RT
Richard, the VoteFair guy
Wed, Jul 28, 2021 3:43 AM
On 7/25/2021 2:00 PM, Richard Lung wrote:
... eliminating candidates, during the count, loses voting
information, before the count is over. ...
My favorite way to count ranked-choice ballots for a single-winner
election is the Condorcet-Kemeny method.
Not only does it not eliminate candidates one at a time, it also does
not identify the first-place winner as a first step, and the
second-place winner as a second step, etc. Instead it isn't finished
until the entire sequence from most popular, second-most popular, and so
on down to least popular has been determined.
I repeat, in case you missed it, science demands one truth (to aspire
to) not two. Therefore an election count and an exclusion/elimination
count must be symmetrical. Call it symmetrical count requirement. But
that is a binomial count.
The Condorcet-Kemeny method is symmetrical.
Whereas, FAB STV is the whole caboodle perhaps relevant to data
mining. This system is monotonic, not vulnerable to strategic shuffling
the preference orders. It avoids premature exclusion, and indeed
premature election! It entirely avoids later harm, not just for transfer
of surplus preferences.
It meets the Laplace condition of weighting a whole range of
preferences in order of importance, unlike Condorcet pairing, whether or
not the pairs are weighted in relative importance.
FAB STV recognises elections as statistical estimates of
representation, and employs up to four averages to maximise accuracy.
I have read your earlier posts but I don't recall seeing a description
of your FAB STV method.
I looked at Electowiki but it's not there (under that name).
I'd be happy to look at FAB STV if you can point me to a definition of
the method.
It accepts the "Impossibility" of a deterministic election result,
and moves on. I beseech you all to do the same!
Remember that the RCIPE method (version 1 or 2) is a stepping stone that
allows reaching better methods such as the Condorcet-Kemeny method, and
beyond to PR methods.
Regarding the beyond part, my VoteFair Ranking system includes a
two-seat kind of "STV," and that can be extended to a higher number of
seats. That's why I'm curious to read a definition of your FAB STV
counting method.
To put things into perspective, consider a metaphor. IRV is like a
tricycle for tots who aren't yet ready for a bicycle. The Condorcet
methods are like a bicycle, very useful in many situations, but they
can't do everything. RCIPE is like training wheels for the bicycle. When
enough voters learn how to vote using ranked-choice ballots and RCIPE
counting then they will be ready to move on to Condorcet methods and beyond.
Richard Lung, thank you for your thoughts.
Richard Fobes
The VoteFair guy
On 7/25/2021 2:00 PM, Richard Lung wrote:
A few comments from Richard Lung (not the VoteFair guy, (who is not to be confused, if I remember rightly, with Santucci, the vote guy!).
As pointed out to Susan Simmons, which she acknowledged, eliminating candidates, during the count, loses voting information, before the count is over. It is not necessary with a binomial count, unlike all existing methods (uninomial counts) which employ elimination as an afterthought to an essentially uninomial election count.
I repeat, in case you missed it, science demands one truth (to aspire to) not two. Therefore an election count and an exclusion/elimination count must be symmetrical. Call it symmetrical count requirement. But that is a binomial count.
And a binomial count indeed does imply higher order counts, governed by the binomial theorem. But a simple coherent first order binomial count should be sufficient for democratic representation.
Whereas, FAB STV is the whole caboodle perhaps relevant to data mining. This system is monotonic, not vulnerable to strategic shuffling the preference orders. It avoids premature exclusion, and indeed premature election! It entirely avoids later harm, not just for transfer of surplus preferences.
It meets the Laplace condition of weighting a whole range of preferences in order of importance, unlike Condorcet pairing, whether or not the pairs are weighted in relative importance.
FAB STV recognises elections as statistical estimates of representation, and employs up to four averages to maximise accuracy. It accepts the "Impossibility" of a deterministic election result, and moves on. I beseech you all to do the same!
Yours sincerely,
Richard Lung.
On 25 Jul 2021, at 4:30 pm, Richard, the VoteFair guy electionmethods@votefair.org wrote:
On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
On 7/22/21 5:31 PM, VoteFair wrote:
On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
How about this?
- Eliminate the candidate with the least number of winning subgroups.
- If there is a tie, break that tie by IRV.
...
Isn't the first step basically Copeland's method?
No, because there's no elimination in Copeland (and it doesn't pass
LIIA). It would just elect the candidate/s with the most
winning subgroups.
I see you're right, of course.
I admit your suggestion is clever because it includes Condorcet loser elimination.
Yet I'm sure lots of non-math-savvy voters will not trust that the candidate with the least number of wins is not always the least popular. I too share that lack of trust.
Keep in mind that lots of voter don't trust the idea that the winner of all the pairwise contests is always the most popular.
But then clone independence is not important after all because the
methods are ugly. I can't quite determine whether clone independence is
important or not.
It's important that the failure rate is small. But it doesn't need to be zero.
That's true. You implicitly need some kind of valuation of the different
failure rates. For instance, if you want LNHarm and LNHelp, you have to
give up either monotonicity or mutual majority. Which it's going to be
depends on what values you place on the different criteria.
I'm not concerned about monotonicity, LNHarm, LNHelp or any other On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
On 7/22/21 5:31 PM, VoteFair wrote:
On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
How about this?
- Eliminate the candidate with the least number of winning subgroups.
- If there is a tie, break that tie by IRV.
...
Isn't the first step basically Copeland's method?
No, because there's no elimination in Copeland (and it doesn't pass
LIIA). It would just elect the candidate/s with the most
winning subgroups.
I see you're right, of course.
I admit your suggestion is clever because it includes Condorcet loser elimination.
Yet I'm sure lots of non-math-savvy voters will not trust that the candidate with the least number of wins is not always the least popular. I too share that lack of trust.
Keep in mind that lots of voter don't trust the idea that the winner of all the pairwise contests is always the most popular.
But then clone independence is not important after all because the
methods are ugly. I can't quite determine whether clone independence is
important or not.
It's important that the failure rate is small. But it doesn't need to be zero.
That's true. You implicitly need some kind of valuation of the different
failure rates. For instance, if you want LNHarm and LNHelp, you have to
give up either monotonicity or mutual majority. Which it's going to be
depends on what values you place on the different criteria.
I'm not concerned about monotonicity, LNHarm, LNHelp or any other failures that are difficult to exploit. I'm much more concerned about exploitable failures.
Admittedly, as a fan of Condorcet-Kemeny, I favor looking deep into the ballots, and I favor ways of "sorting" that basically move the biggest pairwise counts into one half of the usual matrix while moving the smallest pairwise counts into the other half, where the dividing line is the diagonal where candidates are paired with themselves.
Or to put it differently: if the method insists on a zero failure rate
for Condorcet loser, why shouldn't it insist on a zero failure rate for
Condorcet winner, say? And, equivalently, if "merely a low rate of
failure" is good enough for the Condorcet criterion (or say, clone
independence), why is it not good enough for Condorcet loser?
I admit I'm intentionally avoiding a zero failure rate for Condorcet winner because that makes the method into a Condorcet method, and those have been vilified (portrayed as evil) by the FairVote organization, and to some extent by STAR fans.
Plus, just as a voter is not likely to trust that the candidate with the fewest wins is least popular, they aren't likely to trust that the candidate who wins all the pairwise matches is most popular.
So at this point I'm still happy with eliminating the Condorcet loser as the top priority and otherwise eliminating the candidate who has the smallest pairwise support count (which basically counts how many remaining candidates are ranked below the candidate being scored).
At this point I continue to be open to suggestions for something better, but that window of time is closing very soon.
Again, thank you Kristofer for your wise feedback!
Richard Fobes
The VoteFair guy
On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
On 7/22/21 5:31 PM, VoteFair wrote:
On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
How about this?
- Eliminate the candidate with the least number of winning subgroups.
- If there is a tie, break that tie by IRV.
...
Isn't the first step basically Copeland's method?
No, because there's no elimination in Copeland (and it doesn't pass
LIIA). It would just elect the candidate/s with the most winning subgroups.
That's an ugly "method" that fails to look beneath the surface.
IRV also fails to look beneath the surface, which is why it too is an
"ugly" method.
That leads me to wonder which is the case.
You said you couldn't replace the IRV tiebreaker with minmax elimination
because IRV is cloneproof and minmax is not -- that clone independence
was important because it "protects against money-based vote splitting
tactics". So I found something that invokes IRV's clone independence
more often.
But then clone independence is not important after all because the
methods are ugly. I can't quite determine whether clone independence is
important or not.
But again, the ungrouped mechanic is not cloneproof.
Being cloneproof is not a goal. The goal is to have a very small
failure rate for clone independence.
Then you could check the alternatives by that metric. A method seeming
ugly may not necessarily have any bearing on the rates of failure.
Also, electing the Condorcet winner is not a goal. The goal is to have
a very small Condorcet criteria failure rate.
To repeat my concern, attempting to get a zero failure rate will cause
other kinds of failure rates to increase.
That's true. You implicitly need some kind of valuation of the different
failure rates. For instance, if you want LNHarm and LNHelp, you have to
give up either monotonicity or mutual majority. Which it's going to be
depends on what values you place on the different criteria.
The same would hold for rates. Say you want to find the method that
minimizes w * x, where x is the rates of each failure type
(monotonicity, vote splitting, teaming, crowding, favorite betrayal...).
Then the weights of the w vector provide a measure of indifference: how
much of failure type 1 is an acceptable trade for one unit of failure
type 2?
Or to put it differently: if the method insists on a zero failure rate
for Condorcet loser, why shouldn't it insist on a zero failure rate for
Condorcet winner, say? And, equivalently, if "merely a low rate of
failure" is good enough for the Condorcet criterion (or say, clone
independence), why is it not good enough for Condorcet loser?
I'm still willing to consider improvements, but it needs to find a
balance between what voters can understand -- both through an animated
video and through words -- and what yields low failure rates.
Again, thank you Kristofer for applying your clear understanding to
this revision from RCIPE 1 to RCIPE 2.
On 7/25/2021 2:00 PM, Richard Lung wrote:
> ... eliminating candidates, during the count, loses voting
information, before the count is over. ...
My favorite way to count ranked-choice ballots for a single-winner
election is the Condorcet-Kemeny method.
Not only does it not eliminate candidates one at a time, it also does
not identify the first-place winner as a first step, and the
second-place winner as a second step, etc. Instead it isn't finished
until the entire sequence from most popular, second-most popular, and so
on down to least popular has been determined.
> I repeat, in case you missed it, science demands one truth (to aspire
to) not two. Therefore an election count and an exclusion/elimination
count must be symmetrical. Call it symmetrical count requirement. But
that is a binomial count.
The Condorcet-Kemeny method is symmetrical.
> Whereas, FAB STV is the whole caboodle perhaps relevant to data
mining. This system is monotonic, not vulnerable to strategic shuffling
the preference orders. It avoids premature exclusion, and indeed
premature election! It entirely avoids later harm, not just for transfer
of surplus preferences.
> It meets the Laplace condition of weighting a whole range of
preferences in order of importance, unlike Condorcet pairing, whether or
not the pairs are weighted in relative importance.
> FAB STV recognises elections as statistical estimates of
representation, and employs up to four averages to maximise accuracy.
I have read your earlier posts but I don't recall seeing a description
of your FAB STV method.
I looked at Electowiki but it's not there (under that name).
I'd be happy to look at FAB STV if you can point me to a definition of
the method.
> It accepts the "Impossibility" of a deterministic election result,
and moves on. I beseech you all to do the same!
Remember that the RCIPE method (version 1 or 2) is a stepping stone that
allows reaching better methods such as the Condorcet-Kemeny method, and
beyond to PR methods.
Regarding the beyond part, my VoteFair Ranking system includes a
two-seat kind of "STV," and that can be extended to a higher number of
seats. That's why I'm curious to read a definition of your FAB STV
counting method.
To put things into perspective, consider a metaphor. IRV is like a
tricycle for tots who aren't yet ready for a bicycle. The Condorcet
methods are like a bicycle, very useful in many situations, but they
can't do everything. RCIPE is like training wheels for the bicycle. When
enough voters learn how to vote using ranked-choice ballots and RCIPE
counting then they will be ready to move on to Condorcet methods and beyond.
Richard Lung, thank you for your thoughts.
Richard Fobes
The VoteFair guy
On 7/25/2021 2:00 PM, Richard Lung wrote:
>
>
> A few comments from Richard Lung (not the VoteFair guy, (who is not to be confused, if I remember rightly, with Santucci, the vote guy!).
>
> As pointed out to Susan Simmons, which she acknowledged, eliminating candidates, during the count, loses voting information, before the count is over. It is not necessary with a binomial count, unlike all existing methods (uninomial counts) which employ elimination as an afterthought to an essentially uninomial election count.
> I repeat, in case you missed it, science demands one truth (to aspire to) not two. Therefore an election count and an exclusion/elimination count must be symmetrical. Call it symmetrical count requirement. But that is a binomial count.
> And a binomial count indeed does imply higher order counts, governed by the binomial theorem. But a simple coherent first order binomial count should be sufficient for democratic representation.
> Whereas, FAB STV is the whole caboodle perhaps relevant to data mining. This system is monotonic, not vulnerable to strategic shuffling the preference orders. It avoids premature exclusion, and indeed premature election! It entirely avoids later harm, not just for transfer of surplus preferences.
> It meets the Laplace condition of weighting a whole range of preferences in order of importance, unlike Condorcet pairing, whether or not the pairs are weighted in relative importance.
> FAB STV recognises elections as statistical estimates of representation, and employs up to four averages to maximise accuracy. It accepts the "Impossibility" of a deterministic election result, and moves on. I beseech you all to do the same!
>
> Yours sincerely,
> Richard Lung.
>
>
>
>
> On 25 Jul 2021, at 4:30 pm, Richard, the VoteFair guy <electionmethods@votefair.org> wrote:
>
> On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
>> On 7/22/21 5:31 PM, VoteFair wrote:
>>> On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
>>> > How about this?
>>> >
>>> > - Eliminate the candidate with the least number of winning subgroups.
>>> > - If there is a tie, break that tie by IRV.
>>> > ...
>>>
>>> Isn't the first step basically Copeland's method?
>>
>> No, because there's no elimination in Copeland (and it doesn't pass
>> LIIA). It would just elect the candidate/s with the most
>> winning subgroups.
>
> I see you're right, of course.
>
> I admit your suggestion is clever because it includes Condorcet loser elimination.
>
> Yet I'm sure lots of non-math-savvy voters will not trust that the candidate with the least number of wins is not always the least popular. I too share that lack of trust.
>
> Keep in mind that lots of voter don't trust the idea that the winner of all the pairwise contests is always the most popular.
>
>> But then clone independence is not important after all because the
>> methods are ugly. I can't quite determine whether clone independence is
>> important or not.
>
> It's important that the failure rate is small. But it doesn't need to be zero.
>
>> That's true. You implicitly need some kind of valuation of the different
>> failure rates. For instance, if you want LNHarm and LNHelp, you have to
>> give up either monotonicity or mutual majority. Which it's going to be
>> depends on what values you place on the different criteria.
>
> I'm not concerned about monotonicity, LNHarm, LNHelp or any other On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
>> On 7/22/21 5:31 PM, VoteFair wrote:
>>> On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
>>> > How about this?
>>> >
>>> > - Eliminate the candidate with the least number of winning subgroups.
>>> > - If there is a tie, break that tie by IRV.
>>> > ...
>>>
>>> Isn't the first step basically Copeland's method?
>>
>> No, because there's no elimination in Copeland (and it doesn't pass
>> LIIA). It would just elect the candidate/s with the most
>> winning subgroups.
>
> I see you're right, of course.
>
> I admit your suggestion is clever because it includes Condorcet loser elimination.
>
> Yet I'm sure lots of non-math-savvy voters will not trust that the candidate with the least number of wins is not always the least popular. I too share that lack of trust.
>
> Keep in mind that lots of voter don't trust the idea that the winner of all the pairwise contests is always the most popular.
>
>> But then clone independence is not important after all because the
>> methods are ugly. I can't quite determine whether clone independence is
>> important or not.
>
> It's important that the failure rate is small. But it doesn't need to be zero.
>
>> That's true. You implicitly need some kind of valuation of the different
>> failure rates. For instance, if you want LNHarm and LNHelp, you have to
>> give up either monotonicity or mutual majority. Which it's going to be
>> depends on what values you place on the different criteria.
>
> I'm not concerned about monotonicity, LNHarm, LNHelp or any other failures that are difficult to exploit. I'm much more concerned about exploitable failures.
>
> Admittedly, as a fan of Condorcet-Kemeny, I favor looking deep into the ballots, and I favor ways of "sorting" that basically move the biggest pairwise counts into one half of the usual matrix while moving the smallest pairwise counts into the other half, where the dividing line is the diagonal where candidates are paired with themselves.
>
>> Or to put it differently: if the method insists on a zero failure rate
>> for Condorcet loser, why shouldn't it insist on a zero failure rate for
>> Condorcet winner, say? And, equivalently, if "merely a low rate of
>> failure" is good enough for the Condorcet criterion (or say, clone
>> independence), why is it not good enough for Condorcet loser?
>
> I admit I'm intentionally avoiding a zero failure rate for Condorcet winner because that makes the method into a Condorcet method, and those have been vilified (portrayed as evil) by the FairVote organization, and to some extent by STAR fans.
>
> Plus, just as a voter is not likely to trust that the candidate with the fewest wins is least popular, they aren't likely to trust that the candidate who wins all the pairwise matches is most popular.
>
> So at this point I'm still happy with eliminating the Condorcet loser as the top priority and otherwise eliminating the candidate who has the smallest pairwise support count (which basically counts how many remaining candidates are ranked below the candidate being scored).
>
> At this point I continue to be open to suggestions for something better, but that window of time is closing very soon.
>
> Again, thank you Kristofer for your wise feedback!
>
> Richard Fobes
> The VoteFair guy
>
>
>> On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
>>> On 7/22/21 5:31 PM, VoteFair wrote:
>>> On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
>>>> How about this?
>>>>
>>>> - Eliminate the candidate with the least number of winning subgroups.
>>>> - If there is a tie, break that tie by IRV.
>>>> ...
>>>
>>> Isn't the first step basically Copeland's method?
>>
>> No, because there's no elimination in Copeland (and it doesn't pass
>> LIIA). It would just elect the candidate/s with the most winning subgroups.
>>
>>> That's an ugly "method" that fails to look beneath the surface.
>>>
>>> IRV also fails to look beneath the surface, which is why it too is an
>>> "ugly" method.
>>
>> That leads me to wonder which is the case.
>>
>> You said you couldn't replace the IRV tiebreaker with minmax elimination
>> because IRV is cloneproof and minmax is not -- that clone independence
>> was important because it "protects against money-based vote splitting
>> tactics". So I found something that invokes IRV's clone independence
>> more often.
>>
>> But then clone independence is not important after all because the
>> methods are ugly. I can't quite determine whether clone independence is
>> important or not.
>>
>>>> But again, the ungrouped mechanic is not cloneproof.
>>>
>>> Being cloneproof is not a goal. The goal is to have a very small
>>> failure rate for clone independence.
>>
>> Then you could check the alternatives by that metric. A method seeming
>> ugly may not necessarily have any bearing on the rates of failure.
>>
>>> Also, electing the Condorcet winner is not a goal. The goal is to have
>>> a very small Condorcet criteria failure rate.
>>>
>>> To repeat my concern, attempting to get a zero failure rate will cause
>>> other kinds of failure rates to increase.
>>
>> That's true. You implicitly need some kind of valuation of the different
>> failure rates. For instance, if you want LNHarm and LNHelp, you have to
>> give up either monotonicity or mutual majority. Which it's going to be
>> depends on what values you place on the different criteria.
>>
>> The same would hold for rates. Say you want to find the method that
>> minimizes w * x, where x is the rates of each failure type
>> (monotonicity, vote splitting, teaming, crowding, favorite betrayal...).
>> Then the weights of the w vector provide a measure of indifference: how
>> much of failure type 1 is an acceptable trade for one unit of failure
>> type 2?
>>
>> Or to put it differently: if the method insists on a zero failure rate
>> for Condorcet loser, why shouldn't it insist on a zero failure rate for
>> Condorcet winner, say? And, equivalently, if "merely a low rate of
>> failure" is good enough for the Condorcet criterion (or say, clone
>> independence), why is it not good enough for Condorcet loser?
>>
>>> I'm still willing to consider improvements, but it needs to find a
>>> balance between what voters can understand -- both through an animated
>>> video and through words -- and what yields low failure rates.
>>>
>>> Again, thank you Kristofer for applying your clear understanding to
>>> this revision from RCIPE 1 to RCIPE 2.
>>
>> You're welcome :-)
>>
>> -km
> ----
> Election-Methods mailing list - see https://electorama.com/em for list info
>
KM
Kristofer Munsterhjelm
Wed, Jul 28, 2021 8:08 PM
On 27.07.2021 00:16, Richard, the VoteFair guy wrote:
On 7/26/2021 6:20 AM, Kristofer Munsterhjelm wrote:
...
... Is that right? ...
Your calculations are correct.
Then that's just Borda. Congrats, you've reinvented Borda-elimination :-)
Well, it's not quite Borda-elimination since the early elimination of a
Condorcet loser may reshuffle the order of later eliminations, and due
to the same effect that causes nonmonotonicity in elimination methods,
lead someone else to win.
But ordinary Borda-elimination does eventually remove every Condorcet
loser, so it should have little effect on your criterion compliances.
In particular, although if I were selfish I shouldn't point this out,
Borda-elimination passes Condorcet and the Condorcet loser elimination
stage doesn't change that. Since your simulator says that it doesn't
pass Condorcet, I would view its other results with some suspicion,
particularly given how it's also been wrong about clone independence.
But, since your method is pretty close to Borda elimination, and Borda
elimination also eliminates Condorcet losers as a matter of course[1],
you could simplify it (if you want to retain its Borda-elimination
nature) into this:
- Determine the candidate with the fewest arrows, and eliminate that
candidate. Break ties by IRV (and further ties by Ext-Minmax).
- Repeat from 1 until a single candidate remains. That candidate is the
winner.
Even if you don't want to, you could run tests comparing this simplified
version to the one with Condorcet loser elimination - though as
mentioned, I wouldn't be too confident of the results.
As for Condorcet winner, what I was objecting to (besides me being a
Condorcetist and preferring Condorcet) is that there doesn't seem to be
a principled approach to your methods' violation of CW. You used a
bicycle metaphor: but the attempt to deliberately avoid Condorcet winner
seems like removing the chain (and subsequently creating a
penny-farthing) just so that you can say that it's not a bicycle,
because the pedestrian organization (FairVote) dislikes bicycles.
If you're deliberately setting out to not make a bicycle because
bicycles aren't liked, you should create something that's consistently
not a bicycle: that gets some kind of return for not being a bicycle,
and then has more of a justification of what a good winner is than "this
isn't Condorcet, but you can see Condorcet from here".
Consider something like a method that explicitly eliminates the
Condorcet loser, and where the base method otherwise does not pass
Condorcet. Then the elimination method will pass Condorcet when there's
an unambiguous order of losers (X is the loser, Y beats only X, Z beats
only Y and X, etc.). The iterated Condorcet losers provide a
straightforward order of candidates from most relevant to least
relevant. But now introduce a cycle somewhere "downstream" of the
winner. Then by introduction of such a cycle, the Condorcet winner may
be eliminated early (depending on how the method is constructed). It's
difficult to see why that should make a difference; I'd say you'd need
some kind of explicit reason for why something that appears irrelevant
(the relation between people who would otherwise be losers) matters.
In any case, I don't think FairVote itself would support your method as
long as it's not IRV. They're first and foremost an advocacy
organization focused laser-like on IRV and STV, and as your method isn't
it, they probably won't support it. But even being generous that their
"Core Support criterion" isn't just a fig leaf (see my post to Forest),
any method that is Condorcet in the "straight line scenario" above must
by necessity fail that criterion.
Now, don't get me wrong: I don't think passing Core Support would win
them over - because again, their thing is IRV. I'm more just
re-emphasizing that a method that fails Condorcet should have some
reason for doing so. (E.g. you can get very close to Condorcet while
still passing the favorite betrayal condition - then passing the FBC
becomes the reason.)
-km
[1] For that matter, every sequential elimination method that respects
majority rule will also pass Condorcet Loser, because in the worst case
that the Condorcet loser remains until the final round, the other
candidate will by definition beat the loser in the final round.
On 27.07.2021 00:16, Richard, the VoteFair guy wrote:
> On 7/26/2021 6:20 AM, Kristofer Munsterhjelm wrote:
>> ...
>> ... Is that right? ...
>
> Your calculations are correct.
Then that's just Borda. Congrats, you've reinvented Borda-elimination :-)
Well, it's not quite Borda-elimination since the early elimination of a
Condorcet loser may reshuffle the order of later eliminations, and due
to the same effect that causes nonmonotonicity in elimination methods,
lead someone else to win.
But ordinary Borda-elimination does eventually remove every Condorcet
loser, so it should have little effect on your criterion compliances.
In particular, although if I were selfish I shouldn't point this out,
Borda-elimination passes Condorcet and the Condorcet loser elimination
stage doesn't change that. Since your simulator says that it doesn't
pass Condorcet, I would view its other results with some suspicion,
particularly given how it's also been wrong about clone independence.
But, since your method is pretty close to Borda elimination, and Borda
elimination also eliminates Condorcet losers as a matter of course[1],
you could simplify it (if you want to retain its Borda-elimination
nature) into this:
1. Determine the candidate with the fewest arrows, and eliminate that
candidate. Break ties by IRV (and further ties by Ext-Minmax).
2. Repeat from 1 until a single candidate remains. That candidate is the
winner.
Even if you don't want to, you could run tests comparing this simplified
version to the one with Condorcet loser elimination - though as
mentioned, I wouldn't be too confident of the results.
As for Condorcet winner, what I was objecting to (besides me being a
Condorcetist and preferring Condorcet) is that there doesn't seem to be
a principled approach to your methods' violation of CW. You used a
bicycle metaphor: but the attempt to deliberately avoid Condorcet winner
seems like removing the chain (and subsequently creating a
penny-farthing) just so that you can say that it's not a bicycle,
because the pedestrian organization (FairVote) dislikes bicycles.
If you're deliberately setting out to not make a bicycle because
bicycles aren't liked, you should create something that's consistently
not a bicycle: that gets some kind of return for not being a bicycle,
and then has more of a justification of what a good winner is than "this
isn't Condorcet, but you can see Condorcet from here".
Consider something like a method that explicitly eliminates the
Condorcet loser, and where the base method otherwise does not pass
Condorcet. Then the elimination method will pass Condorcet when there's
an unambiguous order of losers (X is the loser, Y beats only X, Z beats
only Y and X, etc.). The iterated Condorcet losers provide a
straightforward order of candidates from most relevant to least
relevant. But now introduce a cycle somewhere "downstream" of the
winner. Then by introduction of such a cycle, the Condorcet winner may
be eliminated early (depending on how the method is constructed). It's
difficult to see why that should make a difference; I'd say you'd need
some kind of explicit reason for why something that appears irrelevant
(the relation between people who would otherwise be losers) matters.
In any case, I don't think FairVote itself would support your method as
long as it's not IRV. They're first and foremost an advocacy
organization focused laser-like on IRV and STV, and as your method isn't
it, they probably won't support it. But even being generous that their
"Core Support criterion" isn't just a fig leaf (see my post to Forest),
any method that is Condorcet in the "straight line scenario" above must
by necessity fail that criterion.
Now, don't get me wrong: I don't think passing Core Support would win
them over - because again, their thing is IRV. I'm more just
re-emphasizing that a method that fails Condorcet should have some
reason for doing so. (E.g. you can get very close to Condorcet while
still passing the favorite betrayal condition - then passing the FBC
becomes the reason.)
-km
[1] For that matter, every sequential elimination method that respects
majority rule will also pass Condorcet Loser, because in the worst case
that the Condorcet loser remains until the final round, the other
candidate will by definition beat the loser in the final round.
RL
Richard Lung
Thu, Jul 29, 2021 5:19 PM
Dear Richard Fobes, the VoteFair guy, and all at election methods,
Personal family misfortunes prevent me from giving a proper reply.
I believe I say something about Kemeny in an appendix to my Smashwords free ebook,
FAB STV: Four Averages Binomial Single Transferable Vote.
https://www.smashwords.com/books/view/806030
The system is fully described in the second part, but there are plenty of summaries, from simple list of attributes, to summary convenient for those familiar with Meek method.
I may mention what the system looks like from the voters point of view. It could be any preference voting ballot. But it counts differently. Last preferences help as much to exclude candidates, as first preferences help to elect candidates. That is to say it is a binomial count.
It is not necessary to fill in all the preferences. Vacant preferences count towards a NOTA quota, leaving a seat unfilled.
Circumstances permitting, I hope to say more about your post and others.
Regards
Richard Lung.
On 28 Jul 2021, at 4:43 am, Richard, the VoteFair guy electionmethods@votefair.org wrote:
On 7/25/2021 2:00 PM, Richard Lung wrote:
... eliminating candidates, during the count, loses voting information, before the count is over. ...
My favorite way to count ranked-choice ballots for a single-winner election is the Condorcet-Kemeny method.
Not only does it not eliminate candidates one at a time, it also does not identify the first-place winner as a first step, and the second-place winner as a second step, etc. Instead it isn't finished until the entire sequence from most popular, second-most popular, and so on down to least popular has been determined.
I repeat, in case you missed it, science demands one truth (to aspire to) not two. Therefore an election count and an exclusion/elimination count must be symmetrical. Call it symmetrical count requirement. But that is a binomial count.
The Condorcet-Kemeny method is symmetrical.
Whereas, FAB STV is the whole caboodle perhaps relevant to data mining. This system is monotonic, not vulnerable to strategic shuffling the preference orders. It avoids premature exclusion, and indeed premature election! It entirely avoids later harm, not just for transfer of surplus preferences.
It meets the Laplace condition of weighting a whole range of preferences in order of importance, unlike Condorcet pairing, whether or not the pairs are weighted in relative importance.
FAB STV recognises elections as statistical estimates of representation, and employs up to four averages to maximise accuracy.
I have read your earlier posts but I don't recall seeing a description of your FAB STV method.
I looked at Electowiki but it's not there (under that name).
I'd be happy to look at FAB STV if you can point me to a definition of the method.
It accepts the "Impossibility" of a deterministic election result, and moves on. I beseech you all to do the same!
Remember that the RCIPE method (version 1 or 2) is a stepping stone that allows reaching better methods such as the Condorcet-Kemeny method, and beyond to PR methods.
Regarding the beyond part, my VoteFair Ranking system includes a two-seat kind of "STV," and that can be extended to a higher number of seats. That's why I'm curious to read a definition of your FAB STV counting method.
To put things into perspective, consider a metaphor. IRV is like a tricycle for tots who aren't yet ready for a bicycle. The Condorcet methods are like a bicycle, very useful in many situations, but they can't do everything. RCIPE is like training wheels for the bicycle. When enough voters learn how to vote using ranked-choice ballots and RCIPE counting then they will be ready to move on to Condorcet methods and beyond.
Richard Lung, thank you for your thoughts.
Richard Fobes
The VoteFair guy
On 7/25/2021 2:00 PM, Richard Lung wrote:
A few comments from Richard Lung (not the VoteFair guy, (who is not to be confused, if I remember rightly, with Santucci, the vote guy!).
As pointed out to Susan Simmons, which she acknowledged, eliminating candidates, during the count, loses voting information, before the count is over. It is not necessary with a binomial count, unlike all existing methods (uninomial counts) which employ elimination as an afterthought to an essentially uninomial election count.
I repeat, in case you missed it, science demands one truth (to aspire to) not two. Therefore an election count and an exclusion/elimination count must be symmetrical. Call it symmetrical count requirement. But that is a binomial count.
And a binomial count indeed does imply higher order counts, governed by the binomial theorem. But a simple coherent first order binomial count should be sufficient for democratic representation.
Whereas, FAB STV is the whole caboodle perhaps relevant to data mining. This system is monotonic, not vulnerable to strategic shuffling the preference orders. It avoids premature exclusion, and indeed premature election! It entirely avoids later harm, not just for transfer of surplus preferences.
It meets the Laplace condition of weighting a whole range of preferences in order of importance, unlike Condorcet pairing, whether or not the pairs are weighted in relative importance.
FAB STV recognises elections as statistical estimates of representation, and employs up to four averages to maximise accuracy. It accepts the "Impossibility" of a deterministic election result, and moves on. I beseech you all to do the same!
Yours sincerely,
Richard Lung.
On 25 Jul 2021, at 4:30 pm, Richard, the VoteFair guy electionmethods@votefair.org wrote:
On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
On 7/22/21 5:31 PM, VoteFair wrote:
On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
How about this?
- Eliminate the candidate with the least number of winning subgroups.
- If there is a tie, break that tie by IRV.
...
Isn't the first step basically Copeland's method?
