[EM] Replying to Demorep

Mike Ossipoff dfb at bbs.cruzio.com
Mon Mar 4 02:50:38 PST 1996

DEMOREP1 at aol.com writes:
> My standard for a single winner office is whether a majority of the voters
> absolutely (and not relatively) want a candidate to be elected to that
> office.

Fair enough. If you believe in a standard that doesn't include letting
voters express more than a 2-level preference ordering, then by your
standard, methods like that are fine. I don't believe you'll find many
electoral reformers who feel that way.

> Back to the Condorcet examples--
> A. Middle is lowest (original example)
> Hitler         Stalin            Middle
>    42              41                  16    First Choice votes
> B. Middle is middle
> Hitler         Stalin            Middle
>     16               42                  41     First Choice votes
> C. Middle is highest
> Hitler         Stalin            Middle
>      41            16                   42    First Choice votes
> Such 3 examples assume that one of the candidates has a relatively small
> percentage of the total votes.
> D. All candidates are very close with around 33 First Choice votes each.
> There will obviously be polling before each election.

There's better be, because Approval makes voters especially dependent
on reliable polling, compared to Condorcet's method.

> Would every voter's second choice within each first choice group be the same
> in every example regardless of the number of votes for his/her first choice
> (as estimated by polling)? Obviously not.

Obviously not with Approval. Obviously yes with Condorcet's method unless
there's danger of order-reversal on a scale suffincient to change the
election result. Condorcet's method doesn't force the voter to figure
out strategic voting under plausible conditions, as Approval does.

> The candidates will obviously attempt to influence the second choice votes of
> their supporters for strategic purposes --- especially (a) to get Condorcet

Again, yes that's true in Approval. In Condorcet's method it's doubtful
that voters or candidates would want to risk order-reversal. And nothing
could ever be gained by truncation, in Condorcet's method. Another thing,
you speak of organized campaigns to promote strategic voting, but if
Buchanan's election committee is widely publicizing a plan to attempt
order-reversal cheating, do you believe that the Clinton voters wouldn't
hear about it? Since the Clinton voters would know about it, they'd
know not to rank Buchanan--the defense against likely order-reversal
by his voters. The order-reversal would then only succeed in electing
Nader. It would backfire. It wouldn't be attempted.

> circular results (Hitler beats Stalin, Stalin beats Middle, Middle beats
> Hitler or the three reversed cases) based on second choice votes that may be
> only slightly less than a minus 100 percent total disapproval and (b) to
> force Condorcet tie breakers. 

As I said, such advice to voters would be a very bad idea.

> I leave it to folks to think of examples of circular results and the fewest
> number of votes needed to elect under the Condorcet tie breaker (especially

You leave it to us? Instead of specifying an example to demonstrate your
claim? I don't know what you mean by fewest numbers needed to elect. If
you're implying a new alleged problem, you need to specify what it is.

> with the D. 33 votes each example).
> Something called "order-reversal cheating" has been mentioned.
> Can there can be "cheating" in election reforms ???

Most definitely. Gibbard & Satterthwaite demonstrated that any reasonable
single-winner method can have situations where voters can benefit from
using strategy. This was mentioned in the _Scientific American_ article
that you cited.

In terms of what's important to voters, maybe a better way to put it
would be to say that there's no method that completely gets rid of
any possibility of need for defensive strategy under every conceivable
set of conditions.

What we can do is reduce that need. For instance, though Condorcet's
method only requires defensive strategy when large numbers of people
attempt risky & devious offensive strategy, Approval requires defensive
strategy pretty much all the time! If that's how you want it, then
for you Approval might be what you want. By the standards of interest
to electoral reformers, Condorcet's method is better.

And, as I've said, even when Condorcet's method does need defensive
strategy, it never requires listing at the same rank-position 2 or
more candidates that the voters doesn't really consider equal. It
certainly never requires you to rank anything else equal to your
favorite unless you really consider it equal. Does Approval often
require that? Yes.

