[EM] No Objection to NOTB or disapproval count

Mike Ossipoff dfb at bbs.cruzio.com
Sat Mar 2 19:05:34 PST 1996

I have absolutely nothing against Steve's suggestion of NOTB, or
Lucien's disapproval count. Neither of those things would interfere
with Condorcet's method's properties. Of those 2 proposals, I prefer
NOTB, because it may have a more precisely-defined meaning: 
NOTB is an alternative in the election, the alternative of rejecting
all the candidates & holding a new election with new candidates.

But, in that case, are you sure that any candidate beaten by NOTB
should be rejected? In a circular tie every candidate is beaten by
some alternative (candidate or NOTB). Why should NOTB have the
privileged status of eliminating anyone it beats? If NOTB were
included, I'd want it to have no more status than the candidates

Of course if you cast a "disapprove" vote against Buchanan, that
means that you'd rather have another election than let Buchanan
win. But Lucien would require a majority who say that, to disqualify
Buchanan, while Steve would merely require that NOTB win, even if
a full majority don't rank it over Buchanan. But, in a way, requiring
NOTB to win in order to disqualify anyone is a tougher requirement
than disqualifying Buchanan if a majority disapprove him (prefer
NOTB to him), because Steve's proposal requires that NOTB win the
whole election before it disqualifies anyone.

I have to admit that I don't know which of those 2 proposals is
better. I guess they both have equally well-defined meanings, but
which one I'd like best would depend on which one would give me
the best opportunity to disqualify a candidate whom I really
dislike. It isn't clear to me whether that's easier by getting
a full majority to dispprove him, or by getting NOTB to win.

But, actually, NOTB can disqualify a candidate even if NOTB doesn't
win: All the people who don't like candidate X, even if they have
nothing else in common, and wouldn't all rank the same candidate
over candidate X, can come together by all ranking NOTB over
candidate X, and this will effective prevent X from being the
least beaten candidate. And it doesn't take a full majority--it
just takes more people than the number by which some other candidate
is beaten. So X is out, even if NOTB doesn't win. And if NOTB beats
everyone, or even if it's otherwise the winning alternative, it
makes sense to hold another election, since voters knew that that's
what NOTB means. So it seems to me that NOTB is easier to disqualify
a disliked candidate with, and so, of those 2 disapproval options,
I'd prefer NOTB.

And no, it isn't that I believe that NOTB would be unlikely
to win. But I do believe that, though I have no objection to NOTB
& the disapproval count, and though I'd like to make use of them in
a Condorcet election, they aren't really necessary.

You said that Condorcet allows a candidate disapproved by a majority
to be declared the winner. True, if we define "disapproved" as "liked
less than a new election with new candidates". Conceivably there could
be a natural circular tie in which someone is the least beaten by
any 1 candidate nevertheless has a majority who have ranked NOTB
over him. But please note that, in that case, the guy is "majority
rejected", he has another alternative ranked over him by a full majority
of all the voters. (that alternative being NOTB)

That means, then, that in order for that candidate to win, _every_
alternative, including every candidate & NOTB, must also be majority-
rejected, by having something else ranked over it/him/her by a full
majority of all the voters. So, Condorcet's method makes it difficult
for a majority-rejected candidate to win, since in order for everyone/
everything to be majority-rejected requires either order-reversal or
an extreme form of natural circular tie--a chaotic & indecisive
situation. That's why I say that NOTB & the disapproval count aren't
really _necessary_, since Condorcet's method does a good job of
defeatng majority-rejected candidates, as-is.

Still, NOTB would be useful, as a gathering place for everyone who
wants to defeat candidate X, even if they don't agree on exactly whom
they like more than him. So I agree that including NOTB would be
a good thing.

That brings me to "dumbed-down". I've proposed Condorcet's method
in more than 1 form. I proposed in in a form that, in the event of
a circular tie, picks the winner from that circular tie. Everyone
rejected that as being too complicated, because the definition of
a circular tie, of which candidates it includes, was the last straw
that made people throw up their hands and say "Forget it! This is
too complicated. You'll have to come up with something simpler."

So I did. I began proposing a version of Condorcet's method, which
I'll call "Plain Condorcet's method", which, when there's a circular
tie, doesn't limit its choice to the members of that tie. As soon as
I began posting that in the newsgroups, I began getting letters 
saying that I _had_ found an adequately simple method.

You, you'll notice that my definition of Condorcet's method doesn't
say anything about limiting the choice to the members of the circular
tie. This avoids the (unacceptably complicated, to most people) 
definition of who is in the circular tie.

You can call that dumbing it down, but if something is rejected by
the public, it doesn't matter how good it is. The Single-Winner
Committee's task is to find the best method to propose to the

So then, in that same spirit, any additional rules, or paragraphs,
or even sentences, in the definition of a proposed method is very
likely to be that last straw that makes a member of the public reject
that proposal, even if he/she would have accepted it had it not been
for that 1 sentence that was 1 sentence too many.

My experience has been that _extreme simplicity_ is absolutely
necessary. So, though I personally like NOTB, and would like to
have it, to use against Clinton, and anyone worse, I have to 
recognize that what's best has to be tempered with what's simple
enough for the public to accept. In this respect, adding NOTB
as an alternative is simpler, in terms of rules, than is the
dispapproval count, which adds rules to the balloting & count.

Still, the addition of NOTB requires an extra paragraph & sentence,
so I have some trepidation about adding even that to a 1st proposal
for Condorcet's method. On the other hand, given voters' cynicism,
maybe if they like it that could make up for it's complication of
the proposal. We'll have to try Condorcet, with & without NOTB
on some sample people. 


But everyone I've talked to likes the idea of rank-balloting, and
immediately understands the desirability of expressing preferences
among several candidates via a ranking. So there's no case for saying
that rank-balloting is too complicated for voters.

On the contrary, it's Approval that I get resistance to, from people
who believe that it violates 1-person-1-vote. Even those who don't
object to it have never expressed any enthusiasm for it. People are
usually enthusiastic about rank-balloting when told about it.
Rank-balloting will easily win when it's offered to people.

I'm not criticising Approval; it would be a big improvement over
Plurality. But we're looking for the _best_ proposal, in terms
of merit & public acceptance.


Mike Ossipoff


More information about the Election-Methods mailing list