[EM] Associational Proportional Representation (APR)

[Richard Fobes]

I'm responding (via Bcc) to Steve Bosworth's earlier reply to my 
responses, which he repeated in a direct message that is copied below. 
I no longer have a copy of the forum message, so please pardon the 
creation of a new thread about a conversation in progress.  For context, 
see below.

Steve, I only had time to quickly look at your two flowcharts (which 
were in PDF format, in contrast to your ".doc" documents which I don't 
open for antivirus reasons), but ...

I saw that your Associational Proportional Representation (APR) method 
involves eliminating a candidate based on having the fewest number of 
votes (after possible transfers of votes).

I favor methods that look deeper than each voter's currently top 
remaining choice.  I don't like methods that only look at one voter's 
currently "top choice" at a time.  Why?  They have the same weaknesses 
as plurality voting and instant-runoff voting (IRV), which look at which 
candidate gets the most, or fewest (respectively) "votes."

Methods that involve the transfer of each voter's vote are open to 
strategic manipulations.  You asked for more specifics.  As a partial 
answer, the election results are vulnerable to strategies that control 
which candidates are nominated.  Usually this manipulation involves 
campaign contributions (with the real source of funds for "spoiler" 
candidates being hidden).

All voting methods fail some fairness criteria, so yours does too. 
Which ones?  I don't know.  That requires time-consuming analysis.
Although your method is not instant-runoff voting, it is similar enough 
that I suspect it would fail many of the same fairness criteria that IRV 
fails.

Of course you can correctly claim that there are no fairness criteria 
for proportional methods, yet I believe your method involves underlying 
algorithms that can be applied to a single-winner method, and that 
related single-winner method has to fail some fairness criteria.

As for the method's proportional aspects, the use of sub-groups -- 
called "associations" in this case -- introduces what can be thought of 
as similar to the mathematics of "rounding" numbers too early (instead 
of waiting until all the calculations are done, and then rounding).

Expressed another way, both the Republican and Democratic parties in the 
U.S. are heavily controlled by the same relatively few people, and the 
result is that voters do not control either political party.  I believe 
that in Canada each party nominates a candidate using voting at a 
convention, but admission to the convention requires paying a fee, so 
that too prevents a majority of voters from controlling any political party.

Looking into the _distant_ future, voting methods will handle 
calculations deeply in ways that do not involve any extra layer of 
subgroups or rounding, and possibly without involving political parties. 
  In the meantime we are stuck with subgroups such as the "electoral 
college" for U.S. Presidential elections, and parliaments/Congress/etc. 
that add an extra voting layer (compared to the future when voters 
eventually will directly vote on issues of concern).  Why not begin now 
to get rid of the need for subgroups?

I am not saying that your voting method is bad.  It might be quite good 
for some voting situations!

I'm just saying -- since you specifically asked me -- that my preference 
is to skip over slight improvements and jump ahead to advanced voting 
methods that look deeply into ballot preferences (beyond one current 
"top" choice at a time) and that avoid the need to segment voters into 
subgroups.

For further context I'll say that years ago a group of people within a 
local food co-op came up with a very carefully designed way of electing 
a group of "representatives" for the purpose of having them make 
decisions instead of letting all the members vote on important 
decisions.  In spite of how well-designed and "fair" (neutral) the 
process was, neither the people who wanted the co-op to sell a few meat 
items nor the people who wanted absolutely no meat in the store were 
willing to let such a group make a decision about that issue.  The point 
of this example is that each layer of decision-making -- even if it gets 
adjusted at every election based on the ballots -- does not truly 
provide proportional representation.  As for what a truly proportional 
solution to that "meat" conflict would have been, I'm not sure.  Selling 
fewer meat items than what a majority of voters wanted would still fail 
to represent the members who didn't want any meat sold.  (It was not 
clear who was in the majority, and probably a middle third of the 
members would have been OK with certain meat choices but not other meat 
choices.)

