My two cents:
If I were to go to Las Vegas and roll the dice at a crap table, then ask
to roll again because there was a clap of thunder which made my arm
twitch just as I released the dice during the first roll, I assure you
that they would not allow me to do so. They would not be persuaded by
an argument that the second roll would have the exact same set of
probabilities of outcomes as the first, or indeed as any other roll.
And their decision would have nothing to do with the actual result of
the first roll (i.e. it would not matter if I won or lost money, on that
roll).
Here we have a similar situation. We have a randomized process which at
worst, was affected by a name on the list that should not have been
there. While the presence of that name did affect the outcome, like the
clap of thunder, it did not do so in any predictable way, and so the
random ("unbiased") nature of that outcome is preserved. Resetting and
re-running the process is not an appropriate response. Doing so inserts
human intervention (read, "bias") into an otherwise unbiased process.
So also, by the way, would a "let's decide what to do based on a coin
flip" solution. That would be rather like me trying to convince the
casino to do a coin flip to see if I should be allowed to roll the dice
again.
Any solution, other than accepting the results as they originally were
generated, will be biased.
In my opinion we should not rerun the process. Rerunning the process is
the exact wrong thing to do. The decision to do so was no doubt made in
good faith, and with the best of intentions, but it is clearly
incorrect.
-Matt
-----Original Message-----
From: James Galvin [mailto:galvin+ietf@xxxxxxxxxx]
Sent: Thursday, August 31, 2006 2:13 PM
To: todd glassey
Cc: 'IETF-Discussion'
Subject: Re: Now there seems to be lack of communicaiton here...
-- On Thursday, August 31, 2006 10:40 AM -0700 todd glassey
<tglassey@xxxxxxxxxxxxx> wrote regarding Re: Now there seems to be lack
of communicaiton here... --
> James I also agree with Donald's logic -
>
> So then what happens when the selection process is restarted and the
> ramdomizer is used again - say the second time it selects six of the
> same candidates and the rest are different out of a pool of 20 or 30
> probably. How is that fair to those selected in the original pick who
> now loses their potential seat to the process.
I'll only say that RFC3777 defines what it means by "fair and unbiased,"
and I believe that what transpired was within that definition.
Specifically, a process is "fair" if any eligible volunteer is equally
likely to be selected.
Even a restart is allowed by the rules.
Now, was a restart the best choice given the issue at hand?
Personally I think there were other good choices that would have served
the purpose. Even so, it is the Chair's job to make that decision and
he obviously saw the situation differently.
Do I want to change the rules to prevent a restart in the future?
Not just yet, but I'm following the discussion and perhaps I'll change
my mind.
Jim
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf