[EM] 12/13/02 - Giving `crutches to weak candidates':

[Adam Tarr]

Don Davison wrote:

>... why are
>you supporting Condorcet and/or Approval Voting?  For, this is what these
>two method do, they give `crutches to weak candidates'.

Do you mean weak candidates like "Centrist", below?

10% FarRight>Right>Centrist>Left>FarLeft
10% Right>FarRight>Centrist>Left>FarLeft
15% Right>Centrist>FarRight>Left>FarLeft
16% Centrist>Right>Left>FarRight>FarLeft
15% Centrist>Left>Right>FarLeft>FarRight
13% Left>Centrist>FarLeft>Right>FarRight
11% Left>FarLeft>Centrist>Right>FarRight
10% FarLeft>Left>Centrist>Right>FarRight

Note that this is an easy, realistic example... I just put five candidates 
on a standard political spectrum.

Let's look at who has the most first-place support:

31% Centrist
25% Right
24% Left
10% FarLeft
10% FarRight

Or who has the most second place support:

28% Centrist
26% Right
25% Left
11% FarLeft
10% FarRight

In addition to having the most first AND second place support, Centrist is 
the only candidate that is never ranked lower than third on any 
ballot.  And yet, as you surely realize, Centrist loses in IRV.  Right will 
beat Left 51%-49% in the final runoff, even though Centrist would beat Left 
or Right by around thirty percentage points -- absolute landslides.

How could anyone call Centrist a "weak candidate" with a straight 
face?  Weaker than who?

>Non-monotonicity is a bad joke, it does not exist, it has never
>happened in a real election,

This is because no real election has ever had three strong parties.  IRV 
keeps any third party from threatening the other two.  History backs me up 
on this.  Is that a positive feature in your mind?

>A few months ago I posted some real ballots to this list and requested
>anyone to use the ballots and prove that Irving or STV can be
>non-monotonicity in the real world.  No one responded.

That's a lie.  Steph responded.  He correctly pointed out that due to the 
extremely strong support for a few candidates, there was no monotonicity 
violation there.  I'd draw an analogy to Condorcet voting - you're not 
going to find a cycle every time.  Monotonicity violations in IRV elections 
would be a bit more common than cyclic ties in Condorcet elections.  Which 
is to say that neither would happen in elections with two strong factions.

It's worth noting that those were STV ballots, and it's not clear that 
candidates and voters would have acted the same way if it had been a 
single-winner election.  It's also worth noting that I could have easily 
drawn out a subset of the votes and found a Monotonicity violation lurking, 
if those had been the only votes.  But I don't think that proves much, 
except that IRV can fail monotonicity (and we already knew that).

I don't think of monotonicity as the great demon of IRV.  I think of it 
more as a red flag.  If a method fails Monotonicity, then it's a sign that 
it can be manipulated or can produce unfair results in other cases.  Which 
is true for IRV.

>Pity votes will have little influence in an Irving election, but the bad
>effects of `pity votes' can easily happen in the two methods, Condorcet and
>Approval.

Approval, granted.  It's easy to get bad results in approval if the voters 
vote very stupidly.

But Condorcet?  Show me the example, Don.  Dream up something.  Show me my 
second-place votes beating my first place votes.  Oh, you can do it, but 
it's not easy at all.  It takes some very strange breakdowns in 
voting.  When you finish making such an example, will you be able to say, 
"that example is more realistic than the monotonicity violations"?  Or will 
you be able to say, "that's a less democratic result than the monotonicity 
violations"?  I'd guess no, and no.

Come up with an example, Don.  Or are "charlatans" the only ones who create 
examples?

-Adam


----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em