[EM] help w. planning/working on a monte-carlo simulation?

[Abd ul-Rahman Lomax]

At 10:28 AM 5/17/2013, David L Wetzell wrote:
>The Droop quota is often presumed for proportional representation
>over the Hare quota that is more proportional, due to how the Hare quota
>can result in a minority being in power.  (I guess the majority get 
>in power only a
>majority of the time with a Hare Quota.  )

It's unlikely to result in a "minority being in power." A minority 
may get a disproportional number of seats, based on what is done when 
no more candidates get the quota. That is only one possible solution; 
it's based on the assumption that *all seats must be filled.* Indeed, 
since all seats meeting the Hare quota is pretty unlikely, setting 
the quota to aim for N+1 seats and then *not awarding that last seat 
unless the quota is met*, is a kind of solution.

Asset demolishes the problem. One would indeed set the number of 
seats one high and tolerate it being met, because if it is met, 
*every voter is represented, no exceptions.*

There is another solution as well. We actually held an Asset 
election, for the Election Science Foundation Steering Committee, and 
it seemed possible that only two candidates would get the Hare quota. 
The election was defined unilaterally by one of the candidates as 
electing the top three after transfers. The transfers made it Asset. 
I'd never have set the "top three" criterion. I was first in votes; 
this candidate was second, and Warren Smith was third, and there were 
two others with votes. I had enough votes to be elected by the Hare 
quota, with some votes left over. Nobody else had the Droop quota 
without transfers. I had enough votes to push Warren over the Hare 
quota, as I recall, certainly it was over the Droop. So since we had 
no clear rules, and the declaration of the candidate as to the 
method, was unilateral, and Warren and I now represented a 
supermajority of voters, I felt free to handle this the way I 
generally desire: push for maximized consensus. So I essentially 
declared Warren elected and, if we had been unable to agree on the 
third seat, we still had two members of the committee and could have 
made any necessary decision on behalf of the full electorate by 
mutual consensus. As it happened, the candidate in second place 
decided to finish the process by awarding his votes to another 
candidate. That left one candidate who was holding two votes. I asked 
him if he approved the election. He did.

So we elected three seats with *unanimity*. All candidates had the 
Hare quota. As policitical scientists about this, they will tell you 
it's impossible. Sure, it was not really difficult because we only 
had 17 voters. But how does one elect a 3-person committee with 17 
voters, and *quickly* manage that it's unanimous?

If the 2nd candidate had held out, we could have then done whatever 
necessary to move on. Personally, I'd have invited him to participate 
fully, but without a deciding vote, unless an agreement had been made 
transferring those votes to him. That wasn't impossible. I wasn't 
actually *opposed*, I merely wanted to see what would happen if I 
held back. With a more formal method, the same thing could easily 
have happened.

Basically, the elector left with the two votes had an obvious way to 
cast them. If they were not cast, all those who voted for that 
elector would not be represented. That's a high social pressure to 
find the best compromise.

What disturbs some people is that *this process cannot be predicted,* 
at least not with high confidence. I'd not have predicted the 2nd 
candidate would do what he did. It was a surprise. It upset one of 
the voters, in fact, who thought that somehow this person had been 
deceptive. "Why did he run if he wasn't willing to serve?" But he 
*was* willing to serve, he simply made a compromise, for his own 
reasons. However, electors in Asset elections need not have any 
intention to actually serve, and they could even announce that in 
advance. They serve in the process of the election, that is the 
promise that they make, not to actually move to some location and 
spend their life dealing with endless minutia, the real service of 
real polticians.

So: an option isto just leave that last seat vacant if there is no 
compromise found. And a way to handle that in the Assembly is to 
treat this vacant seat as part of the basis for a majority. Or just 
neglect it. However, Asset creates a body of *pubic voters* -- we did 
all the negotiations on a public mailing list; we *could* have 
negotiated privately and then announced our decisions as to vote 
transfers, but we didn't -- and it's possible, because these are real 
people with established identities (I'd require that for electors to 
be eligible), internet voting becomes possible with high security. 
It's secret ballot internet voting that's a problem.

