[EM] Hitler-Stalin-Middle Example Again

Mike Ossipoff dfb at bbs.cruzio.com
Fri Mar 1 07:01:21 PST 1996

DEMOREP1 at aol.com writes:

[Rob said most of what I was going to, so I'll just add a few things--Mike]

> Condorcet fans seem unable to comprehend that with 2 of 3 candidates being
> "extremists" on a ballot that each "middle" voter will be in a tough
> situation about what to do about their second choice vote (if any).

With Condorcet's method:

First of all, unless order-reversal cheating is attempted on a scale
sufficient to change the election result (improbable, due to its risk),
there's no penalty for Middle voters ranking a 2nd choice. Doing so can't
count against Middle or help anyone beat him.

But with Approval, Middle voters could be giving the election away
if they approve an extreme candidate.

Besides, with just 3 candidates, with Condorcet's method, Middle voters
don't really have a need to vote a 2nd choice anyway: If Middle isn't
Condorcet winner, then either Hitler or Stalin has an outright majority,
in which case it makes no difference if Middle voters vote a 2nd choice.
If, on the other hand, Middle is Condorcet winner, then it's the extremes
who should vote for Middle. Because Middle is the rightful winner, the
Condorcet winner with majorities over the others, and because a Hitler
victory won't bother Middle voters as much as Stalin voters, the Stalin
voters, if they believe that Hitler might have more 1st choice support
than Stalin, know that the Middle voters are in a better position to
defect than are the Stalin voters and that they will do so if they
have any reason to (the liklihood of large-scale order-reversal would
give Middle voters a reason not to vote a 2nd choice).

To summarize: 1) Unless order-reversal cheating is attempted on a
scale sufficient to change the election result, there's no penalty
for Middle voters voting a 2nd choice--it can't take away Middle's
victory. 2) If order-reversal is likely to be attempted on a large
scale, Middle voters' strategy is to not vote a 2nd choice. This makes
order-reversal backfire badly, every time. And the fact that Middle
voters can be expected to not vote a 2nd choice under those conditions
also means that the smaller extreme would be making a bad mistake to
not rank Middle in 2nd place.

I've mentioned strategy in the preveious paragraph, so I'll re-emphasize
that Condorcet's method is the only method that never requires defensive
strategy under plausible conditions--it never requires defensive strategy
unless order-reversal cheating is attempted on a scale sufficient to
change the eleciton result. And, in a full-size public election, it
won't happen--Hitler voters won't rank Stalin 2nd, knowing that it will
probably elect Stalin.

> A middle voter may hate one extremist by "absolutely" minus 99.9 percent and
> hate the other extremist by minus 100 percent- a "relative" differance of 0.1
> percent. 
> If the middle voter votes for either extremist as a second choice, then such
> extremist may get elected with a majority and claim he/she has a "mandate"
> (to go politically crazy).

You mean in Approval, right? Because Approval is the method that has the
problem that you describe. Condorcet's method doesn't have that problem,
as I've just been discussing in this letter (and as I've repeatedly
shown you in previous letters).

> If the middle voter does not vote for either, then one of the extremists may
> get elected with a plurality.

If the opposite extreme voters make a bad mistaake--a completely 
unnecessary and pointless mistake. As has been said, the extreme
voters have no reason to not vote for Middle in 2nd place, and at
least 1 extreme has very good reason to do so.

> Since all middle voters will probably not be alike, then each of them may
> strategically ask what are the other middle voters doing ?
> Should I vote for an extremist or not ?

No. There's no need to. But doing so is harmless under plausible
> An example again (50 votes for majority)--
> Hitler         Stalin            Middle
>   42              41                  16    First Choice votes
> Does any voter dare vote for a second choice ?

As I've said, and as has been previously pointed out, the extreme
voters have absolutely no reason to not vote a 2nd chocie. The Stalin
voters have an extremely good reason to vote a 2nd choice

> This is the real world. Russian voters in June will very likely nominate 2
> extremists for Russia President putting Russian "middle" voters into the
> above situation in the runoff election (with potential world problems in view
> of the President-is-dictator language in the Russia Constitution).

If the Russian voters eliminate the middle Condorcet winner & get
an extremist, that's because of the defects of Runoff (& MPV), but
what does that have to do with Condorcet's method?

> The approval voting remedy remains to encourage the voters for the extremists
> to vote also for a middle candidate who hopefully could get a majority.

As Rob pointed out, it's the exact opposite. Approval gives the extreme
voters reason to not vote for Middle, if there's a chance that their
favorite could otherwise have a win. If their favorite has a win, & they
vote for Middle, they could be giving the election away to Middle, with
Approval. As I said, with Condorcet's method, the extreme voters have
no reason not to vote for Middle.

You've been naming Approval's problems, & attributing them to Condorcet's


I have to agree, Demorep, that you haven't been completely fair with
us. We've taken your letters seriously enough to take the time to carefully
answer them, but you haven't shown similar respect, and have, instead,
ignored replies to your statements.

> Hitler         Stalin            Middle
>   45              46                  51    Approval votes
> It is directly due to the gerrymander indirect minority rule of legislative
> bodies in the U.S. that there is a major degree of power madness surrounding
> the various chief executive officer positions in the U.S. - U.S. President,
> Governors, Mayors, etc. 
> If no reform comes to U.S. legislative bodies, then U.S. Civil War II is

Because reform is necessary, we've got to seriously & honestly discuss
what reform to propose to the public.

> highly likely (as in the last years of the Roman Republic with civil wars
> between Sulla and Marius, Pompey and Julius Caesar and finally Antony and
> Octavian (later Emperor Augustus Caesar)).
> Witness the "personal" mudslinging among the Republican party candidates for
> U.S. President- Can violence be coming between Republican candidates ?
> Between Republican and Democrat presidential candidates ?

> .-

Because there are so many possible standards & so many possible methods,
it's necessary to have an orderly approach. I suggest that we first
discuss standards. If we all agree on standards, then the next part
is the easy part: demonstrating which method meets the standard that
we've agreed on.

I've already said what standards I consider important: Getting rid
of the lesser-of-2-evils problem & protecting majority rule, &
eliminating the need for defensive strategy. I've told why Condorcet's
method meets these standards. Does anyone disagree with those standards?
Does anyone propose another standard that he believes is more important?


Mike Ossipoff



More information about the Election-Methods mailing list