POLL: Approaching deadline: 2024-05-11 05:15:00 UTC
And I should just as well make a post that the deadline for the EM poll
is approaching, so others don't forget to post their ballots before
time's up :-)
Unless I made a calculation error, there's just short of ten hours left.
I'll make another post with all the manipulability stats in a single
message after this, in case it'll be useful, e.g. for breaking ties.
-km
It probably would have been better to have more warning for the impending end of the poll. I'm not going to have time to give this a proper go now. I know it has been referred to previously, but with so many posts on the topic, it's not that easy to find.
Toby
On Friday, 10 May 2024 at 20:21:54 BST, Kristofer Munsterhjelm km_elmet@t-online.de wrote:
And I should just as well make a post that the deadline for the EM poll
is approaching, so others don't forget to post their ballots before
time's up :-)
Unless I made a calculation error, there's just short of ten hours left.
I'll make another post with all the manipulability stats in a single
message after this, in case it'll be useful, e.g. for breaking ties.
-km
Election-Methods mailing list - see https://electorama.com/em for list info
On 2024-05-10 23:49, Toby Pereira wrote:
It probably would have been better to have more warning for the
impending end of the poll. I'm not going to have time to give this a
proper go now. I know it has been referred to previously, but with so
many posts on the topic, it's not that easy to find.
That's entirely understandable. Sorry about that; in retrospect I should
have posted that earlier.
-km
We could just postpone it.
But in any case I don't know what this poll is supposed to accomplish.
- The number of methods is so large that everyone's ballot is missing half
the methods, I presume on an oversight, because there's just no way people
have that strong of an opinion on - The pairwise matrix is going to end up so degenerate as to be useless.
It'll be full of cycles and ties. The number of voters is so small, and the
number of outcomes is sufficiently large, that the pairwise matrix is
effectively meaningless and dominated by noise. Every outcome will look
something like "4-2"—in other words, random noise. - I think all of us know each other's rough opinions on this topic.
The issue with this poll is it approached the topic as a question of voting
when it isn't. Good voting methods aren't always good statistical methods.
If I had cardinal ballots, I could fit a multilevel hierarchical model and
possibly get something meaningful out of it. But as is the data here is
effectively useless.
On Fri, May 10, 2024 at 3:10 PM Kristofer Munsterhjelm km_elmet@t-online.de
wrote:
On 2024-05-10 23:49, Toby Pereira wrote:
It probably would have been better to have more warning for the
impending end of the poll. I'm not going to have time to give this a
proper go now. I know it has been referred to previously, but with so
many posts on the topic, it's not that easy to find.
That's entirely understandable. Sorry about that; in retrospect I should
have posted that earlier.
-km
Election-Methods mailing list - see https://electorama.com/em for list
info
On Fri, May 10, 2024 at 18:13 Closed Limelike Curves <
closed.limelike.curves@gmail.com> wrote:
We could just postpone it.
I suggested a two day extension of the voting-deadline.
I take your suggestion as a second to that motion.
But in any case I don't know what this poll is supposed to accomplish.
- The number of methods is so large that everyone's ballot is missing
half the methods…
That’s expected. In many-candidate elections, people usually won’t rank all
of the alternatives. No problem. The rankings express their preferences
among their most preferred.
It isn’t clear why you perceive that as a problem.
I presume on an oversight, because there's just no way people have that
strong of an opinion on
Meaning unclear. I didn’t expect so many nominees. So what. Approval & the
Condorcet complying methods don’t have a problem with lots of candidates.
- The pairwise matrix is going to end up so degenerate as to be useless.
Unsupported statement, unclear in meaning.
It'll be full of cycles and ties.
Top-cycles for 1st place are quite rare.
CIVS has done lots of pills with many candidate’s, with no problems.
With many candidates, & successive counts among the remaining no winners
(to determine the finishing-order, the social-ranking), there could be a
spontaneous sincere cycle somewhere in the finishing-order determination.