They electors have the elected seats to participate in deliberation, 
and to vote for them *by default,* but if the electors vote directly, 
the votes that went to them from the elector would devalue the vote 
of the seat fractionally. My sense is that this would be rare as 
making any difference, and it might be routinely neglected in 
reversible procedural decisions (i.e., later it would be revealed 
when the internet voting was considered). It is also possible that 
the internet voting would be real-time. I.e., an elector would be 
watching the proceedings over the net, see a vote, and vote 
immediately. Later is too late. This is the same restriction as is on 
seats. Be there or be square. Or the elector is actually present, in 
the gallery. Internet voting by phone would be simple, and they, as 
electors, would have a secure account on a legislative wireless 
network. Remember, all votes of seats and electors *must be public* 
for this system to work. Some legislative process requires secret 
committees. Sorry, electors, that's not where you can watch and vote. 
You don't have adequate proven public trust, only seats have that.)

A step toward that would be to simply allow electors with unassigned 
votes to vote them. Voting is a small part of what an Assembly 
actually does. Most of the business is in process, in entering 
motions and debating them on the floor or in committee. That would 
only be open to elected seats. That, however, is all public process, 
and can be followed by electors *if they wish*. Electors who are 
widely trusted may wish to do this. An elector holding a couple of 
votes, or even just one (perhaps their own?) may find it more trouble 
than it is worth....

Natural consequences: compromise and gain access, or don't. But, with 
these systems, no votes need be wasted. Period. Unless they are cast 
for a poor candidate. (And it's possible for candidates to name a 
proxy or successor if they can't function. They should be trusted to 
do that. It would be voluntary, except in Asset systems that, for 
voter security, disallow becoming a public voter with less than a 
threshold of votes. That's to make voter coercion more difficult. Not 
necessary in many implementations, only under "difficult conditions." 
Under those conditions, the votes would be amalgamated by proxy and 
assigned to an elector as indicated, by the system, secretly. I don't 
like this at all, but necessity is necessity.)

>And since the amount of proportionality with a droop quota
>  gets watered down as the number of contested seats is reduced, this has led
>some activists/experts, like Douglas Amy,  to insist that PR use at 
>least 5 seats.
>This is often coupled with an insistence on rank choice voting due 
>to the problems with party lists.

Party lists are very attractive. The reason is the same reason that 
led Lewis Carroll to invent Asset Voting. Most voters only know their 
favorite. They really don't know enough to rank others. So they 
bullet vote, when they can. (Much of what FairVote imagines is 
reaction to LNH failure, in the old Bucklin elections, is really just 
this phenomenon, and the same thing happens with IRV when voters are 
not forced to fully rank -- which many then do by just checking down 
the list, which is why they need to do Robson Rotation.... you don't 
get good information out of voters by coercing them. Bad Idea.)

Voters know parties even better than candidates, and they know their 
favorite. The knowledge persists beyond a single election, and they 
can probably rank parties as well.

The reason Demoex was able to gain a seat in their City Council was 
because of party list. It made it simple and cheap. They have not 
been as successful as they might have been because they created a 
no-party party, which then proceeded to behave like a party with 
*high discipline*, i.e., a winner who was pledged to only vote *as 
instructed by the party process*. And when another small party 
started up, that might have been similar in some positions to them, 
they treated it like the enemy because they lost votes. Demoex, then, 
in spite of the brilliant ideas in some of what they did, proceeded 
to demonstrate the hazards of being a political party. Even if it is 
theoretically open, in practice, it can easily become "Us First." 
They didn't cooperate. Had they set up Demoex to *advise* candidates 
-- including their own -- but not to control them, and had they 
elected the person most trusted to represent the voters *by voting 
their own conscience* -- they would have fit precisely into 
democratic traditions. As it is, they were bucking them, creating a 
distaste for them on the City Council. It's obvious from what they 
have reported. And they think they were *right*. They were simply 
naive, that's all. They could still recover if they realize what they did.

The simulations are a great idea, but a common failure of simulations 
is to not understand how people vote. Most people will bullet vote, 
when they have low information. Complicating this is that some people 
will bullet vote when they have high information.

