Hi everyone,

My apologies for the list flakiness and Electowiki.  I admittedly have only
been skimming the list every so often, and I'm considering other mediums
for discussion.  Migration of the wiki software is on the table, too.  I
just need to make the time to play around with all of this.

Jameson's email below prompted me to say something I've been thinking about
anyway:

Jameson Quinn jameson.quinn@gmail.com wrote:

I agree with this wholeheartedly.  I'm not yet familiar with MUMA, so I
can't speak to the dynamics it introduces.  I'm also looking at this with
reasonably fresh eyes.  That may mean my instinct is lacking nuance y'all
have.

Jameson, I'm also not responding to your suggested algorithm just yet.  On
first skim it looks plausible and very good at a technical level, but I
haven't entirely thought through how the complexity is a necessary product.

Here's the simple two round I would propose for U.S. presidential elections:
Round 1: Simple up/down vote on all candidates (approval ballot)  All
candidates over a reasonably high threshold (e.g. 40%) advance to the next
round.  If no candidate reaches 40%, then only the top two advance.
Round 2: Robust single winner system.  Approval would be acceptable, the
various Condorcet-compliant versions, etc.

What voters would be answering in the primary is "is this a viable
candidate that belongs on the debate stage?"  Those candidates who make it
on through the first round would be presumed eligible for debate
participation.

The effect: well, it might be messy involving cases where people aren't
stingy enough with their votes, and allow crackpots through the first
round.  Over time, with enough election experience, primary voters would
learn who to approve.

Thoughts?
Rob

I like a 1-stage system, whether Approval, Score, 3-Slot ICT (Deluxe
Approval), or Bucklin, etc., or something fancier.

Half the cost. ...& it seems to me that, when we discussed it before, there
were other advantages to a 1-stage election.

Michael Ossipoff
On Oct 2, 2016 4:56 PM, "Rob Lanphier" robla@robla.net wrote:

On Sun, Oct 2, 2016 at 6:35 PM, Michael Ossipoff email9648742@gmail.com
wrote:

We've both been advocates for over 20 years, and I used to be convinced of
the mathematical advantages of a single stage election.  However, now that
we all have witnessed the IRV dynamic in more and more settings (e.g.
Burlington 2009), I think I've become convinced that a 2-stage process has
big advantages for making good decisions in high stakes elections.  It's
good to have a first stage for democratic vetting to weed out dangerously
flawed candidates (and to make it possible for the non-crackpots to rise
above the noise created by the crackpots).

The trend in user interface design is toward "anticipatory design"1 (not
overwhelming the user with too many choices).  As advocates, we have touted
the benefits of more choices, but that hasn't helped us.  The problem:

• An election can have 1 clear frontrunner ("establishment candidate"), 2
  other viable choices, and 5 extremists
• The frontrunner can lump his/her opposition in with the extremists, and
  occupy "the center"
• The viable opposition can't raise any more money than any of the
  "extremist" candidates
• The voters don't have the mental energy to tell the difference between a
  viable choice and an extremist
• The frontrunner wins without having to win on merit.

I'm still not an advocate for IRV, but it's also become clear to me that
some of the problems with IRV in practice aren't with the tally method.
I'm living in San Francisco now, and I'm finding that decision fatigue in
elections is also a problem.  An incumbent has an enormous advantage in
the general election with a ranked choice ballot, because the opposition
has a difficult time consolidating around a viable alternative.2

Rob

Let's start from the assumption that having two rounds is good because it improves the quality of the judgement that the voters will make. My first thought is that if Condorcet is a good method for the final round, it could be used also at the first round. One approach would be to replace the first round with one or multiple (official?) polls. It could make sense to arrange the preliminary rounds electronically (in the internet?) to avoid costs and voters getting tired. It is important for a working democracy to maximise the number of voters at the final round.

Arranging multiple similar rounds with the same method has the problem that risk of strategic voting may increase, if the voters and parties know in detain what the current opinion distribution is. One can however fight that by having some time between the rounds. Opinions typically change sufficiently to make at least typical Condorcet strategies difficult to apply.

If the method used in the first round is different from the method used at the final round, then the first round method could be designed so that it has a tendency to introduce new candidates that might be weak signals that will grow before the final round. Your proposal included approval. Approval could be a nice method if voters use it in the spirit of "let's list all the candidates that could be reasonable choices". That could allow weak signals to pop up. One risk is that large party voters would bullet vote for their own candidate in order to eliminate any risk of lifting up new competitors. (Small party voters might be more likely to vote also for some other small party candidates in addition to their favourite candidate.) If voters are dishonest in this way, Condorcet could carry more information, e.g. by pointing out potential midsize compromise candidates. Anyway, slightly different criteria may apply to the first and final rounds.

In many places money appears to be one critical factor in getting someone elected. My preference is to reduce the influence of money. At the internet age there is actually no need for expensive media campaigns. In principle one could distribute all the necessary information to the voters using one official web page that gives sufficient space to all the candidates (and their dynamic debate too). Maybe one round of ads in all newspapers (paid by government?) in addition to that, and maybe one set of street ads next to each polling station, to contact also those voters that are not connected to internet. Few TV debates and interviews (for all candidates, or for those candidates that the first rounds pointed out) would also be good. Multiple rounds means more money spent and less visibility to potential good candidates that are not promoted by some large party or other rich supporters (or huge number of average income supporters).

