27/08/15 16:13, Juho Laatu juho.laatu@gmail.com:

wondering whether there were specific election algorithms and ballot
designs that would allow a voter to express preferences between specific
candidates, without having to specify their preference between the
expressions of preferences themselves.

preference C>D, but could care less whether E>F=G is preferred over the
others

this happening with loops allowed!), but it is an incredibly common
thing in real life: people prefer burgers over pizza and prefer coke
over sprite (YMMV!), but when you ask them wether this mean that they
prefer burgers over coke or pizza over sprite, they'll shrug and say
these are not comparable: (burgers>pizza)=(coke>sprite).

other (ie. one-issue candidates or mutually-complementary ideologies),
and forcing voters to rank (or score, in a cardinal system) incomparable
candidates or ideologies seems to me like a lot of information is lost.

have a matrix comparing every group-candidate-ranking combination to
every other group-candidate-ranking combination. Or something in that
style; or not.

Yes, this is what I was thinking could be done to tally the results:
consider these groupings (eg. "A>B") as "monolithic" in the calculation
(thus behaving like a unique candidate) and then you'd end up, in an
elaborate example, with something like this:

1. A>B
2. B>C
3. C>A
4. B
5. A>C
6. D>A
7. C>D
8. C
9. B>A
10. D

On first sight, this would imply "A>B" to be the winner of the election.
I'm sure there would be some advanced electoral mathematical algorithm
available to check this, but does this also imply that, "A" should
be the generalised winner, because obviously "A>B" implies that more
people prefer A over B.

It's true that maybe in politics such a system isn't really needed, this
would be more useful for highly specific, technical decisions.

I hadn't even considered such a use-case; it does seem like a good idea.
Does anyone have examples of (real-life?) elections where this is
allowed, and how then do they calculate the votes?

I'm not sure if I get your point here. Isn't a candidate grouping not
the same as being allowed to equally rank candidates?

Kind greetings,
Sebastiaan

This example is a perfect demonstration of what I tried to describe a decade or so ago.There is no way to make a linear ordering of pairwise preferences if the voter uses different criteria depending upon what the pair is. I'd like my contribution to the pairwise matrix be based upon a ballot that gave "A or B, neither?" for every combination of choices.

 On Saturday, August 29, 2015 1:43 PM, Sebastiaan Snoeckx <ikke@sebastiaansnoeckx.be> wrote:

27/08/15 16:13, Juho Laatu juho.laatu@gmail.com:

  wondering whether there were specific election algorithms and ballot
designs that would allow a voter to express preferences between specific
candidates, without having to specify their preference between the
expressions of preferences themselves.

preference C>D, but could care less whether E>F=G is preferred over the
others

this happening with loops allowed!), but it is an incredibly common
thing in real life: people prefer burgers over pizza and prefer coke
over sprite (YMMV!), but when you ask them wether this mean that they
prefer burgers over coke or pizza over sprite, they'll shrug and say
these are not comparable: (burgers>pizza)=(coke>sprite).

other (ie. one-issue candidates or mutually-complementary ideologies),
and forcing voters to rank (or score, in a cardinal system) incomparable
candidates or ideologies seems to me like a lot of information is lost.

have a matrix comparing every group-candidate-ranking combination to
every other group-candidate-ranking combination. Or something in that
style; or not.

Yes, this is what I was thinking could be done to tally the results:
consider these groupings (eg. "A>B") as "monolithic" in the calculation
(thus behaving like a unique candidate) and then you'd end up, in an
elaborate example, with something like this:
  1. A>B
  2. B>C
  3. C>A
  4. B
  5. A>C
  6. D>A
  7. C>D
  8. C
  9. B>A
  10. D

On first sight, this would imply "A>B" to be the winner of the election.
I'm sure there would be some advanced electoral mathematical algorithm
available to check this, but does this also imply that, "A" should
be the generalised winner, because obviously "A>B" implies that more
people prefer A over B.

It's true that maybe in politics such a system isn't really needed, this
would be more useful for highly specific, technical decisions.

I hadn't even considered such a use-case; it does seem like a good idea.
Does anyone have examples of (real-life?) elections where this is
allowed, and how then do they calculate the votes?

I'm not sure if I get your point here. Isn't a candidate grouping not
the same as being allowed to equally rank candidates?

Kind greetings,
Sebastiaan

Election-Methods mailing list - see http://electorama.com/em for list info

On 8/29/15 6:17 PM, James Kislanko wrote:

while IRV ballots seem to prohibit marking two candidates equally
(except for those unmarked, who are all tied for last place preference
on that particular ballot), there's nothing in a Condorcet ranked-ballot
to stop you from marking A and B equally, whether they be first or last.

how is

A = B = last place

any different an expression from "neither"?

ranked ballot, in which tied ranking is allowed, is the most sensible
form of expressive preference voting.  score voting requires too much
"expression" from the voters and approval voting too little.