> Here it is worth noting that with C number of candidates there will be ((C x
> C)-C) /2 pairs of candidates
> C        Pairs
> 2          1
> 3          3
> 4          6
> 5          10
> 6          15
> 7          21
> For visual purposes divide an expandable checkerboard along one diagonal. The
> squares on half the checkboard minus the diagonal are the number of pairs.

Why are you pointing that out? With a computer counting the election
the pariwise comparisons can all be easily dealt with.

> How many "serious" candidates will be running for U.S. President (or other
> major executive office) with a single winner method ?

I  don't know. Surely more than now, depending on how good the method

> What will be the probability of multiple Condorcet circular ties with
> multiple candidates ?

So what? As I've repeatedly told you, when the electorate's collective
preferences are circular, then the election has a problem, regardless
of what method is used, if the collective prefernces are sincerely

And I've also carefully showed you, repeatedly, that, circular tie or
no circular tie, a candidate with a majority against him can't win
in Condorcet's method unless every candidate has a majority against
him, something possible only under extreme &/or implausible conditions.

> How many candidates will be elected with de facto majority disapproval due to
> the results of Condorcet tie breakers ?

That depends on what you mean by "de facto majority disapproval". I've just
told you how Condorcet's method denies victory to majority-rejected
canddiates. With the addition of NOTB, or a disapproval count, used
in the same balloting as the rankings, the voters would have a further
way of expressing disapproval.

But you never did answer me about "disapproval". What if you disapprove
of Clinton, but you know that Buchanan, whom you also disapprove, and
whom you know to be considerably worse than Clinton, will win unless
you "approve" Clinton, in an Approval election? Do you refuse to 
"approve" Clinton, because you don't really approve of him, or do you
"approve" him in order to defeat the worse Buchanan? And then what
if it turns out that your favorite would have had a win had you not
"approved" anyone else? Can you understand why we don't consider
Approval the best method?? Will you ever answer the replies to your

> No "so what" answers needed.
> Thus, approval voting remedy remains. Each voter may approve one or more
> candidates for the office using multiple same choices (0, 1, 2, etc.). If 2

Whoa. Are you talking about Approval, or that other method that you
never did give us a real definition of?

> or more candidates get a majority, then Condorcet may be used for the
> tie-breaker (or just declare elected the candidate with the highest majority
> depending on whether computer voting is available).
> To a major extent, the whole discussion about single winner methods is
> somewhat irrelevant if (1) legislative bodies are being elected with a
> majority rule proportional representation system, (2) executive officers are
> elected with short terms of office, (3) all officers can be recalled and (4)
> there is a major constitutional reform allowing ordinary citizens to enforce
> the laws directly.

If you believe that single-winner reform is irrelevant, that would explain
why you won't bother to discuss replies to the claims that you make, and
you won't bother to discuss it responsibly. Thank you for explaining that.

> Regarding (2), how many "controversial" U.S. Presidents (e.g. Mr. Johnson
> during the Vietnam War, Mr. Nixon during Watergate, etc.) would have won 4
> consecutive annual elections ?

They can do a lot of damage before they're finally out, whether due
to the next regular election, or whether due to recall. Besides, you
don't understand the problem with recall. If we use a method like
Plurality or Approval that makes us have to vote for a lesser-evil,
and he betrays us after he's elected, the progressive minority who
made it possible for him to win, with their "approval", isn't numerous
enough to recall him, unless he also offends much larger groups of

> Regarding (4), the use of "executive power" in the U.S. and State
> constitutions has been generally interpreted by the courts to mean that
> ordinary citizens lack standing to enforce "public" type laws - especially
> the laws regulating the executive branch. 
> If the laws regulate the activities of the executive branch (i.e. telling
> executive officers absolutely or conditionally to do or not to do something),
> then there is a continuous constitutional crisis if the executive officers
> ignore such laws.
> Witness especially the continuous crisis over activities of the U.S.
> executive branch in foreign countries- Korean War, Vietnam War, Bosnia, etc.
> etc. etc.
> Thus, single winner election reform should be part of a constitutional reform
> package.
> .-


More information about the Election-Methods mailing list