Voting Rules

[DGDanforth]

The 'Short Circuit Voting' measure has brought up the fact that our current voting rules are ambiguous and buggy.
Here I suggest that we explicitly specify what those rules are so they can be referenced in one page and placed on our BlackBox Center Wiki.

I believe our current voting procedure is given by (in pseudo code)

Code: Select all

	PROCEDURE CurrentVote* (VAR option: INTEGER): BOOLEAN;
	BEGIN
		Zero(count);
		REPEAT UNTIL Quorum(count);
		RETURN Preference(option, count)
	END CurrentVote;

Pseudo code has been useful. It pointed out the fact that a time limit on voting had not been well specified or discussed.

Josef has pointed out that termination of a poll when a quorum is reached is not the correct thing to do because the missing members (2) could change the preference of the vote. For example if the counts are
ABSTAIN=1
YES=4
NO=4
a Quorum (9) has been reached but there is no preference and so a vote by the missing members can change the outcome of the vote.
Hence the CurrentVote is inadequate (buggy) and does not work for all cases. If the missing members never vote then the poll continues forever. That is not an acceptable voting rule.

(0) I propose that a finite poll duration always be used and the duration range from 1 day to 1 month with a standard duration of 7 days (1 week) used. The duration is determined by the creator of the poll. A duration of '0' (never ending) is invalid. endtime := starttime + duration.
(1) The number of voting members (numMem: INTEGER) (currently 11) of the center is used to determine a quorum.
(2) The fraction (qFraction: REAL) (currently 0.80) of numMem is used to determine a quorum.
(3) A quorum (qVal: INTEGER) (currently 9) is given by qVal := Ceiling(qFraction*numMem).
(4) Each poll has a number of options (numOps: INTEGER) (depends upon poll) which is 2 or greater.
(5) A poll has a category called ABSTAIN (used in determining the number of votes).
(6) A count (count: ARRAY 1+numOps) is used to record the number of votes for each option. count[0] is ABSTAIN.
(7) The number of votes (votes: INTEGER) is give by votes := Sum(0,numOps, count).
(8) Missing (missing: INTEGER) is given by numMem-votes.
(9) A rank (rank: ARRAY 1 TO numOps OF INTEGER) satisfies rank >= rank[i+1] and is given by Rank(rank, count).
(10) The spread (spread: INTEGER) of a vote is given by spread := rank[1]-rank[2].
(11) A poll can succeed (succeed: BOOLEAN) (succeed=TRUE) if any of the following hold
(a) spread > missing (short cut voting)
(b) endtime is reached and quorum is reached and spread > 0.
Otherwise the poll fails (succeed=FALSE).
(12) The result of a successful poll is the option(option: INTEGER) corresponding to rank 1.
option := Option(count, rank[1]).

I propose that those 13 rules be used to specify what constitutes a Center vote.

[cfbsoftware]

Josef has pointed out that termination of a poll when a quorum is reached is not the correct thing to do because the missing members (2) could change the preference of the vote. For example if the counts are
ABSTAIN=1
YES=4
NO=4
a Quorum (9) has been reached but there is no preference and so a vote by the missing members can change the outcome of the vote.

The whole point of a quorum is so that a vote can be concluded even if some members do not vote (they are unable to attend a meeting, they forget, they cannot be bothered etc.). In real-world situations it is usually when some members are absent from a meeting when a vote is taken. As long as there are enough members present to satisfy the quorum then the vote is counted as valid. A quorum is often a low number (e.g. 5) even if there are hundreds of members.

For an online vote, the concept of being absent or present cannot be used. An alternative is a time limit. I.e. if the number of voters satisfies the quorum and the time limit has passed then the vote is valid.

What to do in the case of a tied vote is a totally separate issue and should not be confused with the concept of a quorum.