--

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."

I cannot respond to what you describe, since it has no relevance to what I said, which is there is no ranked ballot that can reproduce my pairwise preferences if I use different criteria depending upon what the pairs are.

 On Saturday, August 29, 2015 4:46 PM, robert bristow-johnson <rbj@audioimagination.com> wrote:

On 8/29/15 6:17 PM, James Kislanko wrote:

while IRV ballots seem to prohibit marking two candidates equally
(except for those unmarked, who are all tied for last place preference
on that particular ballot), there's nothing in a Condorcet ranked-ballot
to stop you from marking A and B equally, whether they be first or last.

how is

  A = B = last place

    any different an expression from "neither"?

ranked ballot, in which tied ranking is allowed, is the most sensible
form of expressive preference voting.  score voting requires too much
"expression" from the voters and approval voting too little.

--

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."

Election-Methods mailing list - see http://electorama.com/em for list info

On 29 Aug 2015, at 22:45, Sebastiaan Snoeckx ikke@sebastiaansnoeckx.be wrote:

I'm not aware of any real life examples. Maybe there are none.

The most problematic part in vote counting might be the ballot format. I assumed that there are lots of candidates.

If one lists all of them on the ballot paper, and also all the possible groupings, the ballot paper will be very long. It would be enough to have a limited number of columns. The voter would put a mark on some squares of the matrix (at some column next to the chosen name of a group or individual). The example vote that I discussed above could be cast using only five columns.

Another approach would be to give numbers to all the candidates and groupings, The ballot paper could be rather small. In the example vote above it would be enough to write on the ballot paper e.g. numbers "117 105 100 222 256" (where 100 is the number of Party A, 200 is the number of Party B etc.).

The actual counting of the votes (sum of pairwise preferences and final result) will probably be done with computers rather than by hand.

Yes. These proposals included also some new approaches to vote counting.

The simplest approach was to organise all the candidates as a binary tree. The counting process would first decide if the winner should come from the "right" wing or from the "left" wing, and then proceed downwards towards the candidates in the leaves. This would of course limit the "expressiveness" of the election and voters somewhat, but the benefit would be that this approach would eliminate cyclic preferences and cyclic preference related strategic voting possibilities. (I tend to think that in typical real life elections this kind of strategies are not easy and efficient to use, so it is possible that no such defence mechanisms are even needed (I'm thinking about Condorcet methods here).)

Some other approaches were to allow voters or parties/groupings to name groupings within which cyclic preferences would not be counted against the voters within that grouping. I mean that even if votes / preference matrix within party/grouping C is strongly cyclic, and preferences between groupings would be weak and possibly cyclic, a candidate of party C should still be able to win. I'll skip any detailed definitions of such algorithms here.

Juho

On 8/29/15 9:24 PM, James Kislanko wrote:

i can't really decode this, James.

we all understand that with a population of voters, a Condorcet cycle
can possibly result (Rock>Scissors, Scissors>Paper, Paper>Rock) and then
you need to add something or have something more, in the tabulating
criteria, to determine a winner.  (i actually think a cycle would be
rare in real governmental elections using a ranked ballot.)  people on
this list can argue which is best.  probably most agree that Schulze is
best, but i think that if the cycle contains only 3 candidates, that
Schulze and Ranked-Pairs (margins) and MinMax all pick the same winner.
Ranked-Pairs is a helluva lot easier to explain to skeptical legislators
and others i might try to convince to re-adopt the ranked ballot (but
this time not to go with IRV), so i have been sticking with that.
(and, as rare as a cycle might be, i think it would be even more rare
for a cycle to have more than 3 candidates in it, so the difference
between a Schulze and Ranked-Pairs might seem moot.)

but just because the collective vote totals might result in a circular
preference, that doesn't mean that it's reasonable for a single voter
to.  if a voter prefers Candidate Rock over Candidate Scissors and the
same voter prefers Scissors over Paper, i cannot grok how this same
voter could possibly prefer Paper over Rock.  then a single, linear,
ranked ballot works fine in recording all of the contingency vote
preferences of that voter.  the rest of the problem is taking all of
this collection of ranked ballots, and with the principle of "One Person
One Vote" determining what the collective preference of candidates is
and identifying the winner.  outside of a cycle, i think that Condorcet
works pretty well because, as long as there is a Condorcet Winner,
that choice would prevail in any hypothetical one-on-one election and
not the reverse.

people here since 2009 might remember when i joined the list and that i
lived in a municipality with a strong 3rd party (the Progressive Party
of Vermont), we had IRV for our mayoral election and, while a majority
of voters marked their ballots that they preferred Candidate A (for
Andy) over Candidate B (for Bob), nonetheless Candidate B was elected.
a year later IRV was repealed, although a wide majority of voters didn't
understand exactly what went wrong (and would dispute the problems),
enough folks knew something was wrong and IRV and it was repealed and
unfortunately along with it, the ranked-order ballot.

so, please elaborate on exactly what you mean.

r b-j

--

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."

