More on electoral college (was Re: [EM] Reply to Steve's post of

Steve Eppley seppley at alumni.caltech.edu
Mon Oct 28 14:25:52 PST 1996


Donald D wrote:

>Donald: You have changed my thinking in the case of only one
>candidate within reach of a win. I now say we should force the
>Reformed States to elect this one candidate.
-snip-

This is the simplest proposal and I prefer it too.  After the RS
have reached critical mass, this will be moot.

>We still have two cases to consider: One - There is the case of a
>candidate in the UNreformed States receiving enough electoral votes
>to be President-elect.
>Two - there is the case of no candidate within reach of a win.
>For both of these cases I say we let the Reformed States vote for
>whoever they want to vote for using some single winner method.

In both these cases it's moot; the RS can't influence the result:
In One the election has already been decided in the other states 
(the RS are too small).  In Two, the House will decide.

So why not keep the proposal clean and simple?  I suggest we not 
complicate the proposal by treating these moot cases specially.

>>Steve: Why do you think it's so much better to eliminate the
>>national can't_wins before tallying the order in the reformed
>>states?
>
>Donald: I feel that it is a better sell - it avoids the possibility
>of the Reformed States coming up with a winner that is not one of
>the candidates within reach of winning it all.

But that's not the proposed alternative; in neither would the RS
give any delegates to a can't_win.  

The alternative proposal is this:
1. Determine the collective order in the RS before eliminating any 
candidates.
2. Eliminate from the collective order any candidates who can't win.
3. Give all the RS delegates to the candidate highest in what remains
of the collective order.

For example:
  Suppose the RS collective order, determined solely by tallying the
  RS voters' ballots--before considering who can win the EC--is:
     Dole, Perot, Clinton, Nader
  Suppose Perot and Nader can't win a majority of the EC.  
  Eliminate Perot and Nader from the collective order to get this 
  remaining order:
     Dole, Clinton
  Then the RS give all delegates to Dole.

Since *both* proposals meet your "better sell" test, some other basis 
is needed to differentiate them.

>Three - the question of electoral ratio - do we preserve the ratio
>of each state? If not - why not - and can we sell it to the states?
>If we do preserve the ratios I am interested on how you will do that
>with the single method you will be using.

Changing the ratio would require amending the U.S. Constitution, so
that's out of reach of the proposal.  But the imbalance will become
moot once all the states are in the RS: then the electoral college
will always be unanimous!  Even before all the states join the RS,
the imbalance will shrink in importance as the RS grows.

-snip-
>>The IR collective order is:  Dole, Clinton, Nader
>
>Donald: "VERY Interesting" - I would not have done it like that.
>First I would have removed Clinton from the ballot because he is
>the lowest on the first tally - this gives us: 46 Dole-Nader 34
>Nader-Dole Nader is now lowest so I would remove him - giving us 
>80 Dole
>
>Do I misunderstand how Instant Runoff works or are you using
>pairwise methods and logic on Instant Runoff? 

You understand that IR picks Dole after eliminating Clinton then
Nader.  You're inferring that it places Nader second since Nader was
the last eliminated, but that's not explicitly part of the definition
of IR, which (like other methods) is silent on which candidate
finished second.  I don't think that's the best way to determine
who the runner-up should be if the 1st choice can't serve.

I'm not using pairwise logic here.  (It only looks like there's
something pairwise because it's a 3-candidate race and 1st place
finisher Dole has been eliminated, reducing the field to two.)  
I'm suggesting that the best way to determine the runner-up is
to eliminate the 1st place finisher from the voters' rankings and
retally.  It's like saying to the people "well, your first choice
Dole just died of a heart attack, so we'll let you all vote anew." 
The people would then vote to elect Clinton.

This was part of the definition I gave for calculating a collective
order based on any single-winner function.  I didn't intend to imply
that my definition is the only conceivable definition.  I offered it
because it looks fairest.

Also, this way has the advantage of being independent of the sw
method, whereas picking Nader second because he was last eliminated
is specific to IR.

>>I also don't like the "vote-sum" terminology in your later message.
>
>Donald: If you have a better term I am receptive. With all the
>different possible combinations we are going to have sums of votes
>attached to most of these combinations. I call them Vote-Sums for
>want of a better term. 

It may be possible to come up with a better term ("combo count"?),
but my main objection is that this part of the proposal is just a
detail of a particular implementation.  There are other ways to
implement the same sw algorithm; storing vote-sums could be the
optimal solution but it's not a basic requirement since storing
all the ballots could also work fine.  Therefore, I think you're
overspecifying, and this makes the proposal more complex for the
reader.  If I were you I'd separate those sections and put them 
in an appendix.

By the way, have you reconsidered your preference for IR?  I can
understand the argument that the mechanics of Condorcet might be
significantly harder to explain to the public when campaigning for 
it by a poorly-funded initiative, but I recall your reasons were
different.  Something about cultists?  :-)

---Steve     (Steve Eppley    seppley at alumni.caltech.edu)




More information about the Election-Methods mailing list