I would simplify Doug's rules as follows:

(1) Every poll has a duration of 7 days (1 week).
(2) The quorum is 8 members.
(3) Every poll has at least two options including ABSTAIN.
(4) A poll is concluded when
a) Any outstanding votes would not change the outcome
b) The time limit has been reached
whichever happens first.
(5) The poll result is valid if
a) The number of votes (including ABSTAIN) is greater than or equal to the quorum
b) One of the choices (excluding ABSTAIN) has more votes than any one of the other choices.

I would recommend a shorter voting period (e.g. 2-3 days) and a lower quorum (e.g. 5) to allow some progress to be achieved.

[DGDanforth]

"I would recommend a shorting voting period "
"shorting" => "shorter"?

Chris, I like your rewrite.
-Doug

[cfbsoftware]

"shorting" => "shorter"?

Oops! Thanks - now fixed.

[Josef Templ]

> (1) Every poll has a duration of 7 days (1 week).

That is short enough. No need to make it even shorter, for example 2-3 days.
In practice, most polls will be closed because of 4.a anyway. If we run into
4.b there is always the question how to proceed then. The simplest answer would
be to set a new time limit and to get more members to cast their votes.


> (2) The quorum is 8 members.

should be specified as percentage because the number of center members may change.
Note that it may even become smaller.
In order to avoid coupling the discussion of the voting rules
with a discussion of the quorum, which can be seen as an adjustable
parameter of the voting rules, I would remove it from the rule set.


> (5) The poll result is valid if
> a) The number of votes (including ABSTAIN) is greater than or equal to the quorum
> b) One of the choices (excluding ABSTAIN) has more votes than any one of the other choices.

The intention seems to be that a and b are connected by AND.
Assuming AND there is a severe logical error in this rule.
Who can find it?

- Josef

[ReneK]

I disagree that absence is not a real issue online. Especially during university free periods it has been hard to get a wuorum and aftrwards some of the members claimed to have been offline for lengthy periods of time.

Also I do not like the duration of one week. Esprcially with more complex issues that have not been sufficiently discussed, this could ne too short a time to get knowledge about the issue, and good measires could be cloyed without valid vote for this reason.

[DGDanforth]

I disagree that absence is not a real issue online. Especially during university free periods it has been hard to get a wuorum and aftrwards some of the members claimed to have been offline for lengthy periods of time.

Also I do not like the duration of one week. Esprcially with more complex issues that have not been sufficiently discussed, this could ne too short a time to get knowledge about the issue, and good measires could be cloyed without valid vote for this reason.

Its the duty of the Chair to see that an issue has been sufficiently discussed before a poll is started. I think 1 week is sufficient time for a poll. If members know they will be away (not on-line) for an extended period of time then they should inform the Chair of that fact and the Chair can adjust when the poll is held to accommodate them.

[ReneK]

On an ideal world I agree. In tge real world.... not so much

[DGDanforth]

> (5) The poll result is valid if
> a) The number of votes (including ABSTAIN) is greater than or equal to the quorum
> b) One of the choices (excluding ABSTAIN) has more votes than any one of the other choices.

The intention seems to be that a and b are connected by AND.
Assuming AND there is a severe logical error in this rule.
Who can find it?

- Josef

I don't see it.
For the heck of it I wrote a quick program that randomly created up to 10 options with abstain and missing.
I set the number of members to 11 and the quorum to 9 and let the members randomly choose an option
(including abstain and missing). The program reported valid or invalid along with the votes.
Valid if both quorum and preference were TRUE.
It all looks good to me.

So the logical point must be based on the interpretation of Chris's words.
Would you care to elaborate?
-Doug

[cfbsoftware]

Its the duty of the Chair to see that an issue has been sufficiently discussed before a poll is started.

I agree.

If members know they will be away (not on-line) for an extended period of time then they should inform the Chair of that fact and the Chair can adjust when the poll is held to accommodate them.

In the real world an absentee is usually allowed to nominate a proxy to vote on their behalf.