Juho

Alright, but didn't we once decide that a top-2 runoff would spoil
Approval's FBC?

The requirement to get Approval from 40% of the voters, for staying in the
election, seems harmless in that regard.

But does it make it any easier for voters? They must still deal with all of
the candidates in the 1st balloting.

Another possible benefit could be the 1sf balloting could provide
information to inform strategy in the 2nd balloting.

Brams &/or Fishburn suggested a preliminary balloting that wouldn't
eliminate anyone. It's only purpose would be to provide that strategic
information.

But, for that, wouldn't a vote-for-favorite balloting be better?

Maybe a balloting to indicate favorite. If anyone gets majority, S/he wins.
If more than 1 get majority, only they go into the next balloting.

If no majorities, then you could require 40% for the next balloting.

If no 40%, then just elect the highest vote getter.

Anyway, might not that  informational balloting be intentionally
deceptively voted by some?

Maybe adding complications to Approval would only complicate voting.

It would certainly lose Approval's obvious simplicity. From just one
obvious way to count, it would change to one of many possible modifications.

Michael Ossipoff

On Oct 2, 2016 10:31 PM, "Rob Lanphier" robla@robla.net wrote:

wrote:

Approval), or Bucklin, etc., or something fancier.

there were other advantages to a 1-stage election.

of the mathematical advantages of a single stage election.  However, now
that we all have witnessed the IRV dynamic in more and more settings (e.g.
Burlington 2009), I think I've become convinced that a 2-stage process has
big advantages for making good decisions in high stakes elections.  It's
good to have a first stage for democratic vetting to weed out dangerously
flawed candidates (and to make it possible for the non-crackpots to rise
above the noise created by the crackpots).

(not overwhelming the user with too many choices).  As advocates, we have
touted the benefits of more choices, but that hasn't helped us.  The
problem:

2 other viable choices, and 5 extremists

and occupy "the center"

"extremist" candidates

a viable choice and an extremist

some of the problems with IRV in practice aren't with the tally method.
I'm living in San Francisco now, and I'm finding that decision fatigue in
elections is also a problem.  An incumbent has an enormous advantage in
the general election with a ranked choice ballot, because the opposition
has a difficult time consolidating around a viable alternative.[2]

http://qz.com/429929/the-next-design-trend-is-one-that-eliminates-all-choices/

http://www.sfgate.com/bayarea/article/S-F-Mayor-Ed-Lee-may-have-no-challengers-in-5940768.php

You spoke of the need to weed out the non serious or non winnable
candidates, but if we all just approve those we consider the best, those we
really like, without regard to winnability, then is there a need to use a
separate balloting to find out who's winnable?

Michael Ossipoff

On Oct 2, 2016 10:31 PM, "Rob Lanphier" robla@robla.net wrote:

wrote:

Approval), or Bucklin, etc., or something fancier.

there were other advantages to a 1-stage election.

of the mathematical advantages of a single stage election.  However, now
that we all have witnessed the IRV dynamic in more and more settings (e.g.
Burlington 2009), I think I've become convinced that a 2-stage process has
big advantages for making good decisions in high stakes elections.  It's
good to have a first stage for democratic vetting to weed out dangerously
flawed candidates (and to make it possible for the non-crackpots to rise
above the noise created by the crackpots).

(not overwhelming the user with too many choices).  As advocates, we have
touted the benefits of more choices, but that hasn't helped us.  The
problem:

2 other viable choices, and 5 extremists

and occupy "the center"

"extremist" candidates

a viable choice and an extremist

some of the problems with IRV in practice aren't with the tally method.
I'm living in San Francisco now, and I'm finding that decision fatigue in
elections is also a problem.  An incumbent has an enormous advantage in
the general election with a ranked choice ballot, because the opposition
has a difficult time consolidating around a viable alternative.[2]

http://qz.com/429929/the-next-design-trend-is-one-that-eliminates-all-choices/

http://www.sfgate.com/bayarea/article/S-F-Mayor-Ed-Lee-may-have-no-challengers-in-5940768.php

On 10/3/2016 12:19 AM, Juho Laatu wrote:

If we used better ballots and better counting methods then money would
have far less influence.

Currently money is used to take advantage of the limitations of
single-mark ballots (& plurality counting).  I describe the
vote-splitting and vote-concentration tactics in this recent humorous
article:

https://democracychronicles.com/wild-election/

Better election methods do not have those vulnerabilities.  So, those
currently-very-effective money-based tactics will not be useful when we
stop using single-mark ballots.

As Sanders and Trump have demonstrated, the cost of media is not a big
barrier when the candidate is either reform-minded or able to pretend to
be reform-minded.

Richard Fobes

On 10/3/2016 12:19 AM, Juho Laatu wrote:

Correction:

If the 1st balloting were one in which people indicate their favorite, then
if course we wouldn't want a 40% requirement for getting into the 2nd
balloting.

Offen the CWs won't be the favorite of 40%, as was the case in the
Burlington debacle.

Michael Ossipoff
On Oct 2, 2016 10:31 PM, "Rob Lanphier" robla@robla.net wrote: