and influential member,
the three can decide who has the best skills to influence the
NomCom to pick the candidates of their choice. The company
loads the volunteer pool in order to maximize the odds that
*someone* in the company gets chosen, and if lucky (with
enough people in the list the odds are high), that #11 on the
list is a person from the same company and then they get to
choose which of the three will serve on the NomCom.
The key is that nobody should be able to influence the makeup
of the NomCom other than removing themselves.
Pete,
I see an opposite way to game the system and no good solutions
(or none better than what Rich (and Andrew) have decided it is
best to do). Suppose someone (either selected or not) takes a
look at the overall composition of the proposed Nomcom --
particular people as well as companies represented -- and
decides they don't like it, e.g., that some agenda that they are
interested in pushing is not sufficiently represented. If they
are on the list and the plan is to repeat the randomization and
draw, all they have to do to get a complete new draw is to
decline to serve. Will that improve things in the direction
of a Nomcom composition that would make them happy? No way to
guess. But, if they are already upset about the composition on
one draw, the odds that a second draw will make things even
worse (from their perspective) are probably fairly low.
And, if they are not part of the Nomcom on the initial draw, all
they need to do is to persuade someone who is part of the draw
to say "so sorry, I'm not going to serve".
The "if someone declines, start over" model could also lend
itself to a DoS attack. Suppose there were a company who had
decided that the IETF had outlived its usefulness and was
willing to put resources into killing it off. They would make
sure a sufficient number of people (whose behavior they could
control) were part of the pool to just about guarantee that one
or more of them would be selected as part of the initial draw.
Then one of those people would decline to serve, forcing a
second draw (going back to selection of new randomness sources,
etc.). The second draw would then require a challenge period.
Assuming at least one of "their" people were selected that time,
they could decline to serve, forcing a third draw, etc. As
Rich, IIR, more or less pointed out, those processes are
time-consuming. It would not take too many iterations
(especially if there were a basis for throwing in one or more
appeals) to push actual seating of a Nomcom that could start
considering candidates far enough into the year to make their
doing a competent job nearly impossible or at least so
time-consuming as to encourage others to decline to serve.
If we are worried about the selection process being gamed, it
seems to me to be at least as useful to concentrate on the
situations in which people volunteer to serve, are selected, and
then decline (noting that Rich's process of going down the list
turned up two more such people and, of course, we cannot know
their reasons). Can we create disincentives to people ending
up in that position? For the scenario you outline of a selected
person stepping down to allow a more effective person from the
same organization a seat, we could change the rule such that, if
a company got to two (or more) members and one of them stepped
down, any further possible Nomcom members from that company were
excluded as the Chair went down the list. One could make that
even more harsh by deciding that, if anyone were selected and
stepped down, no one else from their organization could be
selected. That way, someone stepping down would cost the
company any possibility of having two employees (or even one) on
the Nomcom rather than working to their advantage.
best,
john
