Re: guidance on avoiding too many people from one company/group

[Eric Rescorla <ekr@rtfm.com>]

Dave Crocker <dhc@dcrocker.net> writes:
> Unfortunately the selection process must rely on a rather small
> "population", namely folks who self-select to offer to participate. (The
> pool of candidate nomcom participants is inherently biased, because it is
> self-selected and because even as many as 100 volunteers makes for
> statistically problematic pool.) So it is not suprising that -- as we have
> seen -- the resulting group of nomcom participants is not random enough to
> ensure community comfort.
Dave,

As you know, it's very hard to draw these sorts of line and I think
that all of these lines are very blurry.  For example:

(1) I spent a year as a contractor working essentially for Nokia,
despite the fact that I was never an employee. Do I count as
working for Nokia?

(2) Charlie Kaufman works for Iris which is wholly owned by IBM.  For
years, his e-mail was Charlie.kaufman@iris.com and he had Iris
Associates on his business card. Does Charlie Kaufman work for Iris or
IBM? This month IBM shut down iris.com e-mail addresses. Who does
Charlie work for now?

(3) Harald Alvestrand is a European working for an American
company. Next year he will be based in the US. Is he American or
European?

(4) I've got a friend who was born in India and is obviously 
ethnically Indian but is a naturalized US citizen. Is he American
or Indian?

I've got a counter suggestion that doesn't require creating quite so
many quotas and drawing bright lines. As I see it, the problem is
primarily the small self-selected sample (as you mention
above). Obviously, any non-random sample won't represent the main
population as well as the population itself.

We'd like to completely remove the element of self-selection as the
state tries to do with jury duty. Since we can't force people to
serve, we can't eliminate it completely. However, we can substantially
reduce it.  The simple fix is to randomly select nomcom members from
the eligible pool. We don't require people to serve but
psychologically people are much more likely to serve when specifically
selected than if asked en masse to volunteer.

If there's some desire to favor volunteers we could draw part of
the nomcom from a volunteer pool as well, as you suggest.

-Ekr

--
[Eric Rescorla                                   ekr@rtfm.com]
                http://www.rtfm.com/