[wplug-plan] Election rule change proposal

[David Ostroske]

Hi Vance,

Thanks for the response, and I did read everything you wrote, too. I
responded to what needed some additional comments.

On Wed, Jun 10, 2009 at 11:37 PM, Vance Kochenderfer<vkochend at nyx.net> wrote:
[...]
> I understand, and honestly wish the current system was workable, as it
> does offer some nice features.  It just doesn't seem to be working in
> practice.

Thanks, Vance. I'd like to make it more practical, without sacrificing
those nice features.

[...]
> Actually, the idea was to leave out as many details as possible and
> use the defaults in Robert's.  I'll explain more below (page and line
> references are to Robert's Rules of Order Newly Revised, 10th ed.).

Seems decent enough, so long as it's clear what those defaults are.

[...]
> Yes, there would be one vote for each seat.  Voters could cast up to
> five votes for up to five candidates (p. 427, ll. 6-11; p. 402, ll.
> 1-11).  Cumulative voting (i.e., permitting a voter to cast more than
> one vote for a particular candidate) is not allowed (p. 429, ll.
> 16-29).

Right. I do not advocate cumulative voting anyway, for theoretical reasons.

[...]
>> It's true that we've had a few invalid ballots each year. From the
>> times I've served as teller, I can say that the invalid ballots were
>> all caused because the voter did not read the instructions and just
>> put X's next to people's names.
>
> I seem to remember seeing someone giving 5's?  Maybe I dreamed that...

You're right, somebody did cast 5's once. I can't recall what we did
with that ballot. I do seem to remember that the ballot would have
made no difference to the outcome. But this case (5's instead of X's)
just proves my point that fill-in-the-blank leads to problems.

>> This can be fixed by putting a 1, 2, and 3 next to the nominees'
>> names and instructing the voters to circle numbers, instead of
>> putting in a blank next to the names and crossing our fingers.
>
> I can't recall seeing this before, and I hope you'll forgive me for
> not going through my 2-1/2" stack of papers from the bylaws-drafting
> process to look for it.  :)  I agree that this would be a great
> improvement, and support it if we decide to stick with the current
> voting system.  However, people put X's (and 5's) in their ballots
> despite BOLD CAPITAL instructions not to, so who knows what creative
> ways of spoiling their ballots they will be able to come up with using
> your proposed format?

Who knows? At least we're not using butterfly ballots. :) But after a
certain amount of fussing, we can't do much else to prevent stupid
voters from messing up their ballots. I take comfort that there's only
been one or two invalid ballots each election.

If we redrew the ballot now, I'd put each nominee in a box with their
three numbers. That way, voters couldn't see the numbers as columns
and get more confused:
+-------------------------------------+
| Able Baker         1      2      3  |
+-------------------------------------+
| Charlie Delta      1      2      3  |
+-------------------------------------+

We might even do something like this, depending on space considerations:
+----------------+
|   Able Baker   |
|  1    2    3   |
+----------------+
+----------------+
| Charlie Delta  |
|  1    2    3   |
+----------------+

> This still doesn't address the problem of people undervoting.  I
> wonder (but don't know for sure) whether voters are limiting the
> number of choices because a) they mistakenly believe they are limited
> to rating only five candidates or b) they believe doing so will
> increase the chances of their favorite candidate winning.

Could be a little of both. Most voters vote for every nominee, though.
A few only vote for 5. In either case, the method is robust enough to
handle such selectivity.

[...]
>> The solution to this problem is NOT to drop the majority-consent
>> conditions, which would be easy but foolish. We don't want vote
>> packing to occur, and majority consent does prevent this, despite the
>> extra complexity it may cause.
>
> I don't understand what you mean by "vote packing" here.  If a small
> evil group joins WPLUG just because they want to elect Steve Ballmer
> to the board, the defense of the other members under a plurality
> system is to vote for someone else.  If the evil group is large
> enough, then Ballmer can be elected under either system.