Ultimately voters don't care about the process.  That's why so few 
citizens "do the math" to discover why they are not represented by the 
people "they" elect.  This same dilemma applies to all the voting 
methods discussed here.  Here we are not only "doing the math," but we 
are developing "the math" relating to voting methods.  Let's eliminate 
extra layers and stop using "start-at-the-top" blinders as we look at 
each ballot.

Thank you for your interest in my opinion.  I hope this helps, either to 
refine your ideas or to refine ways to "sell" whatever method you like 
best.  (All of us here are learning how to "sell" our favorite method(s).)

BTW, thank you for creating the flowcharts.  They do help clarify your 
method.  (Alas, graphics on websites seem to be the only way to make 
flowcharts easy to view, so they are not suitable here in this forum.)

Richard Fobes


On 10/22/2014 6:54 AM, steve bosworth wrote:
> Hi Richard,
>
> Sorry for the late reply. I've been travelling.
> Thank you for your several criticisms, comments and suggestions.
>
> I've *injected my responses within the text of your email bellow, using
> bold print*.
>
> I hope you will see that some of the problems you mentioned are solved
> within the full explanation of my proposed system (Associatonal
> Proportional Representation (*APR*)) that I have fully described in the
> attached article with its illustrative 2 flow charts and 3 tables.
>
> In the light of the more complete information provided, I very much hope
> you will be able to find the time to respond to the additional explanations.
>
> Thank you,
> Steve
>
>  > Date: Tue, 2 Sep 2014 09:10:10 -0700
>  > From: ElectionMethods at VoteFair.org
>  > To: stevebosworth at hotmail.com
>  > Subject: Re: (2) "Severity" of failing
>  >
>  > Steve Bosworth ~
>  >
>  > Thank you for your interest in my opinion.
>  >
>  > Getting to the point of your question, your election method combines
>  > single-winner voting concepts
> *S: No, in effect, APR**is entirely a multi-winner system, e.g. to elect
> the 435 members of the US House of Representatives or the UK House of
> Commons. *
> with proportional representation concepts,
> *S: APR's giving different 'weighted votes' to each rep depending on how
> many citizens had ranked them would provide complete individual
> representative and party proportionality.*
>  > which means that the well-known fairness criteria do not apply.
>
> *S: I know of no such criteria which APR would not satisfy.*
>  > Your idea sounds intriguing. Yet it would encounter time-related
>  > issues,
> *S: Please explain.*
> especially strategy issues,
> *S: Please explain.*
> if it were converted into an actual
>  > election method -- that involves ballots.
> *S: Perhaps you will see that these issues have been solved by the
> detailed presentation of the 'actual method' and the paper 'ballots' to
> be used by APR, and explained by the attachments.*
>  >
>  > The single-winner aspects basically match instant-runoff voting, so the
>  > same fairness-criteria failures would apply.
> *S: No, because it is not an IRV system.*
>  >
>  > As for the proportional part, your method would tend to elect a few
>  > celebrity representatives who are supported by "the media"
> *S: Given APR's 'electoral associations' as selected by citizens months
> before the general election through APR's special 'primary election',
> the relative influence of 'celebrity' and the 'media' might be much
> reduced. In any case, the article stipulates that any very popular rep
> who receives more than 10% of all the votes in the country would be
> required to publish exactly how he will pass on all of his 'extra votes'
> to his trusted fellow reps.*
> and the
>  > other representatives would tend to be "fringe" types who are supported
>  > by fewer voters. Note that this is a tendency, and would be reduced to
>  > the extent that it's noticed, which means that most voters would not
>  > notice this tendency.
>  >
>  > Alas, my time is limited, so I can't offer more feedback at this time.
> *S: Thank you again for your time.*
>  > I hope this is helpful.
>  >
>  > If you want more opinions, I suggest that you present the idea on the
>  > Election Methods forum.
> *S: I keep trying to find out how to do this but have so far failed. Can
> you please explain how one contributes to this forum?*
>  >
>  > Richard Fobes
>  >
>  >