No, because there's no elimination in Copeland (and it doesn't pass
LIIA). It would just elect the candidate/s with the most
winning subgroups.
I see you're right, of course.
I admit your suggestion is clever because it includes Condorcet loser elimination.
Yet I'm sure lots of non-math-savvy voters will not trust that the candidate with the least number of wins is not always the least popular. I too share that lack of trust.
Keep in mind that lots of voter don't trust the idea that the winner of all the pairwise contests is always the most popular.
But then clone independence is not important after all because the
methods are ugly. I can't quite determine whether clone independence is
important or not.
It's important that the failure rate is small. But it doesn't need to be zero.
That's true. You implicitly need some kind of valuation of the different
failure rates. For instance, if you want LNHarm and LNHelp, you have to
give up either monotonicity or mutual majority. Which it's going to be
depends on what values you place on the different criteria.
I'm not concerned about monotonicity, LNHarm, LNHelp or any other On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
On 7/22/21 5:31 PM, VoteFair wrote:
On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
How about this?
- Eliminate the candidate with the least number of winning subgroups.
- If there is a tie, break that tie by IRV.
...
Isn't the first step basically Copeland's method?
No, because there's no elimination in Copeland (and it doesn't pass
LIIA). It would just elect the candidate/s with the most
winning subgroups.
I see you're right, of course.
I admit your suggestion is clever because it includes Condorcet loser elimination.
Yet I'm sure lots of non-math-savvy voters will not trust that the candidate with the least number of wins is not always the least popular. I too share that lack of trust.
Keep in mind that lots of voter don't trust the idea that the winner of all the pairwise contests is always the most popular.
But then clone independence is not important after all because the
methods are ugly. I can't quite determine whether clone independence is
important or not.
It's important that the failure rate is small. But it doesn't need to be zero.
That's true. You implicitly need some kind of valuation of the different
failure rates. For instance, if you want LNHarm and LNHelp, you have to
give up either monotonicity or mutual majority. Which it's going to be
depends on what values you place on the different criteria.
I'm not concerned about monotonicity, LNHarm, LNHelp or any other failures that are difficult to exploit. I'm much more concerned about exploitable failures.
Admittedly, as a fan of Condorcet-Kemeny, I favor looking deep into the ballots, and I favor ways of "sorting" that basically move the biggest pairwise counts into one half of the usual matrix while moving the smallest pairwise counts into the other half, where the dividing line is the diagonal where candidates are paired with themselves.
Or to put it differently: if the method insists on a zero failure rate
for Condorcet loser, why shouldn't it insist on a zero failure rate for
Condorcet winner, say? And, equivalently, if "merely a low rate of
failure" is good enough for the Condorcet criterion (or say, clone
independence), why is it not good enough for Condorcet loser?
I admit I'm intentionally avoiding a zero failure rate for Condorcet winner because that makes the method into a Condorcet method, and those have been vilified (portrayed as evil) by the FairVote organization, and to some extent by STAR fans.
Plus, just as a voter is not likely to trust that the candidate with the fewest wins is least popular, they aren't likely to trust that the candidate who wins all the pairwise matches is most popular.
So at this point I'm still happy with eliminating the Condorcet loser as the top priority and otherwise eliminating the candidate who has the smallest pairwise support count (which basically counts how many remaining candidates are ranked below the candidate being scored).
At this point I continue to be open to suggestions for something better, but that window of time is closing very soon.
Again, thank you Kristofer for your wise feedback!
Richard Fobes
The VoteFair guy
On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
On 7/22/21 5:31 PM, VoteFair wrote:
On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
How about this?
- Eliminate the candidate with the least number of winning subgroups.
- If there is a tie, break that tie by IRV.
...
Isn't the first step basically Copeland's method?
No, because there's no elimination in Copeland (and it doesn't pass
LIIA). It would just elect the candidate/s with the most winning subgroups.
That's an ugly "method" that fails to look beneath the surface.
IRV also fails to look beneath the surface, which is why it too is an
"ugly" method.
That leads me to wonder which is the case.
You said you couldn't replace the IRV tiebreaker with minmax elimination
because IRV is cloneproof and minmax is not -- that clone independence
was important because it "protects against money-based vote splitting
tactics". So I found something that invokes IRV's clone independence
more often.
But then clone independence is not important after all because the
methods are ugly. I can't quite determine whether clone independence is
important or not.
But again, the ungrouped mechanic is not cloneproof.
Being cloneproof is not a goal. The goal is to have a very small
failure rate for clone independence.
Then you could check the alternatives by that metric. A method seeming
ugly may not necessarily have any bearing on the rates of failure.
Also, electing the Condorcet winner is not a goal. The goal is to have
a very small Condorcet criteria failure rate.
To repeat my concern, attempting to get a zero failure rate will cause
other kinds of failure rates to increase.
That's true. You implicitly need some kind of valuation of the different
failure rates. For instance, if you want LNHarm and LNHelp, you have to
give up either monotonicity or mutual majority. Which it's going to be
depends on what values you place on the different criteria.
The same would hold for rates. Say you want to find the method that
minimizes w * x, where x is the rates of each failure type
(monotonicity, vote splitting, teaming, crowding, favorite betrayal...).
Then the weights of the w vector provide a measure of indifference: how
much of failure type 1 is an acceptable trade for one unit of failure
type 2?
Or to put it differently: if the method insists on a zero failure rate
for Condorcet loser, why shouldn't it insist on a zero failure rate for
Condorcet winner, say? And, equivalently, if "merely a low rate of
failure" is good enough for the Condorcet criterion (or say, clone
independence), why is it not good enough for Condorcet loser?
I'm still willing to consider improvements, but it needs to find a
balance between what voters can understand -- both through an animated
video and through words -- and what yields low failure rates.
Again, thank you Kristofer for applying your clear understanding to
this revision from RCIPE 1 to RCIPE 2.
Dear Richard Fobes, the VoteFair guy, and all at election methods,
Personal family misfortunes prevent me from giving a proper reply.
I believe I say something about Kemeny in an appendix to my Smashwords free ebook,
FAB STV: Four Averages Binomial Single Transferable Vote.
https://www.smashwords.com/books/view/806030
The system is fully described in the second part, but there are plenty of summaries, from simple list of attributes, to summary convenient for those familiar with Meek method.
I may mention what the system looks like from the voters point of view. It could be any preference voting ballot. But it counts differently. Last preferences help as much to exclude candidates, as first preferences help to elect candidates. That is to say it is a binomial count.
It is not necessary to fill in all the preferences. Vacant preferences count towards a NOTA quota, leaving a seat unfilled.
Circumstances permitting, I hope to say more about your post and others.
Regards
Richard Lung.
On 28 Jul 2021, at 4:43 am, Richard, the VoteFair guy <electionmethods@votefair.org> wrote:
On 7/25/2021 2:00 PM, Richard Lung wrote:
> ... eliminating candidates, during the count, loses voting information, before the count is over. ...
My favorite way to count ranked-choice ballots for a single-winner election is the Condorcet-Kemeny method.
Not only does it not eliminate candidates one at a time, it also does not identify the first-place winner as a first step, and the second-place winner as a second step, etc. Instead it isn't finished until the entire sequence from most popular, second-most popular, and so on down to least popular has been determined.
> I repeat, in case you missed it, science demands one truth (to aspire to) not two. Therefore an election count and an exclusion/elimination count must be symmetrical. Call it symmetrical count requirement. But that is a binomial count.
The Condorcet-Kemeny method is symmetrical.
> Whereas, FAB STV is the whole caboodle perhaps relevant to data mining. This system is monotonic, not vulnerable to strategic shuffling the preference orders. It avoids premature exclusion, and indeed premature election! It entirely avoids later harm, not just for transfer of surplus preferences.
> It meets the Laplace condition of weighting a whole range of preferences in order of importance, unlike Condorcet pairing, whether or not the pairs are weighted in relative importance.
> FAB STV recognises elections as statistical estimates of representation, and employs up to four averages to maximise accuracy.
I have read your earlier posts but I don't recall seeing a description of your FAB STV method.
I looked at Electowiki but it's not there (under that name).
I'd be happy to look at FAB STV if you can point me to a definition of the method.
> It accepts the "Impossibility" of a deterministic election result, and moves on. I beseech you all to do the same!
Remember that the RCIPE method (version 1 or 2) is a stepping stone that allows reaching better methods such as the Condorcet-Kemeny method, and beyond to PR methods.
Regarding the beyond part, my VoteFair Ranking system includes a two-seat kind of "STV," and that can be extended to a higher number of seats. That's why I'm curious to read a definition of your FAB STV counting method.
To put things into perspective, consider a metaphor. IRV is like a tricycle for tots who aren't yet ready for a bicycle. The Condorcet methods are like a bicycle, very useful in many situations, but they can't do everything. RCIPE is like training wheels for the bicycle. When enough voters learn how to vote using ranked-choice ballots and RCIPE counting then they will be ready to move on to Condorcet methods and beyond.
Richard Lung, thank you for your thoughts.
Richard Fobes
The VoteFair guy
> On 7/25/2021 2:00 PM, Richard Lung wrote:
>
>
> A few comments from Richard Lung (not the VoteFair guy, (who is not to be confused, if I remember rightly, with Santucci, the vote guy!).
>
> As pointed out to Susan Simmons, which she acknowledged, eliminating candidates, during the count, loses voting information, before the count is over. It is not necessary with a binomial count, unlike all existing methods (uninomial counts) which employ elimination as an afterthought to an essentially uninomial election count.
> I repeat, in case you missed it, science demands one truth (to aspire to) not two. Therefore an election count and an exclusion/elimination count must be symmetrical. Call it symmetrical count requirement. But that is a binomial count.
> And a binomial count indeed does imply higher order counts, governed by the binomial theorem. But a simple coherent first order binomial count should be sufficient for democratic representation.
> Whereas, FAB STV is the whole caboodle perhaps relevant to data mining. This system is monotonic, not vulnerable to strategic shuffling the preference orders. It avoids premature exclusion, and indeed premature election! It entirely avoids later harm, not just for transfer of surplus preferences.
> It meets the Laplace condition of weighting a whole range of preferences in order of importance, unlike Condorcet pairing, whether or not the pairs are weighted in relative importance.
> FAB STV recognises elections as statistical estimates of representation, and employs up to four averages to maximise accuracy. It accepts the "Impossibility" of a deterministic election result, and moves on. I beseech you all to do the same!
>
> Yours sincerely,
> Richard Lung.
>
>
>
>
> On 25 Jul 2021, at 4:30 pm, Richard, the VoteFair guy <electionmethods@votefair.org> wrote:
>
> On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
>>> On 7/22/21 5:31 PM, VoteFair wrote:
>>>> On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
>>>> How about this?
>>>>
>>>> - Eliminate the candidate with the least number of winning subgroups.
>>>> - If there is a tie, break that tie by IRV.
>>>> ...
>>>
>>> Isn't the first step basically Copeland's method?
>>
>> No, because there's no elimination in Copeland (and it doesn't pass
>> LIIA). It would just elect the candidate/s with the most
>> winning subgroups.
>
> I see you're right, of course.
>
> I admit your suggestion is clever because it includes Condorcet loser elimination.