Schulze & RP(WC) are regarded by many as the best for sincere cycles.
…regarded as meeting the most criteria with a consensus of importance.
Closed, when arriving at a forum, the standard advice is to listen & ask
questions for a while before asserting. You’re doing far too much
asserting, when you should instead be listening & asking.
The number of voters is so small, and the number of outcomes is
sufficiently large, that the pairwise matrix is effectively meaningless and
dominated by noise.
Nonsense. It aggregates the expressed pairwise-preferences, to find the CW,
the candidate who’d beat everything else in 2-way votes.
In the event of no CW, that absence will be shown. …& Schulz & RP(wv) will
choose in the best accordance with established criteria.
The issue with this poll is it approached the topic as a question of
voting when it isn't.
:-) I don’t know what your question is. If it’s different, feel free to
ask it.
Good voting methods aren't always good statistical methods.
If I had cardinal ballots, I could fit a multilevel hierarchical model and
possibly get something meaningful out of it. But as is the data here is
effectively useless.
As I said, you assert far too much. Whoa, we’re getting ahead of ourselves,
Cowboy.
You spoke of “noise”. You’re making a nuisance noise of yourself.
On Fri, May 10, 2024 at 3:10 PM Kristofer Munsterhjelm <
km_elmet@t-online.de> wrote:
On 2024-05-10 23:49, Toby Pereira wrote:
It probably would have been better to have more warning for the
impending end of the poll. I'm not going to have time to give this a
proper go now. I know it has been referred to previously, but with so
many posts on the topic, it's not that easy to find.
That's entirely understandable. Sorry about that; in retrospect I should
have posted that earlier.
-km
Election-Methods mailing list - see https://electorama.com/em for list
info
Election-Methods mailing list - see https://electorama.com/em for list
info
I am happy with a 2-day extension.
I think the poll could have been conceived a bit better in the first
place, taking account of ballot restrictions in the US that are fixed or
difficult to change. We could have had parallel polls for different
ballot types, including those we most like and those we might have to
put up with.
Chris
On 11/05/2024 12:19 pm, Michael Ossipoff wrote:
On Fri, May 10, 2024 at 18:13 Closed Limelike Curves
closed.limelike.curves@gmail.com wrote:
We could just postpone it.
I suggested a two day extension of the voting-deadline.
I take your suggestion as a second to that motion.
But in any case I don't know what this poll is supposed to
accomplish.
1. The number of methods is so large that everyone's ballot is
missing half the methods…
That’s expected. In many-candidate elections, people usually won’t
rank all of the alternatives. No problem. The rankings express their
preferences among their most preferred.
It isn’t clear why you perceive that as a problem.
I presume on an oversight, because there's just no way people have
that strong of an opinion on
Meaning unclear. I didn’t expect so many nominees. So what. Approval &
the Condorcet complying methods don’t have a problem with lots of
candidates.
2. The pairwise matrix is going to end up so degenerate as to be
useless.
Unsupported statement, unclear in meaning.
It'll be full of cycles and ties.
Top-cycles for 1st place are quite rare.
CIVS has done lots of pills with many candidate’s, with no problems.
With many candidates, & successive counts among the remaining no
winners (to determine the finishing-order, the social-ranking), there
could be a spontaneous sincere cycle somewhere in the finishing-order
determination.
Schulze & RP(WC) are regarded by many as the best for sincere cycles.
…regarded as meeting the most criteria with a consensus of importance.
Closed, when arriving at a forum, the standard advice is to listen &
ask questions for a while before asserting. You’re doing far too much
asserting, when you should instead be listening & asking.
The number of voters is so small, and the number of outcomes is
sufficiently large, that the pairwise matrix is effectively
meaningless and dominated by noise.
Nonsense. It aggregates the expressed pairwise-preferences, to find
the CW, the candidate who’d beat everything else in 2-way votes.
