[EM] RBJ et al.
David L Wetzell
wetzelld at gmail.com
Fri Feb 3 12:17:57 PST 2012
> dlw: When you try out a new piece of technology, you can't expect to get
>> it right right away. A democracy is a function of both the rules and
>> people's habits. If GOPers had seen that their party couldn't win then
>> some of them wd've voted Dem first and the CW wd have won....
> David! That's the point! That's the problem! IRV promised that you
> could vote for your favorite candidate and that would not help elect your
> least favorite.
dlw: They promised it to those who had to vote strategically way too often
with FPTP. They did not promise it was always true.
> it explicitly failed to do that on the second try. In this town that, at
> least 3 years ago, had 3 major parties (so the spoiler wasn't some kinda
> Ron Paul or Ralph Nader gadfly who had no hope of election but could still
> rob victory from the majority candidate). In the context where the 3 (or
> more) candidates are *all* plausible, Condorcet would have elected a
> candidate where, by definition, no other candidate was preferred over this
> CW and, at least in the Burlington 2009 example, would not have suffered
> spoiler, punishment for sincere voting, non-monotonicity, and
dlw: non-monotonicity is not at fault here, unless you expect a large no.
of GOP supporters to have a huge change of heart to support the Prog party
first....Neither was there a problem with summability/transparency...
And how do you know there wouldn't be other foibles that emerge as folks
got adjusted to a Condorcet method?
Perhaps the number of candidates would proliferate so much that it'd be a
At the end of the day, 3-way competitive elections for single-seat
positions are hard to sustain. IRV wd have made the parties around the
true center be the major parties. Now, it seems that won't be the case...
> rbj: It *failed*, David. (but it still beats Plurality and, unfortunately
> the voters of Burlington, who adopted IRV by 65% in 2005, tossed the baby
> out with the bathwater in 2010 and *really* did in 2011 when they rejected
> the 50% threshold.)
dlw: Depends on your loss-function and whether you take a single-period or
multi-period assessment of the outcomes.
I refuse to accept a pass-fail assessment of IRV wrt Burlington. It's not
appropriate. It's playing into the hands of the opponents of electoral
reform by repeating their frames.
> rbj: now, elections are something that we (any particular group of
> people) do not do every day. it's not like you got your iPhone or iPad and
> it worked the day you bought it, and had trouble the second day, but you
> are willing to see how well it works the next day. it's more like a
> high-rise building technique or bridge-building technique (e.g. Tacoma
> Narrows Bridge). if you use some new technique and it fails the first time
> you use it, you better believe there will be hesitation and controversy the
> next time its use is proposed. and very similar if it happens the second
It depends on the severity of the loss. You are exaggerating the practical
bads of the election of a non-CW somewhat left of the CW.
Micronumerosity says we got to not draw strong conclusions from very
limited use of something new. It tells us we need to turn away from our
fallen human natures driven by our fears.
> rbj: on the other hand, if the technique was used 50 times before it
> failed, you would more likely look at the failure as a fluke or outlier.
> elections happen once or twice a year (if you're politically active, if
> you're not it's more like once in four years) and their consequences are
> significant, in some cases worse than a building collapse.
dlw: Once again, assess the "damage" and take the longer view of how this
will play into the next election. If IRV had been continued the Prog
candidate wd have moved to the right some to woo Democrats so the outcome
wd have been preferred by most people.
"a failure that occurs so soon after adoption might very well be an
indication of something systemic, not just an outlier."
dlw: It ain't necessarily so... and you got to consider the relative import
of type one vs type two errors. A sample of type 2 is not going to be
powerful and when you try to make it powerful, you increase the likelihood
of a type one error, ending the use of a good election rule before it had a
chance to prove itself among a populace that understands it better.
>> dlw:To prevent all tactical voting is not the greatest good.
> The *primary* reason for adopting ranked-choice voting, the greatest good
> promised, is to remove the *burden* of tactical voting from voters so that
> they do not experience voter's regret the day after the election (which,
> here in Burlington, soured many voters that do not return to the polls,
> thus reducing participation in democracy). i don't suggest that we can
> prevent all tactical voting, but the common burden of tactical voting, the
> tactic called "compromising", is avoidable and *should* be avoided where at
> all possible.
Think about it. Really? Preventing anyone from being pressured to
tactical voting is the greatest good? Shouldn't it be to make the parties
responsive to the general views of the population? To reduce the distance
between the de facto and true political center?
I don't have a problem if a major party chooses to get ideologically stuck
so some of its supporters have to abandon it because of its
In our context where $peech is so strong the "tactical voters" are more
likely to be the ones who've been gaming the system for their own bottom
line for quite some time. It isn't the same thing for them to be pressured
to vote insincerely as it is when third party dissenters from "dumb and
dumber" get pressured to vote that way. The former bonds the de facto and
true center. The latter severs the two.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Election-Methods