Y'all are still jumping into areas I didn't mention.

Succinctly, all I'm saying is that you cannot derive my personal pairwise preferences from a ranked ballot. I'm not talking about the even more complicated question of counting votes.
Given more than 2 "issues" I might rank A>B>C>D>E given 5 choices, but asked whether I prefer C or D might choose D based upon the only subset of the issues where they disagree. You can build up a pairwise matrix from my pairwise choices, but you cannot infer my pairwise preferences from something that isn't pairwise.
The list (and literature) is full of examples of why my ranked ballot may not match my pairwise preferences: maybe I put C>D because that would help A. But if A, B, C and E were NOT running I'd prefer D.

A decade or so ago I tried to point this out in the context of someone describing a "condorcet election" as being the same as a "round robin" tournament. It would be if the voters got to choose between each pair of alternatives (THAT would be a "round robin") but it is logically incorrect to infer from an ordered list that the voter's list would always be the same for all subsets of that list.

Personally, I'd like a voting method that gave me pairwise choices, like my opthalmologist's "better or worse?" tests when trying to figure out what eyeglass prescription to write. Just record my answers in a matrix and sum all voters' matrixes. Then you can count the result however you want but I'll know that my pairwise choices weren't made by you.

 On Sunday, August 30, 2015 3:07 PM, robert bristow-johnson <rbj@audioimagination.com> wrote:

On 8/29/15 9:24 PM, James Kislanko wrote:

i can't really decode this, James.

we all understand that with a population of voters, a Condorcet cycle
can possibly result (Rock>Scissors, Scissors>Paper, Paper>Rock) and then
you need to add something or have something more, in the tabulating
criteria, to determine a winner.  (i actually think a cycle would be
rare in real governmental elections using a ranked ballot.)  people on
this list can argue which is best.  probably most agree that Schulze is
best, but i think that if the cycle contains only 3 candidates, that
Schulze and Ranked-Pairs (margins) and MinMax all pick the same winner. 
Ranked-Pairs is a helluva lot easier to explain to skeptical legislators
and others i might try to convince to re-adopt the ranked ballot (but
this time not to go with IRV), so i have been sticking with that. 
(and, as rare as a cycle might be, i think it would be even more rare
for a cycle to have more than 3 candidates in it, so the difference
between a Schulze and Ranked-Pairs might seem moot.)

but just because the collective vote totals might result in a circular
preference, that doesn't mean that it's reasonable for a single voter
to.  if a voter prefers Candidate Rock over Candidate Scissors and the
same voter prefers Scissors over Paper, i cannot grok how this same
voter could possibly prefer Paper over Rock.  then a single, linear,
ranked ballot works fine in recording all of the contingency vote
preferences of that voter.  the rest of the problem is taking all of
this collection of ranked ballots, and with the principle of "One Person
One Vote" determining what the collective preference of candidates is
and identifying the winner.  outside of a cycle, i think that Condorcet
works pretty well because, as long as there is a Condorcet Winner,
that choice would prevail in any hypothetical one-on-one election and
not the reverse.

people here since 2009 might remember when i joined the list and that i
lived in a municipality with a strong 3rd party (the Progressive Party
of Vermont), we had IRV for our mayoral election and, while a majority
of voters marked their ballots that they preferred Candidate A (for
Andy) over Candidate B (for Bob), nonetheless Candidate B was elected. 
a year later IRV was repealed, although a wide majority of voters didn't
understand exactly what went wrong (and would dispute the problems),
enough folks knew something was wrong and IRV and it was repealed and
unfortunately along with it, the ranked-order ballot.

so, please elaborate on exactly what you mean.

r b-j

--

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."

Election-Methods mailing list - see http://electorama.com/em for list info

On 8/30/15 9:30 PM, James Kislanko wrote:

that may be true.

you may be saying that, but it isn't true.  unless your personal
pairwise preferences are circular.  and that i don't buy.

for a single voter who is not schizophrenic, if the voter prefers A over
B and prefers B over C, that voter also prefers A over C.

understood.  we'll fry them fish when it comes to that.

--

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."

CIt makes sen

Correction.
It makes sense to sometimes read what you have written yourself. When I wrote "preferences would not be counted against the voters within that grouping" I meant "preferences would not be counted against the candidates within that grouping". Also, it would have been clearer not to write "If one lists all of them on the ballot paper" but "If one lists all of them on the preprinted ballot paper".

Juho