In the event of no CW, that absence will be shown. …& Schulz & RP(wv)
will choose in the best accordance with established criteria.
The issue with this poll is it approached the topic as a question
of voting when it isn't.
:-) I don’t know what your question is. If it’s different, feel free
to ask it.
Good voting methods aren't always good statistical methods.
If I had cardinal ballots, I could fit a multilevel hierarchical
model and possibly get something meaningful out of it. But as is
the data here is effectively useless.
As I said, you assert far too much. Whoa, we’re getting ahead of
ourselves, Cowboy.
You spoke of “noise”. You’re making a nuisance noise of yourself.
On Fri, May 10, 2024 at 3:10 PM Kristofer Munsterhjelm
<km_elmet@t-online.de> wrote:
On 2024-05-10 23:49, Toby Pereira wrote:
It probably would have been better to have more warning for the
impending end of the poll. I'm not going to have time to
give this a
proper go now. I know it has been referred to previously,
but with so
many posts on the topic, it's not that easy to find.
That's entirely understandable. Sorry about that; in
retrospect I should
have posted that earlier.
-km
----
Election-Methods mailing list - see https://electorama.com/em
for list info
----
Election-Methods mailing list - see https://electorama.com/em for
list info
Election-Methods mailing list - seehttps://electorama.com/em for list info
On 2024-05-11 11:57, Chris Benham wrote:
I am happy with a 2-day extension.
What do you think about the five-day idea?
-km
5 days works for me! Though I might be able to get in my vote with two days.
On Saturday, 11 May 2024 at 14:24:22 BST, Kristofer Munsterhjelm km_elmet@t-online.de wrote:
On 2024-05-11 11:57, Chris Benham wrote:
I am happy with a 2-day extension.
What do you think about the five-day idea?
-km
Election-Methods mailing list - see https://electorama.com/em for list info
If 5 days gives more people time to choose a ranking, the deadline having
caught them unprepared, then that sounds better. I vote for 5.
Also, more time for people to say what they meant to about method-merit.
On Sat, May 11, 2024 at 06:24 Kristofer Munsterhjelm km_elmet@t-online.de
wrote:
On 2024-05-11 11:57, Chris Benham wrote:
I am happy with a 2-day extension.
What do you think about the five-day idea?
-km
Election-Methods mailing list - see https://electorama.com/em for list
info
Very much agreed. I move we postpone the poll indefinitely.
On Sat, May 11, 2024 at 3:13 AM Closed Limelike Curves <
closed.limelike.curves@gmail.com> wrote:
We could just postpone it.
But in any case I don't know what this poll is supposed to accomplish.
- The number of methods is so large that everyone's ballot is missing
half the methods, I presume on an oversight, because there's just no way
people have that strong of an opinion on - The pairwise matrix is going to end up so degenerate as to be useless.
It'll be full of cycles and ties. The number of voters is so small, and
the number of outcomes is sufficiently large, that the pairwise matrix is
effectively meaningless and dominated by noise. Every outcome will look
something like "4-2"—in other words, random noise. - I think all of us know each other's rough opinions on this topic.
The issue with this poll is it approached the topic as a question of
voting when it isn't. Good voting methods aren't always good statistical
methods.
If I had cardinal ballots, I could fit a multilevel hierarchical model and
possibly get something meaningful out of it. But as is the data here is
effectively useless.
On Fri, May 10, 2024 at 3:10 PM Kristofer Munsterhjelm <
km_elmet@t-online.de> wrote:
On 2024-05-10 23:49, Toby Pereira wrote:
It probably would have been better to have more warning for the
impending end of the poll. I'm not going to have time to give this a
proper go now. I know it has been referred to previously, but with so
many posts on the topic, it's not that easy to find.
That's entirely understandable. Sorry about that; in retrospect I should
have posted that earlier.
-km
Election-Methods mailing list - see https://electorama.com/em for list
info
Election-Methods mailing list - see https://electorama.com/em for list
info