>
> Yet I'm sure lots of non-math-savvy voters will not trust that the candidate with the least number of wins is not always the least popular. I too share that lack of trust.
>
> Keep in mind that lots of voter don't trust the idea that the winner of all the pairwise contests is always the most popular.
>
>> But then clone independence is not important after all because the
>> methods are ugly. I can't quite determine whether clone independence is
>> important or not.
>
> It's important that the failure rate is small. But it doesn't need to be zero.
>
>> That's true. You implicitly need some kind of valuation of the different
>> failure rates. For instance, if you want LNHarm and LNHelp, you have to
>> give up either monotonicity or mutual majority. Which it's going to be
>> depends on what values you place on the different criteria.
>
> I'm not concerned about monotonicity, LNHarm, LNHelp or any other On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
>>> On 7/22/21 5:31 PM, VoteFair wrote:
>>>> On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
>>>> How about this?
>>>>
>>>> - Eliminate the candidate with the least number of winning subgroups.
>>>> - If there is a tie, break that tie by IRV.
>>>> ...
>>>
>>> Isn't the first step basically Copeland's method?
>>
>> No, because there's no elimination in Copeland (and it doesn't pass
>> LIIA). It would just elect the candidate/s with the most
>> winning subgroups.
>
> I see you're right, of course.
>
> I admit your suggestion is clever because it includes Condorcet loser elimination.
>
> Yet I'm sure lots of non-math-savvy voters will not trust that the candidate with the least number of wins is not always the least popular. I too share that lack of trust.
>
> Keep in mind that lots of voter don't trust the idea that the winner of all the pairwise contests is always the most popular.
>
>> But then clone independence is not important after all because the
>> methods are ugly. I can't quite determine whether clone independence is
>> important or not.
>
> It's important that the failure rate is small. But it doesn't need to be zero.
>
>> That's true. You implicitly need some kind of valuation of the different
>> failure rates. For instance, if you want LNHarm and LNHelp, you have to
>> give up either monotonicity or mutual majority. Which it's going to be
>> depends on what values you place on the different criteria.
>
> I'm not concerned about monotonicity, LNHarm, LNHelp or any other failures that are difficult to exploit. I'm much more concerned about exploitable failures.
>
> Admittedly, as a fan of Condorcet-Kemeny, I favor looking deep into the ballots, and I favor ways of "sorting" that basically move the biggest pairwise counts into one half of the usual matrix while moving the smallest pairwise counts into the other half, where the dividing line is the diagonal where candidates are paired with themselves.
>
>> Or to put it differently: if the method insists on a zero failure rate
>> for Condorcet loser, why shouldn't it insist on a zero failure rate for
>> Condorcet winner, say? And, equivalently, if "merely a low rate of
>> failure" is good enough for the Condorcet criterion (or say, clone
>> independence), why is it not good enough for Condorcet loser?
>
> I admit I'm intentionally avoiding a zero failure rate for Condorcet winner because that makes the method into a Condorcet method, and those have been vilified (portrayed as evil) by the FairVote organization, and to some extent by STAR fans.
>
> Plus, just as a voter is not likely to trust that the candidate with the fewest wins is least popular, they aren't likely to trust that the candidate who wins all the pairwise matches is most popular.
>
> So at this point I'm still happy with eliminating the Condorcet loser as the top priority and otherwise eliminating the candidate who has the smallest pairwise support count (which basically counts how many remaining candidates are ranked below the candidate being scored).
>
> At this point I continue to be open to suggestions for something better, but that window of time is closing very soon.
>
> Again, thank you Kristofer for your wise feedback!
>
> Richard Fobes
> The VoteFair guy
>
>
>>> On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
>>>> On 7/22/21 5:31 PM, VoteFair wrote:
>>>> On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
>>>> How about this?
>>>>
>>>> - Eliminate the candidate with the least number of winning subgroups.
>>>> - If there is a tie, break that tie by IRV.
>>>> ...
>>>
>>> Isn't the first step basically Copeland's method?
>>
>> No, because there's no elimination in Copeland (and it doesn't pass
>> LIIA). It would just elect the candidate/s with the most winning subgroups.
>>
>>> That's an ugly "method" that fails to look beneath the surface.
>>>
>>> IRV also fails to look beneath the surface, which is why it too is an
>>> "ugly" method.
>>
>> That leads me to wonder which is the case.
>>
>> You said you couldn't replace the IRV tiebreaker with minmax elimination
>> because IRV is cloneproof and minmax is not -- that clone independence
>> was important because it "protects against money-based vote splitting
>> tactics". So I found something that invokes IRV's clone independence
>> more often.
>>
>> But then clone independence is not important after all because the
>> methods are ugly. I can't quite determine whether clone independence is
>> important or not.
>>
>>>> But again, the ungrouped mechanic is not cloneproof.
>>>
>>> Being cloneproof is not a goal. The goal is to have a very small
>>> failure rate for clone independence.
>>
>> Then you could check the alternatives by that metric. A method seeming
>> ugly may not necessarily have any bearing on the rates of failure.
>>
>>> Also, electing the Condorcet winner is not a goal. The goal is to have
>>> a very small Condorcet criteria failure rate.
>>>
>>> To repeat my concern, attempting to get a zero failure rate will cause
>>> other kinds of failure rates to increase.
>>
>> That's true. You implicitly need some kind of valuation of the different
>> failure rates. For instance, if you want LNHarm and LNHelp, you have to
>> give up either monotonicity or mutual majority. Which it's going to be
>> depends on what values you place on the different criteria.
>>
>> The same would hold for rates. Say you want to find the method that
>> minimizes w * x, where x is the rates of each failure type
>> (monotonicity, vote splitting, teaming, crowding, favorite betrayal...).
>> Then the weights of the w vector provide a measure of indifference: how
>> much of failure type 1 is an acceptable trade for one unit of failure
>> type 2?
>>
>> Or to put it differently: if the method insists on a zero failure rate
>> for Condorcet loser, why shouldn't it insist on a zero failure rate for
>> Condorcet winner, say? And, equivalently, if "merely a low rate of
>> failure" is good enough for the Condorcet criterion (or say, clone
>> independence), why is it not good enough for Condorcet loser?
>>
>>> I'm still willing to consider improvements, but it needs to find a
>>> balance between what voters can understand -- both through an animated
>>> video and through words -- and what yields low failure rates.
>>>
>>> Again, thank you Kristofer for applying your clear understanding to
>>> this revision from RCIPE 1 to RCIPE 2.
>>
>> You're welcome :-)
>>
>> -km
> ----
> Election-Methods mailing list - see https://electorama.com/em for list info
>
RT
Richard, the VoteFair guy
Fri, Jul 30, 2021 4:48 PM
On 7/28/2021 1:08 PM, Kristofer Munsterhjelm wrote:
On 27.07.2021 00:16, Richard, the VoteFair guy wrote:
On 7/26/2021 6:20 AM, Kristofer Munsterhjelm wrote:
...
... Is that right? ...
Your calculations are correct.
Then that's just Borda. Congrats, you've reinvented Borda-elimination :-)
The fact that you believe the RCIPE version 2 method is "borda
elimination" tells me:
-
Other people will make the same mistake of associating the method with
the Borda count method, which deserves its poor reputation.
-
The similarity would be used to attack the RCIPE (v2) method.
So, I'm choosing to abandon creating a version 2, and stick with RCIPE
version 1.
So, RCIPE continues to refer to IRV with the addition of a "safety net"
that eliminates Condorcet losers when they occur. That's less likely to
be confusing about what the method is.
Your opinion that RCIPE version 2 is borda elimination is supported by
the following opening sentence in the Wikipedia article titled "Borda
count":
"The Borda count is a family of positional voting rules which gives each
candidate, for each ballot, a number of points corresponding to the
number of candidates ranked lower."
Yet the words "positional voting" link to an article that clearly states
that the points are determined by the candidate's ranking.
And every example in the Borda count article uses a candidate's ranking
position. None of the examples counts how many other candidates are
ranked below the candidate.
It's true that "a number of points corresponding to the number of
candidates ranked lower" can be equal to the point count associated with
the ranking level. However that equivalence requires that every voter
is forced to rank every candidate at a different (unique) ranking level.
Interestingly I see that Baldwin's method is "borda elimination." Correct?
I had expected that RCIPE version 2 was likely to be a method that had
already been invented. It's so simple, and yet it yields very good results.
So if anyone knows about a method, besides Instant Pairwise Elimination
(IPE), that is similar to what I was calling RCIPE version 2, I'd still
be interested in knowing about that similarity.
Now I'll switch back to creating an animation of RCIPE (version 1).
Fortunately in the process of imagining ways to animate version 2 I've
thought of improved ways to animate version 1.
If I have time I may also animate IPE, which does use the code that
manipulates up arrows with candidate initials on them.
And ideally I'd like to animate the counting done using the
Condorcet-Kemeny method. (It would show that swapping the
sequence/order moves larger numbers of up arrows into the upper right
triangular area, while the cells with smaller numbers of up arrows move
into the lower left triangular area.)
One final interesting point is that my favorite single-winner method,
the Condorcet-Kemeny method, yields the same popularity sequence as the
Borda count IF every voter can be forced to rank each candidate at a
unique ranking level (and there are no unmarked ranking levels). If I'm
mistaken about this, please let me know.
Once again, thank you Kristofer for your valuable feedback!!!
Richard Fobes
On 7/28/2021 1:08 PM, Kristofer Munsterhjelm wrote:
On 27.07.2021 00:16, Richard, the VoteFair guy wrote:
On 7/26/2021 6:20 AM, Kristofer Munsterhjelm wrote:
...
... Is that right? ...
Your calculations are correct.
Then that's just Borda. Congrats, you've reinvented Borda-elimination :-)
Well, it's not quite Borda-elimination since the early elimination of a
Condorcet loser may reshuffle the order of later eliminations, and due
to the same effect that causes nonmonotonicity in elimination methods,
lead someone else to win.
But ordinary Borda-elimination does eventually remove every Condorcet
loser, so it should have little effect on your criterion compliances.
In particular, although if I were selfish I shouldn't point this out,
Borda-elimination passes Condorcet and the Condorcet loser elimination
stage doesn't change that. Since your simulator says that it doesn't
pass Condorcet, I would view its other results with some suspicion,
particularly given how it's also been wrong about clone independence.
But, since your method is pretty close to Borda elimination, and Borda
elimination also eliminates Condorcet losers as a matter of course[1],
you could simplify it (if you want to retain its Borda-elimination
nature) into this:
- Determine the candidate with the fewest arrows, and eliminate that
candidate. Break ties by IRV (and further ties by Ext-Minmax).
- Repeat from 1 until a single candidate remains. That candidate is the
winner.
Even if you don't want to, you could run tests comparing this simplified
version to the one with Condorcet loser elimination - though as
mentioned, I wouldn't be too confident of the results.
As for Condorcet winner, what I was objecting to (besides me being a
Condorcetist and preferring Condorcet) is that there doesn't seem to be
a principled approach to your methods' violation of CW. You used a
bicycle metaphor: but the attempt to deliberately avoid Condorcet winner
seems like removing the chain (and subsequently creating a
penny-farthing) just so that you can say that it's not a bicycle,
because the pedestrian organization (FairVote) dislikes bicycles.
