[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Condorcet Voting and Supermajorities (Re: [CONSTITUTIONAL AMENDMENT] Disambiguation of 4.1.5)

On Fri, Dec 08, 2000 at 04:52:38PM +1000, Anthony Towns wrote:
> > > You're mixing and matching what you apply the word "preference"
> > > to. The option ranked first is more important than the others
> > > because the voter has expressed that it's preferred to all the
> > > others. That is the following preferences are expressed:
> > > 	B > A
> > > 	B > S
> > > The option ranked second, however only gets:
> > > 	A > S
> > > But "A > S" counts just as much as "B > S" or "B > A". "A"
> > > however, doesn't count as much as "B", because, well, "B > A".

On Fri, Dec 08, 2000 at 01:04:42PM -0500, Raul Miller wrote:
> > I don't see that you can say "count as much" without assuming a
> > counting methodology. If you're going to use this argument as
> > an argument for that methodology you're introducing a circular
> > argument. The way I see it, first preference is more important to
> > the voter than the other preferences because, well, the voter rated
> > it as the first preference.

On Sat, Dec 09, 2000 at 01:14:34PM +1000, Anthony Towns wrote:
> Again, you're mixing terms. What are you trying to say? That by being
> ranked first, "B" should be treated as more important than "A"? Then,
> sure, I agree, and I just explained how Condorcet schemes deal with
> that.

Of course I'm saying that.  And, I know how the Condorcet schemes deal
with that.

> If you're trying to say that the preference "B is better than A" is
> much more important than the preference "A is better than S", then I
> think you're making things up as you go, and that they're not very
> good things either.

Once again: the 1st preference vote is more important than the other

> There are schemes that'll work that way. You could invert a Borda
> counting method, for example, and get something that works that way.
> Condorcet, however doesn't. The constitution doesn't, either.

Your "the constitution doesn't" requires that you use an interpretation
where some number of potential votes are undecidable.  Correct?

> > The way I see it, the second (etc.) preferences are only to be used
> > when there's an ambiguity in the cumulative impact of the first
> > preferences.
> In which case you'd be wrong.

Counter example?  [Please limit this to something which fits within
the constraints of the constitution.  I'm fully aware that you can pick
arbitrary voting systems with arbitrary properties.]

> > > > > Please rate your preferences for the final form of the draft resolution:
> > > > >       [ _ ] P
> > > > >       [ _ ] P+A
> > > > >       [ _ ] P+B
> > > > >       [ _ ] P+A+B
> > > > >       [ _ ] P+A+C
> > > > >       [ _ ] P+B+C
> > > > >       [ _ ] Further Discussion
> > > > >
> > > > > Should P be the final form of the draft resolution, please rate your
> > > > > preferences for its acceptance:
> > > > >       [ _ ] Yes
> > > > >       [ _ ] No
> > > > >       [ _ ] Further Discussion
> > > > >
> > > > > Should P+A be the fin...
> > > > > (etc)
> > > > This looks like one amendment ballot and six other ballots, one which
> > > > would be a final ballot, and five which would be neither amendment
> > > > ballots nor final ballots.  
> > > Not exactly. It's one amendment ballot and the final ballot repeated six
> > > times, one for each different form the final form of the draft resolution
> > > can take so that each voter can "vote differently in the final ballot
> > > for each of the possible forms of the final draft resolution".
> > Eh?  You're saying this is a single final ballot?
> > That's not one ballot, it's six.
> It is the same ballot, repeated six times. Is that really difficult to
> understand?

They're not the same ballot because they don't have the same options
on them.

Different ballots have different options on them.  Different instances
of the same ballot would have the same options on them.  These ballots
are similar, but they're not the same.

Uh.. you do understand the distinction between "similar" and "the same",

> I'm also failing to see anything particularly unfair or worrying about it.

I'm concerned about its unnecessary complexity.

> It's *exactly* equivalent to having N+1 separate ballots (spaced out
> in time), except that it's quicker (and doesn't give people time to
> change their minds between votes). Every outcome that's possible is
> the same, each of the ballots listed is in a form exactly specified
> in the constitution. There's no ambiguities that have to be carefully
> worked around.

That's not what I see.  

First off, the internal conflicts in what you're saying: I see one
amendment ballot, and six final ballots.  N+1, where N=1 is 2.  You've got
seven ballots.  To space them out in time, in exactly the same fashion,
you'd have to have seven votes.

Second off, the inconsistencies between your interpretation and the
constitution.  These aren't as blatent -- essentially, you've added
constraints which are not present in the constitution:

[a] You've decided that the form of the final ballot, specified in
A.3(2), when there's a single draft resolution to be considered, is the
only form possible for a final ballot.  Except, since it's physically
impossible to live within this constraint and still respect A.3(3),
you've invented this idea that some number (greater than 1) of distinct
ballots are really one final ballot.

[b] You've also decided that the clause "may arrange these ballots to be
held simultaneously, even (for example) using a single voting message"
says something about the form of the ballots, and therefore of that
single voting message.  Yet that clause is deliberately ambiguous as to
the form of the ballot.

What I see is that this part of the constitution specifies the form of
certain ballots for certain conditions (the simple cases), then adds
on a general clause for doing things more expediently.  In particular,
it talks about combining ballots into a single voting message -- here,
it gives some requirements of the message, but doesn't specify its form.

I do see that this flexibility allows you to come up with the
interpretation you've been presenting.

