Hi Joe,

I'm curious: what do you think the best voting system would be for
WQXR's purposes?

Rob

On Fri, Dec 5, 2025 at 6:52 PM Joseph Malkevitch via Election-Methods <
election-methods@lists.electorama.com> wrote:

Let's see if this works. I've been having problems with my mail provider
dropping EM messages due to Dreamhost being on a blacklist. So (after
telling my provider about this), I did a quick and dirty hack by
importing mbox messages from the list's web page.

On 2025-12-06 03:50, Joseph Malkevitch wrote:

I've been thinking about this more broadly before. In a sense, every
deterministic election has a kind of "compromising incentive" where you
rank candidates who you're sure other people know about. There might be
some candidate who, if the public saw them listed, would go "now there's
an idea", but who doesn't have the ability to make themselves visible.
Getting yourself elected requires that you can put yourself in the
public's mind.

(This problem is just a lot more severe when there's no formal
coordination process, like the classical countdown example.)

Some nondeterministic methods don't have this problem. Random Ballot is
one of them. If the election is by random ballot, you're free to list
your true favorite as your favorite. If you know someone you think would
do a good job, just put that someone in first. No matter how popular
they are, if your ballot is chosen, they're elected, and if not, you had
no influence anyway.

But Random Ballot is too variable. So perhaps the following would work
as a compromise:

Stage 1: Voters rank or rate a given number of candidates by writing
them down in order (as with the classical countdown).
Stage 2: The method picks five candidates by bloc Condorcet or a house
monotone multiwinner method; and five more candidates by random ballot.
(If this is an election for public office, then some organization or
organizations should probably be given time to provide information about
these candidates -- or a citizen's assembly could do this.) Voters then
approve of subsets of these candidates.
Stage 3: The field is narrowed down to four based on the approval counts
(either by bloc approval or a house monotone PR approval method like
PAV). The voters rank the candidates and a winner is selected by an
ordinal method.

This is probably too cumbersome for something as simple as a classical
music election, but it would reduce the indirect compromise problem. The
first and second phases serve as a particularly unbiased reduction of
the field of candidates, then the second and third phases focus on
determining a quality outcome from that unbiased selection.

There's a related paradox-of-choice problem where the public doesn't
know about truly good candidates. I kinda referred to it with the stage
2 parenthetical, but it's a much harder problem to fix than just the
indirect compromising problem of "there exists some high-quality
alternative that people would agree was good if they saw it".

And Random Ballot might not be the ideal method -- it does have a kind
of center squeeze after all -- but it is a good illustration of the point.

-km

Dear Kris,

WQXR tends to program to program warhorses rather than less well known pieces by well known composers. Relatively recently it has given more exposure to women composers and composers in various minorities.

I think some voters like to vote for pieces that will get played rather than vote for pieces they want to "signal" are "great" music.  Beethoven's Symphony Number 9 regularly comes in Number 1, but I would rather hear one of Beethoven's late string quartets so I vote for some of these but none of these will be played in the Countdown.

Last year if memory serves no pieces by Haydn were played though many of his compositions were nominated. One approach to counting ballots would be to think of composers names like political party names, and tallies for composers be maintained, and be used in some way to play his/her music under some circumstances.

Best wishes,

Joe

From: Kristofer Munsterhjelm km-elmet@munsterhjelm.no
Sent: Thursday, December 11, 2025 7:29 PM
To: Joseph Malkevitch jmalkevitch@york.cuny.edu; Election Methods Mailing List election-methods@electorama.com
Subject: Re: [EM] Classical music countdown balloting

• This email originates from a sender outside of CUNY. Verify the sender before replying or clicking on links and attachments. *

This email originated from km-elmet@munsterhjelm.no, a sender outside of CUNY.  Never send login credentials, financial information, or sensitive information by email. Report suspicious email to reportspam@york.cuny.edu

Let's see if this works. I've been having problems with my mail provider
dropping EM messages due to Dreamhost being on a blacklist. So (after
telling my provider about this), I did a quick and dirty hack by
importing mbox messages from the list's web page.

On 2025-12-06 03:50, Joseph Malkevitch wrote:

I've been thinking about this more broadly before. In a sense, every
deterministic election has a kind of "compromising incentive" where you
rank candidates who you're sure other people know about. There might be
some candidate who, if the public saw them listed, would go "now there's
an idea", but who doesn't have the ability to make themselves visible.
Getting yourself elected requires that you can put yourself in the
public's mind.

(This problem is just a lot more severe when there's no formal
coordination process, like the classical countdown example.)

Some nondeterministic methods don't have this problem. Random Ballot is
one of them. If the election is by random ballot, you're free to list
your true favorite as your favorite. If you know someone you think would
do a good job, just put that someone in first. No matter how popular
they are, if your ballot is chosen, they're elected, and if not, you had
no influence anyway.

But Random Ballot is too variable. So perhaps the following would work
as a compromise:

Stage 1: Voters rank or rate a given number of candidates by writing
them down in order (as with the classical countdown).
Stage 2: The method picks five candidates by bloc Condorcet or a house
monotone multiwinner method; and five more candidates by random ballot.
(If this is an election for public office, then some organization or
organizations should probably be given time to provide information about
these candidates -- or a citizen's assembly could do this.) Voters then
approve of subsets of these candidates.
Stage 3: The field is narrowed down to four based on the approval counts
(either by bloc approval or a house monotone PR approval method like
PAV). The voters rank the candidates and a winner is selected by an
ordinal method.

This is probably too cumbersome for something as simple as a classical
music election, but it would reduce the indirect compromise problem. The
first and second phases serve as a particularly unbiased reduction of
the field of candidates, then the second and third phases focus on
determining a quality outcome from that unbiased selection.

There's a related paradox-of-choice problem where the public doesn't
know about truly good candidates. I kinda referred to it with the stage
2 parenthetical, but it's a much harder problem to fix than just the
indirect compromising problem of "there exists some high-quality
alternative that people would agree was good if they saw it".

And Random Ballot might not be the ideal method -- it does have a kind
of center squeeze after all -- but it is a good illustration of the point.

-km

On 2025-12-12 03:50, Joseph Malkevitch wrote:

Yeah, that's a good example. If everything is a write-in then the pieces
everybody knows about already has an advantage. (If you're going to meet
somebody at a place in town but they were unable to tell you where,
you're more likely to find them at the central station, say.)

That's an interesting solution, a sort of party list for composers.

If you take it further, you could think of a feature landscape of music,
and a PR method as one that tries to find clusters in the same region of
the landscape and pick a "representative" from each. This would
ordinarily work, but when everything's a write-in, it's hard to get the
voters to list a bunch of strings-heavy music, a bunch of complex
mathematical (Bach-like) music, etc. So if you don't want to go the
random ballot way, "party list" might be the best you can do.

E.g. round 1, people name five composers they'd like to hear. There's a
vote, five composers are picked by a PR method. The voters then get to
pick five pieces by each winning composer. There's another PR vote which
picks five winning pieces, which are played.

-km