If you're deliberately setting out to not make a bicycle because
bicycles aren't liked, you should create something that's consistently
not a bicycle: that gets some kind of return for not being a bicycle,
and then has more of a justification of what a good winner is than "this
isn't Condorcet, but you can see Condorcet from here".
Consider something like a method that explicitly eliminates the
Condorcet loser, and where the base method otherwise does not pass
Condorcet. Then the elimination method will pass Condorcet when there's
an unambiguous order of losers (X is the loser, Y beats only X, Z beats
only Y and X, etc.). The iterated Condorcet losers provide a
straightforward order of candidates from most relevant to least
relevant. But now introduce a cycle somewhere "downstream" of the
winner. Then by introduction of such a cycle, the Condorcet winner may
be eliminated early (depending on how the method is constructed). It's
difficult to see why that should make a difference; I'd say you'd need
some kind of explicit reason for why something that appears irrelevant
(the relation between people who would otherwise be losers) matters.
In any case, I don't think FairVote itself would support your method as
long as it's not IRV. They're first and foremost an advocacy
organization focused laser-like on IRV and STV, and as your method isn't
it, they probably won't support it. But even being generous that their
"Core Support criterion" isn't just a fig leaf (see my post to Forest),
any method that is Condorcet in the "straight line scenario" above must
by necessity fail that criterion.
Now, don't get me wrong: I don't think passing Core Support would win
them over - because again, their thing is IRV. I'm more just
re-emphasizing that a method that fails Condorcet should have some
reason for doing so. (E.g. you can get very close to Condorcet while
still passing the favorite betrayal condition - then passing the FBC
becomes the reason.)
-km
[1] For that matter, every sequential elimination method that respects
majority rule will also pass Condorcet Loser, because in the worst case
that the Condorcet loser remains until the final round, the other
candidate will by definition beat the loser in the final round.
On 7/28/2021 1:08 PM, Kristofer Munsterhjelm wrote:
> On 27.07.2021 00:16, Richard, the VoteFair guy wrote:
>> On 7/26/2021 6:20 AM, Kristofer Munsterhjelm wrote:
>>> ...
>>> ... Is that right? ...
>>
>> Your calculations are correct.
>
> Then that's just Borda. Congrats, you've reinvented Borda-elimination :-)
The fact that you believe the RCIPE version 2 method is "borda
elimination" tells me:
* Other people will make the same mistake of associating the method with
the Borda count method, which deserves its poor reputation.
* The similarity would be used to attack the RCIPE (v2) method.
So, I'm choosing to abandon creating a version 2, and stick with RCIPE
version 1.
So, RCIPE continues to refer to IRV with the addition of a "safety net"
that eliminates Condorcet losers when they occur. That's less likely to
be confusing about what the method is.
Your opinion that RCIPE version 2 is borda elimination is supported by
the following opening sentence in the Wikipedia article titled "Borda
count":
"The Borda count is a family of positional voting rules which gives each
candidate, for each ballot, a number of points corresponding to the
number of candidates ranked lower."
Yet the words "positional voting" link to an article that clearly states
that the points are determined by the candidate's ranking.
And every example in the Borda count article uses a candidate's ranking
position. None of the examples counts how many other candidates are
ranked below the candidate.
It's true that "a number of points corresponding to the number of
candidates ranked lower" can be equal to the point count associated with
the ranking level. However that equivalence requires that every voter
is forced to rank every candidate at a different (unique) ranking level.
Interestingly I see that Baldwin's method is "borda elimination." Correct?
I had expected that RCIPE version 2 was likely to be a method that had
already been invented. It's so simple, and yet it yields very good results.
So if anyone knows about a method, besides Instant Pairwise Elimination
(IPE), that is similar to what I was calling RCIPE version 2, I'd still
be interested in knowing about that similarity.
Now I'll switch back to creating an animation of RCIPE (version 1).
Fortunately in the process of imagining ways to animate version 2 I've
thought of improved ways to animate version 1.
If I have time I may also animate IPE, which does use the code that
manipulates up arrows with candidate initials on them.
And ideally I'd like to animate the counting done using the
Condorcet-Kemeny method. (It would show that swapping the
sequence/order moves larger numbers of up arrows into the upper right
triangular area, while the cells with smaller numbers of up arrows move
into the lower left triangular area.)
One final interesting point is that my favorite single-winner method,
the Condorcet-Kemeny method, yields the same popularity sequence as the
Borda count IF every voter can be forced to rank each candidate at a
unique ranking level (and there are no unmarked ranking levels). If I'm
mistaken about this, please let me know.
Once again, thank you Kristofer for your valuable feedback!!!
Richard Fobes
On 7/28/2021 1:08 PM, Kristofer Munsterhjelm wrote:
> On 27.07.2021 00:16, Richard, the VoteFair guy wrote:
>> On 7/26/2021 6:20 AM, Kristofer Munsterhjelm wrote:
>>> ...
>>> ... Is that right? ...
>>
>> Your calculations are correct.
>
> Then that's just Borda. Congrats, you've reinvented Borda-elimination :-)
>
> Well, it's not quite Borda-elimination since the early elimination of a
> Condorcet loser may reshuffle the order of later eliminations, and due
> to the same effect that causes nonmonotonicity in elimination methods,
> lead someone else to win.
>
> But ordinary Borda-elimination does eventually remove every Condorcet
> loser, so it should have little effect on your criterion compliances.
>
> In particular, although if I were selfish I shouldn't point this out,
> Borda-elimination passes Condorcet and the Condorcet loser elimination
> stage doesn't change that. Since your simulator says that it doesn't
> pass Condorcet, I would view its other results with some suspicion,
> particularly given how it's also been wrong about clone independence.
>
> But, since your method is pretty close to Borda elimination, and Borda
> elimination also eliminates Condorcet losers as a matter of course[1],
> you could simplify it (if you want to retain its Borda-elimination
> nature) into this:
>
> 1. Determine the candidate with the fewest arrows, and eliminate that
> candidate. Break ties by IRV (and further ties by Ext-Minmax).
> 2. Repeat from 1 until a single candidate remains. That candidate is the
> winner.
>
> Even if you don't want to, you could run tests comparing this simplified
> version to the one with Condorcet loser elimination - though as
> mentioned, I wouldn't be too confident of the results.
>
> As for Condorcet winner, what I was objecting to (besides me being a
> Condorcetist and preferring Condorcet) is that there doesn't seem to be
> a principled approach to your methods' violation of CW. You used a
> bicycle metaphor: but the attempt to deliberately avoid Condorcet winner
> seems like removing the chain (and subsequently creating a
> penny-farthing) just so that you can say that it's not a bicycle,
> because the pedestrian organization (FairVote) dislikes bicycles.
>
> If you're deliberately setting out to not make a bicycle because
> bicycles aren't liked, you should create something that's consistently
> not a bicycle: that gets some kind of return for not being a bicycle,
> and then has more of a justification of what a good winner is than "this
> isn't Condorcet, but you can see Condorcet from here".
>
> Consider something like a method that explicitly eliminates the
> Condorcet loser, and where the base method otherwise does not pass
> Condorcet. Then the elimination method will pass Condorcet when there's
> an unambiguous order of losers (X is the loser, Y beats only X, Z beats
> only Y and X, etc.). The iterated Condorcet losers provide a
> straightforward order of candidates from most relevant to least
> relevant. But now introduce a cycle somewhere "downstream" of the
> winner. Then by introduction of such a cycle, the Condorcet winner may
> be eliminated early (depending on how the method is constructed). It's
> difficult to see why that should make a difference; I'd say you'd need
> some kind of explicit reason for why something that appears irrelevant
> (the relation between people who would otherwise be losers) matters.
>
> In any case, I don't think FairVote itself would support your method as
> long as it's not IRV. They're first and foremost an advocacy
> organization focused laser-like on IRV and STV, and as your method isn't
> it, they probably won't support it. But even being generous that their
> "Core Support criterion" isn't just a fig leaf (see my post to Forest),
> any method that is Condorcet in the "straight line scenario" above must
> by necessity fail that criterion.
>
> Now, don't get me wrong: I don't think passing Core Support would win
> them over - because again, their thing is IRV. I'm more just
> re-emphasizing that a method that fails Condorcet should have some
> reason for doing so. (E.g. you can get very close to Condorcet while
> still passing the favorite betrayal condition - then passing the FBC
> becomes the reason.)
>
> -km
>
> [1] For that matter, every sequential elimination method that respects
> majority rule will also pass Condorcet Loser, because in the worst case
> that the Condorcet loser remains until the final round, the other
> candidate will by definition beat the loser in the final round.
>
RL
Richard Lung
Sun, Aug 15, 2021 5:05 AM
Looking at election method as a purely mathematical problem, the
objection to existing voting method is that it lacks a complete scale of
measurement of candidate support, positive and negative. This is achievd
by making an exclusion count the polar opposite of an election count, on
the same continuum. The zero point in the middle is the zero surplus
votes of just elected candidates. Or alternatively the zero deficit
votes of just not unelected candidates.
Once youve got this bipolar (or indeed binomial) count youve got one
complete dimension, a basic standard of scientific measurement.
(It's possible to go onto more than one dimension, as used in natural
science.)
Richard Lung.
Dear Richard Fobes, the VoteFair guy, and all at election methods,
Personal family misfortunes prevent me from giving a proper reply.
I believe I say something about Kemeny in an appendix to my Smashwords
free ebook,
FAB STV: Four Averages Binomial Single Transferable Vote.
https://www.smashwords.com/books/view/806030
https://www.smashwords.com/books/view/806030
The system is fully described in the second part, but there are plenty
of summaries, from simple list of attributes, to summary convenient for
those familiar with Meek method.
I may mention what the system looks like from the voters point of view.
It could be any preference voting ballot. But it counts differently.
Last preferences help as much to exclude candidates, as first
preferences help to elect candidates. That is to say it is a binomial
count.
It is not necessary to fill in all the preferences. Vacant preferences
count towards a NOTA quota, leaving a seat unfilled.
Circumstances permitting, I hope to say more about your post and others.
Regards
Richard Lung.
On 28 Jul 2021, at 4:43 am, Richard, the VoteFair guy
<electionmethods@votefair.org mailto:electionmethods@votefair.org> wrote:
On 7/25/2021 2:00 PM, Richard Lung wrote:
... eliminating candidates, during the count, loses voting
information, before the count is over. ...
My favorite way to count ranked-choice ballots for a single-winner
election is the Condorcet-Kemeny method.
Not only does it not eliminate candidates one at a time, it also does
not identify the first-place winner as a first step, and the
second-place winner as a second step, etc. Instead it isn't finished
until the entire sequence from most popular, second-most popular, and so
on down to least popular has been determined.
I repeat, in case you missed it, science demands one truth (to aspire
to) not two. Therefore an election count and an exclusion/elimination
count must be symmetrical. Call it symmetrical count requirement. But
that is a binomial count.
The Condorcet-Kemeny method is symmetrical.
Whereas, FAB STV is the whole caboodle perhaps relevant to data
mining. This system is monotonic, not vulnerable to strategic
shuffling the preference orders. It avoids premature exclusion, and
indeed premature election! It entirely avoids later harm, not just for
transfer of surplus preferences.
It meets the Laplace condition of weighting a whole range of
preferences in order of importance, unlike Condorcet pairing, whether
or not the pairs are weighted in relative importance.
FAB STV recognises elections as statistical estimates of
representation, and employs up to four averages to maximise accuracy.
I have read your earlier posts but I don't recall seeing a description
of your FAB STV method.
I looked at Electowiki but it's not there (under that name).
I'd be happy to look at FAB STV if you can point me to a definition of
the method.
It accepts the "Impossibility" of a deterministic election result, and
moves on. I beseech you all to do the same!
Remember that the RCIPE method (version 1 or 2) is a stepping stone that
allows reaching better methods such as the Condorcet-Kemeny method, and
beyond to PR methods.
Regarding the beyond part, my VoteFair Ranking system includes a
two-seat kind of "STV," and that can be extended to a higher number of
seats. That's why I'm curious to read a definition of your FAB STV
counting method.
To put things into perspective, consider a metaphor. IRV is like a
tricycle for tots who aren't yet ready for a bicycle. The Condorcet
methods are like a bicycle, very useful in many situations, but they
can't do everything. RCIPE is like training wheels for the bicycle. When
enough voters learn how to vote using ranked-choice ballots and RCIPE
counting then they will be ready to move on to Condorcet methods and beyond.
Richard Lung, thank you for your thoughts.
Richard Fobes
The VoteFair guy
On 7/25/2021 2:00 PM, Richard Lung wrote:
A few comments from Richard Lung (not the VoteFair guy, (who is not to
be confused, if I remember rightly, with Santucci, the vote guy!).
As pointed out to Susan Simmons, which she acknowledged, eliminating
candidates, during the count, loses voting information, before the
count is over. It is not necessary with a binomial count, unlike all
existing methods (uninomial counts) which employ elimination as an
afterthought to an essentially uninomial election count.
I repeat, in case you missed it, science demands one truth (to aspire
to) not two. Therefore an election count and an exclusion/elimination
count must be symmetrical. Call it symmetrical count requirement. But
that is a binomial count.
And a binomial count indeed does imply higher order counts, governed
by the binomial theorem. But a simple coherent first order binomial
count should be sufficient for democratic representation.
Whereas, FAB STV is the whole caboodle perhaps relevant to data
mining. This system is monotonic, not vulnerable to strategic
shuffling the preference orders. It avoids premature exclusion, and
indeed premature election! It entirely avoids later harm, not just for
transfer of surplus preferences.
It meets the Laplace condition of weighting a whole range of
preferences in order of importance, unlike Condorcet pairing, whether
or not the pairs are weighted in relative importance.
FAB STV recognises elections as statistical estimates of
representation, and employs up to four averages to maximise accuracy.
It accepts the "Impossibility" of a deterministic election result, and
moves on. I beseech you all to do the same!
Yours sincerely,
Richard Lung.
On 25 Jul 2021, at 4:30 pm, Richard, the VoteFair guy
<electionmethods@votefair.org mailto:electionmethods@votefair.org>
wrote:
On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
On 7/22/21 5:31 PM, VoteFair wrote:
On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
How about this?
- Eliminate the candidate with the least number of winning subgroups.
- If there is a tie, break that tie by IRV.
...
Isn't the first step basically Copeland's method?
No, because there's no elimination in Copeland (and it doesn't pass
LIIA). It would just elect the candidate/s with the most
winning subgroups.
I see you're right, of course.
I admit your suggestion is clever because it includes Condorcet loser
elimination.
Yet I'm sure lots of non-math-savvy voters will not trust that the
candidate with the least number of wins is not always the least
popular. I too share that lack of trust.
Keep in mind that lots of voter don't trust the idea that the winner
of all the pairwise contests is always the most popular.
But then clone independence is not important after all because the
methods are ugly. I can't quite determine whether clone independence is
important or not.
It's important that the failure rate is small. But it doesn't need to
be zero.
That's true. You implicitly need some kind of valuation of the different
failure rates. For instance, if you want LNHarm and LNHelp, you have to
give up either monotonicity or mutual majority. Which it's going to be
depends on what values you place on the different criteria.
I'm not concerned about monotonicity, LNHarm, LNHelp or any other On
7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
On 7/22/21 5:31 PM, VoteFair wrote:
On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
How about this?
- Eliminate the candidate with the least number of winning subgroups.
- If there is a tie, break that tie by IRV.
...
Isn't the first step basically Copeland's method?
No, because there's no elimination in Copeland (and it doesn't pass
LIIA). It would just elect the candidate/s with the most
winning subgroups.
I see you're right, of course.
I admit your suggestion is clever because it includes Condorcet loser
elimination.
Yet I'm sure lots of non-math-savvy voters will not trust that the
candidate with the least number of wins is not always the least
popular. I too share that lack of trust.
Keep in mind that lots of voter don't trust the idea that the winner
of all the pairwise contests is always the most popular.
But then clone independence is not important after all because the
methods are ugly. I can't quite determine whether clone independence is
important or not.
It's important that the failure rate is small. But it doesn't need to
be zero.
That's true. You implicitly need some kind of valuation of the different
failure rates. For instance, if you want LNHarm and LNHelp, you have to
give up either monotonicity or mutual majority. Which it's going to be
depends on what values you place on the different criteria.
I'm not concerned about monotonicity, LNHarm, LNHelp or any other
failures that are difficult to exploit. I'm much more concerned about
exploitable failures.
Admittedly, as a fan of Condorcet-Kemeny, I favor looking deep into
the ballots, and I favor ways of "sorting" that basically move the
biggest pairwise counts into one half of the usual matrix while moving
the smallest pairwise counts into the other half, where the dividing
line is the diagonal where candidates are paired with themselves.
Or to put it differently: if the method insists on a zero failure rate
for Condorcet loser, why shouldn't it insist on a zero failure rate for
Condorcet winner, say? And, equivalently, if "merely a low rate of
failure" is good enough for the Condorcet criterion (or say, clone
independence), why is it not good enough for Condorcet loser?
I admit I'm intentionally avoiding a zero failure rate for Condorcet
winner because that makes the method into a Condorcet method, and
those have been vilified (portrayed as evil) by the FairVote
organization, and to some extent by STAR fans.
Plus, just as a voter is not likely to trust that the candidate with
the fewest wins is least popular, they aren't likely to trust that the
candidate who wins all the pairwise matches is most popular.
So at this point I'm still happy with eliminating the Condorcet loser
as the top priority and otherwise eliminating the candidate who has
the smallest pairwise support count (which basically counts how many
remaining candidates are ranked below the candidate being scored).
At this point I continue to be open to suggestions for something
better, but that window of time is closing very soon.
Again, thank you Kristofer for your wise feedback!
Richard Fobes
The VoteFair guy
On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
On 7/22/21 5:31 PM, VoteFair wrote:
On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
How about this?
- Eliminate the candidate with the least number of winning subgroups.
- If there is a tie, break that tie by IRV.
...
Isn't the first step basically Copeland's method?
No, because there's no elimination in Copeland (and it doesn't pass
LIIA). It would just elect the candidate/s with the most winning
subgroups.
That's an ugly "method" that fails to look beneath the surface.
IRV also fails to look beneath the surface, which is why it too is an
"ugly" method.
That leads me to wonder which is the case.
You said you couldn't replace the IRV tiebreaker with minmax elimination
because IRV is cloneproof and minmax is not -- that clone independence
was important because it "protects against money-based vote splitting
tactics". So I found something that invokes IRV's clone independence
more often.
But then clone independence is not important after all because the
methods are ugly. I can't quite determine whether clone independence is
important or not.
But again, the ungrouped mechanic is not cloneproof.
Being cloneproof is not a goal. The goal is to have a very small
failure rate for clone independence.
Then you could check the alternatives by that metric. A method seeming
ugly may not necessarily have any bearing on the rates of failure.
Also, electing the Condorcet winner is not a goal. The goal is to have
a very small Condorcet criteria failure rate.
To repeat my concern, attempting to get a zero failure rate will cause
other kinds of failure rates to increase.
That's true. You implicitly need some kind of valuation of the different
failure rates. For instance, if you want LNHarm and LNHelp, you have to
give up either monotonicity or mutual majority. Which it's going to be
depends on what values you place on the different criteria.
The same would hold for rates. Say you want to find the method that
minimizes w * x, where x is the rates of each failure type
(monotonicity, vote splitting, teaming, crowding, favorite betrayal...).
Then the weights of the w vector provide a measure of indifference: how
much of failure type 1 is an acceptable trade for one unit of failure
type 2?
Or to put it differently: if the method insists on a zero failure rate
for Condorcet loser, why shouldn't it insist on a zero failure rate for
Condorcet winner, say? And, equivalently, if "merely a low rate of
failure" is good enough for the Condorcet criterion (or say, clone
independence), why is it not good enough for Condorcet loser?
I'm still willing to consider improvements, but it needs to find a
balance between what voters can understand -- both through an animated
video and through words -- and what yields low failure rates.
Again, thank you Kristofer for applying your clear understanding to
this revision from RCIPE 1 to RCIPE 2.
Looking at election method as a purely mathematical problem, the
objection to existing voting method is that it lacks a complete scale of
measurement of candidate support, positive and negative. This is achievd
by making an exclusion count the polar opposite of an election count, on
the same continuum. The zero point in the middle is the zero surplus
votes of just elected candidates. Or alternatively the zero deficit
votes of just not unelected candidates.
Once youve got this bipolar (or indeed binomial) count youve got one
complete dimension, a basic standard of scientific measurement.
(It's possible to go onto more than one dimension, as used in natural
science.)
Richard Lung.
Dear Richard Fobes, the VoteFair guy, and all at election methods,
Personal family misfortunes prevent me from giving a proper reply.
I believe I say something about Kemeny in an appendix to my Smashwords
free ebook,
FAB STV: Four Averages Binomial Single Transferable Vote.
https://www.smashwords.com/books/view/806030
<https://www.smashwords.com/books/view/806030>
The system is fully described in the second part, but there are plenty
of summaries, from simple list of attributes, to summary convenient for
those familiar with Meek method.
I may mention what the system looks like from the voters point of view.
It could be any preference voting ballot. But it counts differently.
Last preferences help as much to exclude candidates, as first
preferences help to elect candidates. That is to say it is a binomial
count.
It is not necessary to fill in all the preferences. Vacant preferences
count towards a NOTA quota, leaving a seat unfilled.
Circumstances permitting, I hope to say more about your post and others.
Regards
Richard Lung.
On 28 Jul 2021, at 4:43 am, Richard, the VoteFair guy
<electionmethods@votefair.org <mailto:electionmethods@votefair.org>> wrote:
On 7/25/2021 2:00 PM, Richard Lung wrote:
> ... eliminating candidates, during the count, loses voting
> information, before the count is over. ...
My favorite way to count ranked-choice ballots for a single-winner
election is the Condorcet-Kemeny method.
Not only does it not eliminate candidates one at a time, it also does
not identify the first-place winner as a first step, and the
second-place winner as a second step, etc. Instead it isn't finished
until the entire sequence from most popular, second-most popular, and so
on down to least popular has been determined.