> > > > Also.. the constitution does specify that the user be able to vote
> > > > differently >>in the final ballot<< for each of the forms of final draft
> > > > resolution.  You're only allowing the user to vote on one form of the
> > > > final draft resolution in the final ballot.
> > > 
> > > *Oh*, is that how you're reading this?
> > > 
> > > You seem to be saying that should be read as:
> > > 
> > > 	``In the final ballot, each voter must be able to vote for each of
> > > 	  the possible forms of the final draft resolution.''
> > > 
> > > whereas I'm saying it should be read as, ummm:
> > > 
> > > 	``If the amendment and final ballots are combined, then there are
> > > 	  multiple forms of the final resolution that are possible. For
> > > 	  each of these, each voter must be able to express a *different*
> > > 	  preference for the options in the final vote (the Y/N/F one),
> > > 	  even though it won't be clear which of these preferences will
> > > 	  be used when the voter votes.''
> > > 
> > > Can you see how I'm breaking that clause up to read it that way?
> > To some degree.  Can you see that my interpretation is a whole lot
> > simpler?
> Not really. To make your interpretation work you have to:
> 	* invent which options get included on the final ballot (since
> 	  while to constitution explicitly describes what options are
> 	  mean to be on every other ballot its possible to case, it
> 	  somehow leaves this one out)

Not really.  The options on the final ballot, in my interpretation,
are the possible final forms of the draft resolution, plus "No" and
"Further Discussion".  All of these clearly mentioned in the constitution.

> 	* treat the words "Yes" "No" and "Further Discussion" as variables
> 	  rather than explicitly stated options present on the ballot

How is "yes on proposal plus option A" in my interpertation any more of
a variable than "yes on proposal plus option A" in your interprertation?

How is "no" a variable?

How is "further discussion" a variable?

>	* treat the different language "ballot" and "single voting
>	  message" as though they were synonymous, even though the
>         constitution specifically uses different words for them

False.  In my interpertation, "single voting message" may represent ballot
or ballots.  For certain kinds of votes (several distinct amendment votes
at the same time), it's quite reasonable to present several ballots in
the same voting message.  And, indeed, if we're talking about a situation
with, say, 10 independent options, it makes sense to narrow down the
possibilities before holding the final vote.

My interpretation is that "single voting message" doesn't distinguish
between "ballot" or "ballots".  The form of the voting message is not
specified in that fashion.

> > > You seem to be joing the ``to vote'' with the ``for each'' to make
> > > it ``to vote for'', while I'm treating ``vote differently'' to stand
> > > alone, and the ``for each'' to mean they get to ``vote'' many times,
> > > each of which may well be ``different'', and what you're voting for or
> > > against is only mentioned in the previous clause.
> > Um.. ok, you can describe the way I'm reading "to vote for each" as "'to'
> > 'vote for' 'each'".  [Or, as "'to vote' 'for each'".]
> >
> > However, although my interpretation allows "to vote for each" to be taken
> > as "to vote differently for each", I still don't really understand...
> Yeesh. Your interpretation:
> 	The vote take or the voters may arrange for these ballots to be
> 	held simultaneously, even using a single [ballot]. 
> 	If amendment ballot(s) and the final ballot are combined in this
> 	way THEN
> 		it must be possible for a voter TO VOTE FOR 
> 			each of the possible final forms of the final draft 
> 			resolution
> 		IN the final ballot.
> That is "each of the final forms" is what you're "voting for".
> My interpretation:
> 	The vote take or the voters may arrange for these ballots to be
> 	held simultaneously, even using a single voting message (although
> 	that ballots are still separate).
> 	If amendment ballot(s) and the final ballot are combined in this
> 	way THEN
> 		FOR EACH of the possible final forms of the final draft 
> 		resolution
> 			it must be possible for a voter TO VOTE DIFFERENTLY
> 			IN the final ballot.

What I don't get here is -- if you're going to take the plurality of words
as implying constraints -- where nothing that's not explicitly permitted
is denied -- what do you call those "final ballots which never become
final".  Why isn't there some mention of "alternative final ballots"?

> Honestly, I think that's a more straightforward reading of the
> constitution (you'll note I only had to rearrange the last clause to
> get it into a vague pseuo-codish nature, rather than having to take
> a clause out of the middle of a split infinitive (?) or so), and it
> involves minimal creativity on the vote taker's part, both of which I
> consider good things.

Yes, I do notice that you changed th wording of the specification language
to turn it into psuedo-code.

I don't see that mine requires any more creativity on the vote takers

> Further, you ocassionally assert that the constitution is (or should
> be) stating properties of how the vote should be handled, not an
> algorithm. I think this is completely mistaken: the constitution
> should be (and is) describing exactly how decisions are made, so that
> there's absolutely no room for doubt, or (accusations of) personal
> bias, to slip in and muddy the waters.

[Sarcasticly:] Don't you think the constitution would have mentioned
"alternative final ballots" if that's what it meant?  Instead, it says
"on the final ballot".

I can agree that the constitution, as stated, is ambiguous -- that
it allows for multiple interpertations where you'd like there to
be a single interperation.  But just wishing isn't going to make
those other interpertations invalid -- that would require changing
the constitution.

> > Well, let me just ask you this:  Do you still think that my interpretation
> > of this clause is incorrect.
> I'm not sure; I *think* it's valid solely from a parsing point of
> view; but I don't think it's as straightforward, and I don't think
> it's at all desirable to then have to invent or generalise other
> procedures to handle this interpretation.

Heh, that's pretty much what I think of yours.

I suppose that means this conversation is near close.

If you agree, would you like to summarize what we've managed to agree
on, for people in debian to refer back to (just in case there are
non-masochists who don't have the time to plow through every exciting
word of our exchange)?  [And, if you only partially agree, maybe put
that aspect in a different message?]

Of course, we can hold off on summarizing for a bit longer.  And, if
you'd prefer, I can write the summary and let you point out my mistakes.



Reply to: