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

[Dave Ketchum]

On Sat, 14 Dec 2002 20:27:43 -0000 James Gilmour wrote:

>>>Dave wrote, in part:
>>>
>>>
>>>>Like IRV, separate runoffs have been around a long time.  Separate runoffs
>>>>almost frustrated French voters into riots this year and, given a similar
>>>>set of candidates and voters, IRV could easily have stumbled into the same
>>>>result.
>>>>
>>>>
>>>>
>>>I must correct two wrong statements here before they are repeated yet again.
>>>
>>>The only reason there were problems with the run-offs in the French
>>>Presidential
>>>election this year is that only the top two candidates are allowed to
>>>stand in the
>>>run-off.  If a proper process of successive elimination had been used (the
>>>"Exhaustive Ballot"), there would have been none of the problems we saw.
>>>Restricted run-off, like the UK's Supplementary Vote, is a highly
>>>defective voting
>>>system.
>>>
>>BTW - assume you meant "effective" rather than "defective" above.
>>
> 
> No, Dave, I meant EXACTLY what I said.  Both Run-off restricted to two rounds with
> only the top two in round two, and the UK's Supplementary Vote, are highly
> defective.
> 
Oops - sorry.

> 
>>
>>Ok, I could have said more clearly, "Separate runoffs, by the method the
>>French were using,".  Do not see that as especially significant.
>>
> 
> Au contraire - this is extremely significant.  There is a world of difference
> between successive run-offs (Exhaustive Ballot) and a two-round run-off with only
> the top two from round one allowed to contest round two.
> 

I still do not see how this got such great significance.  We seem agreed 
the French got in trouble with the runoffs they were using.  I started by 
saying so; you assumed I made a different and incorrect statement; I did a 
very simple restatement that I meant what I said.  Let's move on to more 
significant problems.

> 
> 
>>If I remember "UK's Supplementary Vote" correctly from a previous post,
>>that is IRV with the voter restricted to voting for 2 candidates.  If so,
>>I would see it as sharing IRV's problems, perhaps extended a bit by the
>>difference.
>>
> 
> Yes, but that's not all.  In our version of the Supplementary Vote you have only
> two preferences ("X" in column one, "X" in column two), but it only the
> preferences for the top two candidates that are counted.  In this respect, it is
> even worse than the French two-round run-off.  At least in the French system you
> know who the top two are when you cast you second vote.
> 
> 
>>>This could not happen with IRV because there is no restriction on the
>>>
>>number of
>>
>>>preferences a voter can mark and all the preferences are
>>>
>>transferable.  Given the
>>
>>>pattern of voting in the first round of the French Presidential, most
>>>
>>voters would
>>
>>>have marked several or many preferences (there were 16 candidates representing
>>>four main political groupings of parties).  Successive elimination of
>>>
>>candidates
>>
>>>among parties of the left and among parties of the right, would
>>>
>>almost certainly
>>
>>>have resulted in the final stage of the count being a contest between
>>>
>>Chirac and
>>
>>>Jospin.  That would have been in accord with the wishes of at least
>>>
>>two-thirds of
>>
>>>the voters.
>>>
>>>James
>>>
>>>
>>I stand by what I said about IRV - that it could fail with a French
>>collection of
>>
>>voters and candidates, not that it would surely fail.  Not clear why
>>French voters
>>could be expected to "have marked several or many preferences", but I am only
>>dealing with possibilities here.  Try a sample election:
>>
> 
> I accept that, theoretically, IRV could fail like the French run-off, especially
> if very few voters marked second preferences.  But I said IRV would not fail as
> you suggested, because I am interested in practical politics and what real voters
> will do in real elections.  I believe there is enough evidence of the behaviour of
> French voters for me to completely confident that what you suggested would never
> happen in an IRV election.
> 
> Presented with 16 candidates from four main groups across the political spectrum
> and the possibility of voting preferentially and of having those votes transferred
> to successive preferences if required, I am sure the French electors would use IRV
> to the full.  Then you would have seen progressive consolidation on the
> centre-right and on the centre-left, with the outcome of a Chirac-Jospin contest
> as I suggested.


One thing that is important to learn about design is that if you include a 
breakable component, you had best plan on the users breaking it sooner or 
later.  Just dreaming that they will never find it is dangerous.
      I suspect that this class of dreaming has afflicted Microsoft 
products significantly.

My sample election was set up for IRV to elect Chirac or Le Pen, as 
happened in the real election, and for Condorcet to elect Jospin, the 
apparent compromise candidate.  Forgetting Condorcet for the moment, and 
looking at what could happen while demonstrating IRV's weakness:
      Whatever lower choices may get voted, Chirac and Le Pen must 
continue to share the two top slots.  Therefore it does not matter what 
voters who specify these as first choice do for lower choices, for IRV 
will never see such.
      I specified half the voters as voting for Jospin as first or second 
choice.  Does not matter what these voters do for lower choices, except 
these must not disturb Chirac and Le Pen ownership of two top slots.
      I specified 10% strays.  Care not how these vote, provided Chirac 
and Le Pen continue to own the two top slots and Jospin as first choice 
remains in slot 6.

Looking again at flexibility:
      Chirac and Le Pen being most popular first choices is consistent 
with the real election.
      So is Jospin as first choice getting a smaller vote count.  Above 
description would tolerate all 90 non-strays choosing Jospin first or 
second.  Even the strays could vote for Jospin as a lower choice provided 
their intermediate choices saw to it that this did not disturb the 
specified order (if half the strays vote S1, M1, Jospin and half vote S2, 
M2, Jospin, the order is not disturbed).

> 
> James

-- 
  davek at clarityconnect.com    http://www.clarityconnect.com/webpages3/davek
   Dave Ketchum    108 Halstead Ave, Owego, NY  13827-1708    607-687-5026
              Do to no one what you would not want done to you.
                    If you want peace, work for justice.

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