> I repeat, in case you missed it, science demands one truth (to aspire
> to) not two. Therefore an election count and an exclusion/elimination
> count must be symmetrical. Call it symmetrical count requirement. But
> that is a binomial count.
The Condorcet-Kemeny method is symmetrical.
> Whereas, FAB STV is the whole caboodle perhaps relevant to data
> mining. This system is monotonic, not vulnerable to strategic
> shuffling the preference orders. It avoids premature exclusion, and
> indeed premature election! It entirely avoids later harm, not just for
> transfer of surplus preferences.
> It meets the Laplace condition of weighting a whole range of
> preferences in order of importance, unlike Condorcet pairing, whether
> or not the pairs are weighted in relative importance.
> FAB STV recognises elections as statistical estimates of
> representation, and employs up to four averages to maximise accuracy.
I have read your earlier posts but I don't recall seeing a description
of your FAB STV method.
I looked at Electowiki but it's not there (under that name).
I'd be happy to look at FAB STV if you can point me to a definition of
the method.
> It accepts the "Impossibility" of a deterministic election result, and
> moves on. I beseech you all to do the same!
Remember that the RCIPE method (version 1 or 2) is a stepping stone that
allows reaching better methods such as the Condorcet-Kemeny method, and
beyond to PR methods.
Regarding the beyond part, my VoteFair Ranking system includes a
two-seat kind of "STV," and that can be extended to a higher number of
seats. That's why I'm curious to read a definition of your FAB STV
counting method.
To put things into perspective, consider a metaphor. IRV is like a
tricycle for tots who aren't yet ready for a bicycle. The Condorcet
methods are like a bicycle, very useful in many situations, but they
can't do everything. RCIPE is like training wheels for the bicycle. When
enough voters learn how to vote using ranked-choice ballots and RCIPE
counting then they will be ready to move on to Condorcet methods and beyond.
Richard Lung, thank you for your thoughts.
Richard Fobes
The VoteFair guy
> On 7/25/2021 2:00 PM, Richard Lung wrote:
>
>
> A few comments from Richard Lung (not the VoteFair guy, (who is not to
> be confused, if I remember rightly, with Santucci, the vote guy!).
>
> As pointed out to Susan Simmons, which she acknowledged, eliminating
> candidates, during the count, loses voting information, before the
> count is over. It is not necessary with a binomial count, unlike all
> existing methods (uninomial counts) which employ elimination as an
> afterthought to an essentially uninomial election count.
> I repeat, in case you missed it, science demands one truth (to aspire
> to) not two. Therefore an election count and an exclusion/elimination
> count must be symmetrical. Call it symmetrical count requirement. But
> that is a binomial count.
> And a binomial count indeed does imply higher order counts, governed
> by the binomial theorem. But a simple coherent first order binomial
> count should be sufficient for democratic representation.
> Whereas, FAB STV is the whole caboodle perhaps relevant to data
> mining. This system is monotonic, not vulnerable to strategic
> shuffling the preference orders. It avoids premature exclusion, and
> indeed premature election! It entirely avoids later harm, not just for
> transfer of surplus preferences.
> It meets the Laplace condition of weighting a whole range of
> preferences in order of importance, unlike Condorcet pairing, whether
> or not the pairs are weighted in relative importance.
> FAB STV recognises elections as statistical estimates of
> representation, and employs up to four averages to maximise accuracy.
> It accepts the "Impossibility" of a deterministic election result, and
> moves on. I beseech you all to do the same!
>
> Yours sincerely,
> Richard Lung.
>
>
>
>
> On 25 Jul 2021, at 4:30 pm, Richard, the VoteFair guy
> <electionmethods@votefair.org <mailto:electionmethods@votefair.org>>
> wrote:
>
> On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
>>> On 7/22/21 5:31 PM, VoteFair wrote:
>>> On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
>>>> How about this?
>>>>
>>>> - Eliminate the candidate with the least number of winning subgroups.
>>>> - If there is a tie, break that tie by IRV.
>>>> ...
>>>
>>> Isn't the first step basically Copeland's method?
>>
>> No, because there's no elimination in Copeland (and it doesn't pass
>> LIIA). It would just elect the candidate/s with the most
>> winning subgroups.
>
> I see you're right, of course.
>
> I admit your suggestion is clever because it includes Condorcet loser
> elimination.
>
> Yet I'm sure lots of non-math-savvy voters will not trust that the
> candidate with the least number of wins is not always the least
> popular. I too share that lack of trust.
>
> Keep in mind that lots of voter don't trust the idea that the winner
> of all the pairwise contests is always the most popular.
>
>> But then clone independence is not important after all because the
>> methods are ugly. I can't quite determine whether clone independence is
>> important or not.
>
> It's important that the failure rate is small. But it doesn't need to
> be zero.
>
>> That's true. You implicitly need some kind of valuation of the different
>> failure rates. For instance, if you want LNHarm and LNHelp, you have to
>> give up either monotonicity or mutual majority. Which it's going to be
>> depends on what values you place on the different criteria.
>
> I'm not concerned about monotonicity, LNHarm, LNHelp or any other On
> 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
>>> On 7/22/21 5:31 PM, VoteFair wrote:
>>> On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
>>>> How about this?
>>>>
>>>> - Eliminate the candidate with the least number of winning subgroups.
>>>> - If there is a tie, break that tie by IRV.
>>>> ...
>>>
>>> Isn't the first step basically Copeland's method?
>>
>> No, because there's no elimination in Copeland (and it doesn't pass
>> LIIA). It would just elect the candidate/s with the most
>> winning subgroups.
>
> I see you're right, of course.
>
> I admit your suggestion is clever because it includes Condorcet loser
> elimination.
>
> Yet I'm sure lots of non-math-savvy voters will not trust that the
> candidate with the least number of wins is not always the least
> popular. I too share that lack of trust.
>
> Keep in mind that lots of voter don't trust the idea that the winner
> of all the pairwise contests is always the most popular.
>
>> But then clone independence is not important after all because the
>> methods are ugly. I can't quite determine whether clone independence is
>> important or not.
>
> It's important that the failure rate is small. But it doesn't need to
> be zero.
>
>> That's true. You implicitly need some kind of valuation of the different
>> failure rates. For instance, if you want LNHarm and LNHelp, you have to
>> give up either monotonicity or mutual majority. Which it's going to be
>> depends on what values you place on the different criteria.
>
> I'm not concerned about monotonicity, LNHarm, LNHelp or any other
> failures that are difficult to exploit. I'm much more concerned about
> exploitable failures.
>
> Admittedly, as a fan of Condorcet-Kemeny, I favor looking deep into
> the ballots, and I favor ways of "sorting" that basically move the
> biggest pairwise counts into one half of the usual matrix while moving
> the smallest pairwise counts into the other half, where the dividing
> line is the diagonal where candidates are paired with themselves.
>
>> Or to put it differently: if the method insists on a zero failure rate
>> for Condorcet loser, why shouldn't it insist on a zero failure rate for
>> Condorcet winner, say? And, equivalently, if "merely a low rate of
>> failure" is good enough for the Condorcet criterion (or say, clone
>> independence), why is it not good enough for Condorcet loser?
>
> I admit I'm intentionally avoiding a zero failure rate for Condorcet
> winner because that makes the method into a Condorcet method, and
> those have been vilified (portrayed as evil) by the FairVote
> organization, and to some extent by STAR fans.
>
> Plus, just as a voter is not likely to trust that the candidate with
> the fewest wins is least popular, they aren't likely to trust that the
> candidate who wins all the pairwise matches is most popular.
>
> So at this point I'm still happy with eliminating the Condorcet loser
> as the top priority and otherwise eliminating the candidate who has
> the smallest pairwise support count (which basically counts how many
> remaining candidates are ranked below the candidate being scored).
>
> At this point I continue to be open to suggestions for something
> better, but that window of time is closing very soon.
>
> Again, thank you Kristofer for your wise feedback!
>
> Richard Fobes
> The VoteFair guy
>
>
>>> On 7/24/2021 2:19 PM, Kristofer Munsterhjelm wrote:
>>> On 7/22/21 5:31 PM, VoteFair wrote:
>>>> On 7/22/2021 6:04 AM, Kristofer Munsterhjelm wrote:
>>>> How about this?
>>>>
>>>> - Eliminate the candidate with the least number of winning subgroups.
>>>> - If there is a tie, break that tie by IRV.
>>>> ...
>>>
>>> Isn't the first step basically Copeland's method?
>>
>> No, because there's no elimination in Copeland (and it doesn't pass
>> LIIA). It would just elect the candidate/s with the most winning
>> subgroups.
>>
>>> That's an ugly "method" that fails to look beneath the surface.
>>>
>>> IRV also fails to look beneath the surface, which is why it too is an
>>> "ugly" method.
>>
>> That leads me to wonder which is the case.
>>
>> You said you couldn't replace the IRV tiebreaker with minmax elimination
>> because IRV is cloneproof and minmax is not -- that clone independence
>> was important because it "protects against money-based vote splitting
>> tactics". So I found something that invokes IRV's clone independence
>> more often.
>>
>> But then clone independence is not important after all because the
>> methods are ugly. I can't quite determine whether clone independence is
>> important or not.
>>
>>>> But again, the ungrouped mechanic is not cloneproof.
>>>
>>> Being cloneproof is not a goal. The goal is to have a very small
>>> failure rate for clone independence.
>>
>> Then you could check the alternatives by that metric. A method seeming
>> ugly may not necessarily have any bearing on the rates of failure.
>>
>>> Also, electing the Condorcet winner is not a goal. The goal is to have
>>> a very small Condorcet criteria failure rate.
>>>
>>> To repeat my concern, attempting to get a zero failure rate will cause
>>> other kinds of failure rates to increase.
>>
>> That's true. You implicitly need some kind of valuation of the different
>> failure rates. For instance, if you want LNHarm and LNHelp, you have to
>> give up either monotonicity or mutual majority. Which it's going to be
>> depends on what values you place on the different criteria.
>>
>> The same would hold for rates. Say you want to find the method that
>> minimizes w * x, where x is the rates of each failure type
>> (monotonicity, vote splitting, teaming, crowding, favorite betrayal...).
>> Then the weights of the w vector provide a measure of indifference: how
>> much of failure type 1 is an acceptable trade for one unit of failure
>> type 2?
>>
>> Or to put it differently: if the method insists on a zero failure rate
>> for Condorcet loser, why shouldn't it insist on a zero failure rate for
>> Condorcet winner, say? And, equivalently, if "merely a low rate of
>> failure" is good enough for the Condorcet criterion (or say, clone
>> independence), why is it not good enough for Condorcet loser?
>>
>>> I'm still willing to consider improvements, but it needs to find a
>>> balance between what voters can understand -- both through an animated
>>> video and through words -- and what yields low failure rates.
>>>
>>> Again, thank you Kristofer for applying your clear understanding to
>>> this revision from RCIPE 1 to RCIPE 2.
>>
>> You're welcome :-)
>>
>> -km
> ----
> Election-Methods mailing list - see https://electorama.com/em
> <https://electorama.com/em> for list info
>
