Strategy thoughts:
Assuming as candidates, Good, Soso, and lice: My preference is G but
S is better than any lice. Thus I desire to vote for both G and S
with G preferred.
Plurality - can not vote for both. On days when I expect G to
certainly lose I vote for S to protect, as best I can, against lice.
Approval - can vote for both but this can cause G to lose. Simple
rules and a bit better than plurality.
IRV - can vote for both. Vote counting is both much labor and can
fail to elect G even though best liked, if this is not seen by the way
the counters look at the ballots.
Range - can vote for both. After giving G top rating, S has a
strategy headache: Rate S high and risk S winning over G; rate S low
and risk S losing to lice.
Condorcet - can vote for both and show clear preference for G over S.
On Aug 15, 2011, at 9:20 PM, Jameson Quinn wrote:
2011/8/15 Jonathan Lundell <[email protected]>
On Aug 15, 2011, at 11:58 AM, Jameson Quinn wrote:
It's true that I might agree to a statement if all it said were "We
believe that approval is marginally superior to plurality" (thought
to the extent that I agreed, I don't think it's enough better to
merit any energy in advocating it). But that's not what you're
proposing. Is it?
No. I'm proposing saying that, in different words, along with a
number of other things with which you haven't disagreed. Including
that we believe that approval is a step towards systems which we
see as significantly superior to plurality. (Remember - just as
approval is 2-level Range, approval is also 2-level Schulze or what
have you, and also no-intercandidate-preference SODA, etc.) So,
either propose some specific change in the language relating to
approval, or bring some other objection, or both.
The statement says, in effect, "Range is good, IRV is bad". I
disagree.
Perhaps I'm the only one, in which case it's inconsequential that
I'm not aboard.
(What Schulze are we talking about? I associate the name with a
Condorcet-cycle-breaking method.)
It doesn't say that. It says, we can agree that range is at least
marginally better than plurality, we cannot agree on that for IRV. I
would happily sign a separate statement saying IRV is better than
plurality, but I think including that here would lose too many.
Schulze is just my default example of a complex but good Condorcet
tiebreaker. And if you run it with only two-level ballots, it is
equivalent to approval.
If you want to suggest rewording to make it clear that you're only
giving the weakest possible endorsement to Range, then go ahead. But
remember, any amount you weaken the "these are good systems"
section, weakens it for all of the listed systems. Because we are
not going to get many people to sign on to a statement that makes
distinctions between those systems.
Or say clearly that you can't sign the statement in any form, and
we'll stop worrying about you. I want this to get as much support as
possible, but I know that I'll never get everyone.
Again, I personally agree with much of what you are saying. Approval
does force strategic thinking on the voter, more than many other
options. (That's also true of Range, but not of MJ, so you shouldn't
generalize to "rating systems".) But this is not about just me.
JQ
----
Election-Methods mailing list - see http://electorama.com/em for list info