With the current consent condition, a majority of voters would
necessarily have to give Ballmer at least 1 vote for him to get
elected. It would take a pretty big group of "evil" voters to swing
the election.

But at the same time, if we went by points alone, it would still take
a lot of monkey enthusiasts to get Ballmer on the Board. So I admit
that the vote-packing argument is fairly weak.

Still, the condition does lead us to nominate a lot of folks, and that
in itself makes the condition worth keeping. I doubt that we'd be
self-disciplined enough to nominate enough folks without a separate
need for it, and the nomination process does give us a way to
acknowledge all the people that are willing to help WPLUG out.

[...]
> I have to give you props for actually having served as a teller.  :)

Hey, props to you, too! :)

> However, you forgot to mention applying the multipliers and totaling
> up the points received by each candidate.  It really isn't that
> difficult when you understand the system, but my experience from
> serving as a teller in one election was that the other tellers never
> really did understand the system, and I ended up filling in the sheet.

If all you're going by is the teller sheets to understand the system,
then I think we'd rightly be confused. And no offense intended here,
Vance, because I know you had a hand in these templates.

What could we do? We could have the tellers just write down the counts
for each candidate from the ballots and just add them up:
+-------------------------------------------+------------+
| Able Baker:     3 + 3 + 1 + 3       = 10  | 4 ballots  |
| Charlie Delta:  2 + 3 + 1 + 3 + 1   = 10  | 5 ballots  |
+-------------------------------------------+------------+

That's fairly simple to wrap our minds around. Tellers would add up
the digits for scores, and count the (non-zero) digits to get the
number of ballots for consent. And if we have to break ties, then it's
easy to just pick out the 3's.
+------------------+--------+------+------+-----------+
| Tied candidates  | threes | twos | ones | Result    |
+------------------+--------+------+------+-----------+
| Able Baker       |   ** 3 |    0 |    1 | ** Winner |
| Charlie Delta    |      2 |    1 |    2 |           |
+------------------+--------+------+------+-----------+

We should produce a set of step-by-step instructions to speed up the
tellers' process. Make it a simple algorithm.

[...]
> Robert's does generally speak against a plurality vote (p. 391, l.35-
> p. 392, l. 13).  A major reason for this is because multiple ballots
> do not pose undue hardship under its standard election method.  Our
> requirement that elections be conducted by absentee ballot makes
> reballoting difficult and time-consuming, and Robert's recognizes that
> organizations which vote by mail can avoid delay and expense by
> electing by plurality.  It does recommend preferential voting over
> plurality voting in such cases, though.

Yeah, preferential voting has the great advantage that you only need
one ballot, and you can go by the lower preferences if needed instead
of calling for another round of voting.

My problem with plurality voting stems from its theoretical shoddiness
as a voting method. And also, for an all-volunteer group like ours, we
should expect our voting method to reflect the voters' real
assessments of the nominees, which is something that block voting
doesn't allow for. There can be good people who do not get
acknowledged for putting their hat in the ring. The limited number of
votes in the block voting method prevents that.

> My main reason for choosing plurality voting is because it is a system
> that people are most likely to be familiar with from school board and
> municipal elections.  Because it is conceptually simple, it is easier
> for the tellers, and easier for the voters to understand.  I feel that
> complexity tends to lessen voters' confidence in the system, since
> they may not understand how their vote influences the result.  And we
> have seen that voters who misread the instructions spoil their ballots
> and end up having zero influence on the result.

True, block voting is simple. In fact, it's too simple, as I've argued before.

I do want to improve the current system by making it easier to vote
and easier to tally. But I'm not inclined to move to a traditional but
faulty method. I'd like to hear some more perceptions and experiences
about using the current system.

I'll likely see Bob Turley this Sunday, so I'll ask him about his
recollections of being a teller for us a few years ago. That was the
year that you were also a teller, wasn't it, Vance?

-- Dave O