IRO Arguments & Rebuttals

Mike Ositoff ntk at
Mon Jul 27 01:52:34 PDT 1998

Jim Lindsay once asked me to relay some of his arguments regarding
IRO to the members of the EM list. When we were discussing IRO,
I did relay his arguments. But having asked me to do that, he
surely wouldn't mind if I relay some of them here too. Because
IRO is being aggressively promoted, and a CVD poll has indicated
that its members are interested in single-winner reform. If they're
interested in reform, then surely they're interested in the
merits of a widely-promoted reform that's being offered to them.

Jim said that pairwise-count methods (such as the ones that
EM has recommended here) are for people who like compromise.
What? American voters like compromise. Ask a voter why he voted
Democrat, and he'll say that it was only to keep someone worse
from winning. So if Condorcet is for people who like compromise,
then it's for American voters.

And, in fact, do some people who disdain compromise actually
vote Democrat themselves, even if that isn't their favorite?

And compromise is the basis of rank-balloting: Being able to
list your lower compromise choices while still voting your favorite
in 1st place. That's really the whole reason why most of us want
single-winner reform.

And, in any case, shouldn't it be up to the voter to decide
whether or not he wants to compromise?

Jim asked for empirical evidence or hard data, showing that
IRO is regarded as having a lesser-of-2-evils problem. Steve
furnished that evidence by pointing out there are usually onlyk
2 candidates in the Australian elections that use IRO. Steve
also pointed out that Roberts Rules of Order recognize that IRO
has that problem, and advise against its use when something better
can be used.

Majority rule? IRO advocates have claimed that IRO carries out
majority rule when it lets the majority get its way in the
choice between the last 2 candidates, after eliminating everyone
else (!) Never mind that a majority preferred one of the eliminated
alternatives to both of the remaining ones :-)

What if I took you to a restaurant, and gave you the menu, and
said "Get whatever you want". You point to something in the menu
& say "I'd like that." I say "You can't have that. Which of these
2 would you prefer." That's how IRO treats majority wishes.


One IRO advocate told me that he didn't believe that IRO's
"Condorcet problems" would matter in the real world. I guess
he was over-ruling the academic authors and the many others who
_do_ believe that Condorcet's criterion is important, and that
it's important to elect a Condorcet winner.

When there's a candidate for whom the voters have expressed
preference for, over each one of the other candidates, failing
to elect that candidate is a clear violation of expressed public

Suppose we did a rank-balloting election for President, and
counted the ballots by IRO & by Condorcet(EM). And then we
hold a public election between the winners by those 2
methods. When Condorcet(EM) chooses the Condorcet winner
(the candidate who'd beat each one of the others in separate
2-way elections), and IRO doesn't, then guess which method's
winner would win that 2nd vote?

In fact, in an early list that was a precurser of EM, we had
to vote on a multi-alternative issue. I suggested a procedure
like that in the previous paragraph, and Jim accused me of 
proposing something that was essentially the same as just
using Condorcet.

When there is a Condorcet winner, a candidate who'd beat each
one of the others in separate 2-way elections, IRO is the
only rank-balloting method that can fail to elect him even
if everyone ranks him alone in 1st or 2nd place.

The methods recommended here by EM, on the other hand, were
chosen because of their remarkable abililty to protect the win
of a Condorcet winner, and to protect majority wishes.


Non-monotonicity. Here's a newspaper headline the morning after
a Presidential election by IRO. IRO can do this:

"President Mo failed in his re-election bid. The only difference
between the voting this year & last time is that this year some
of the voters ranked Mo higher & Shemp lower. Presient-elect Shemp
gave his acceptance speech at 11:30 p.m. Eastern Time."


That's a problem of IRO, but not of the methods recommended by EM.


One IRO advocate asked me "Isn't it true that you don't like
IRO because you believe that people will strategize, and abandon
their favorite?"

Yes they will. Some IRO advocates seem to like IRO because they
believe that, by abandoning the middle compromise, it will swing
back & forth between extremes, and thereby give us more variety
& dynamism.

Even if you think that sounds good, there's no reason to believe
that the IRO pendulum will swing both ways. The Democrats &
Republicans are much closer to eachother than to the progressives.
Wouldn't the Democrat candidate ask his voters to rank the Republican
2nd? Wouldn't the people who pay the Democrat campaign also 
advertise that the Republican should be the 2nd choice? The voter
who considers the Democrat his 1st choice is going to be someone
who likes the Republican better than the progressive. So the
compromise will be abandoned for excursions one way only, 
so it would seem.

But let's ignore that, & suppose that IRO's noncentrist generosity
does swing both ways. It would be unstable. Say you're a progressive,
and there are several progressives, like Nader, etc. The one
who is better positioned as a compromise, a Condorcet winner, isn't
your favorite. They're all roughly the same 1st choice popularity,
and that compromise progressive could easily be eliminated in the
1st IRO round, if you don't vote him in 1st place. So you would
vote him 1st. But if you'd do that to elect a progressive, wouldn't
a Republican do it to _avoid_ the election of a progressive. Surely
a progressive Presidency would be as abhorrent to a Republican as
it would be desirable to a progressive. 

Say the Republicans try to vote the Democrat 1st when they feel
they need to, to avoid a progressive victory. But only if they
need to. Let's say that, when in doubt, they try to err on the
side of caution (and protect the Democrat compromise). 

And let's say the progressives tend more to err on the side
of incaution. Well the result will be that sometimes the
Republicans will win without a majority, when the Democrat is
Condorcet winner, but the progressives will never be the recipients
of that boon.


I wanted to give space to that, because the belif that everyone
is better off if compromise is scuttled seems to be shared by
a some IRO advocates. 


Thanks for listening to me on this.


Mike Ossipoff


More information about the Election-Methods mailing list