I do suggest discussing the simulations on the Center for Election 
Science mailing list.

https://groups.google.com/forum/#!forum/electionscience

>
>
>So I'd like to simulate the effects of using 3-seat LR Hare for a 13 
>seat city council election, like in MInneapolis, MN.
>
>We'd consider 7 cases:
>1. 13 FPTP elections.
>2. 13 IRV elections, as are used now.
>3. Four 3-seat LR Hare elections with 1 at-large seat with IRV.
>4. A 6 and a 7 seat with Droop quota election.
>5. A 6 and a 7 seat with Hare quota election.
>6. A 13 seat with Droop quota election.
>7. A 13 seat with Hare quota election.
>
>I'd like to measure relative proportionality and the probability of 
>a majority getting a ruling majority, the portion of 
>close/competitive elections, and maybe some other stuff that cd be of interest.
>
>Anybody interested?
>
>My intuition is that smaller-order PRs retain the 
>constituent-legislator relationship and would be preferred by many 
>who like having their council-person.  I also think that the Hare 
>quota is more important for increasing the likelihood of having a 
>competitive election and giving minority groups a higher chance of 
>being swing voters.  If this is paired with the use of an at-large 
>seat or some other way of establishing a hierarchy who can get 
>things done, it might be a winning combination.
>dlw
>
>
>On Wed, May 15, 2013 at 2:02 PM, 
><<mailto:election-methods-request at lists.electorama.com>election-methods-request at lists.electorama.com> 
>wrote:
>Send Election-Methods mailing list submissions to
> 
><mailto:election-methods at lists.electorama.com>election-methods at lists.electorama.com
>
>To subscribe or unsubscribe via the World Wide Web, visit
> 
><http://lists.electorama.com/listinfo.cgi/election-methods-electorama.com>http://lists.electorama.com/listinfo.cgi/election-methods-electorama.com
>
>or, via email, send a message with subject or body 'help' to
> 
><mailto:election-methods-request at lists.electorama.com>election-methods-request at lists.electorama.com
>
>You can reach the person managing the list at
> 
><mailto:election-methods-owner at lists.electorama.com>election-methods-owner at lists.electorama.com
>
>When replying, please edit your Subject line so it is more specific
>than "Re: Contents of Election-Methods digest..."
>
>
>Today's Topics:
>
>    1. My cycle definition of the Schwartz set was incorrect
>       (Michael Ossipoff)
>
>
>----------------------------------------------------------------------
>
>Message: 1
>Date: Tue, 14 May 2013 16:18:25 -0400
>From: Michael Ossipoff <<mailto:email9648742 at gmail.com>email9648742 at gmail.com>
>To: <mailto:election-methods at electorama.com>election-methods at electorama.com
>Subject: [EM] My cycle definition of the Schwartz set was incorrect
>Message-ID:
> 
><CAOKDY5Dv9vjoYwO=tXpgfjh=<mailto:CgG73XjPr5_7W6bLkYZ95D4ayQ at mail.gmail.com>CgG73XjPr5_7W6bLkYZ95D4ayQ at mail.gmail.com>
>Content-Type: text/plain; charset=ISO-8859-1
>
>I wanted to express the beatpath definition of the Schwartz set in a
>simpler and more compelling or appealing way, and the cycle definition
>(that I've posted here) seemed such a simplification.
>
>But the cycle definition doesn't define the Schwartz set. A candidate
>that doesn't have a defeat that isn't in a cycle isn't necessarily in
>the Scwhartz set (as defined by the unbeaten set definition and the
>beatpath definition].
>
>Of the two definitions (unbeaten set and beatpath), the beatpath
>definition desn't have much compellingness. For compellingness, I much
>prefer the unbeaten set definition.
>
>Let me state both definitions here:
>
>Unbeaten set definition of the Schwartz set::
>
>1. An unbeaten sets is a set of alternatives none of which are beaten
>by anything outside the set.
>
>2. An innermost unbeaten set is an unbeaten set that doesn't contain a
>smaller unbeaten set.
>
>3.The Schwartz set is the set of alternatives that are in innermost
>unbeaten sets.
>
>[end of unbeaten set definition of Schwartz set]
>
>---------------------------------------
>
>Beatpath definition of Schwartz set:
>
>There is a beatpath from X to Y if X beats Y, or if X beats A and
>there is a beatpath from A to Y.
>
>If there is a beatpath from Y to X, but not from X to Y, then X is not
>in the Schwartz set.
>
>Otherwise X is in the Schwartz set.
>
>[end of beatpath definition of the Schwartz set]
>
>Michael Ossipoff
>
>
>------------------------------
>
>_______________________________________________
>Election-Methods mailing list
><mailto:Election-Methods at lists.electorama.com>Election-Methods at lists.electorama.com
>http://lists.electorama.com/listinfo.cgi/election-methods-electorama.com
>
>
>End of Election-Methods Digest, Vol 107, Issue 9
>************************************************
>
>
>----
>Election-Methods mailing list - see http://electorama.com/em for list info