Exaggerated opinions

[David Marsay]

Dear Mike,

Thanks for your response of  Wed, 29 Apr 1998 to my:
 
<< I think that some variation of Condorcet is the 'natural' method 
for  ranking, with complexity being its only real draw-back >>

I was intending to comment on how to derive a 'majority' verdict 
from collection of ballots, each of which gives a ranking. I do, 
though, think that using such ballots is also fairly natural. 
However, first-past-the-post (FPP) would also have a claim, so I 
couldn't say that using ranked ballots is 'the' natural method. I 
currently thinking about alternatives that are not so 'natural', but 
which may be expedient.

> I note that in order to conduct a ranked vote, there must first be prepared a
> single ballot showing "all of the available choices".

I am really thinking of situations where these are 'given'.

> Naturally, it is
> impossible to enumerate "every possible choice" therefore some
> dictatorial/hierarchical element must be created to create "the official
> ballot".

I agree that there is a problem here. One could invite the voters to 
propose options. This might lead to too many. A good thing about 
FPP is that it encourages the creation of a few parties who 'thin 
out' the candidates.

Note, though, that there may be a natural compromise option that is 
no voter's first choice.

> And if that ballot neglects to include "none of the above" then it
> can and will be used to force an outcome even if the voters uniformly dislike
> all of the proferred choices.

Agreed. Arrow showed that we must allow for draws, so there also 
needs to be a good action in the event of a draw. Hardly anyone will 
have voted for it, and yet if there are only a few voters it could be 
quite likely.

> That same dictatorial/hierarchical element can
> also delay a vote with the plausible "we haven't gathered all of the possible
> options yet".

Isn't this just a 'do nothing' option? Should the ballot have to give 
a reason for doing nothing?
 
> Ranked votes (i.e. first, second, third) of ANY sort, regardless of the
> scoring method being used, necessarily involve a dictatorial ballot-creating
> function.

Unless there is some accepted option-generating method. I regard this 
as outside the pure voting system.

> In order to get group decision WITHOUT invoking a dictatorial element, there
> are only a couple of possible choices.  Namely, each option or choice gets its
> own ballot, and new choices/ballots are created by the members of the group
> when desired (not the "official ballot committee").

Why not allow members to put post options on a board and discuss them 
for a week (say). Like-minded members could then agree to work 
together on improved options and withdraw the old ones. Some members 
may even be motivated to develop compromise options. The final list 
would then be put to the vote.

One problem here might be  groups attempting to modify their 
option in the light of those of other options, for example to find 
'the middle ground'. With Condorcet they could put forward a few 
options to mark out the common ground.

> The voting system used
> should then allow each member to express an honest opinion about each choice
> when offered.  Yes/no is possible but pretty simplistic.  Better is
> Support/Oppose/Neither/Both.  Other versions can allow more gradations for
> better expression.  

I think that using four grades and applying Condorcet is about as 
good as one will get for a general-purpose method. (I've posted on 
this before!) I wonder if, between us, we aren't beginning to make it 
seem pretty natural.

I'm not sure if you intend all the votes to take place at once, or as 
the ballots are created. If the latter, I would be concerned about 
comparing results of ballots made at different times. I suppose I 
could always re-run an old ballot. I might think your description 
appropriate to decision-making in a club, but is vulnerable to 
'spoilers' flooding voters with ballots, leading to degraded 
decision-making. Perhaps every ballot should have a box where voters 
can indicate how soon they would be happy to entertain another ballot 
from the same source. The default could be a week (say), and the 
results could be averaged.

Cheers.
--------------------------------------------------
Sorry folks, but apparently I have to do this. :-(
The views expressed above are entirely those of the writer
and do not represent the views, policy or understanding of
any other person or official body.