[EM] Voting by selecting a published ordering

[Abd ul-Rahman Lomax]

At 08:16 PM 4/3/2006, Dave Ketchum wrote:

>Looking much like Condorcet.  From there we know that cycles can occur,
>needing more thought here.

The proposal was one which uses any method allowing a ranked ballot. 
The difference between it an standard plurality -- it uses the 
standard plurality ballot -- is that instead of voting for a 
candidate, strictly, one is voting for a list provided by a candidate.

While it is theoretically possible for that list to not be headed by 
the candidate, we may presume that it is. I.e., we may presume that 
any candidate will put his or her own name at the top of the list.

It is also possible to do this without using candidate-provided 
lists, in which case it is essentially Asset Voting. But the idea 
here is that candidates have provided a list, which is published, so 
when voters vote for a candidate, they know where their vote goes if 
that candidate is eliminated.

It seems to have been missed by Mr. Ketchum that the voting is for a 
candidate, and that each candidate provides only one list, in advance 
of the election. Hence he asked:

>Starts out looking good but, how many lists might there be with half a
>dozen candidates?

Half a dozen.

>       What would this quantity do to the voting machine?
>       How much might this confuse the voter looking for an acceptable list?

Not at all. I suppose I might, as a voter, be mildly interested in 
what my favorite candidate would do with my vote, but, for me, the 
issue is this: if I consider someone trustworthy enough to serve in 
an office, why not consider them trustworthy to vote for someone who 
is also relatively worthy of election.

This is an extremely simple ballot and process. The method by which 
the lists are used would be any ranked ballot method, as if the 
voters had marked the ballot with one of the provided lists. I 
understand that in STV elections, parties often provide such lists 
anyway, and very many voters, out of party loyalty, use them.

This method is quite clearly superior to standard plurality and, in 
my view, to IRV. It allows experts to make ranking decisions, with 
the public choosing their favorite *expert*. (Candidates frequently 
know other candidates, far better than does the public.)

> > This is exactly the kind of simplicity that we need to get a 
> viable improvement over plurality for public elections.

I agree that it is totally simple. Any complexity would be the 
complexity of the exact election method chosen. However, reporting 
and counting the vote would be much simpler, because the vote can be 
reported just the same as plurality. It is only in the analysis that 
the difference comes. And anyone can do that calculation, using the 
provided lists.

>Sounds like as much trouble as Condorcet and in counting complexity,
>though doable with present voting machines - provided they can tolerable
>the number of choices.

Of course, the method was misunderstood. The number of choices is the 
number of candidates, same as in plurality, and the counting 
complexity is actually low, since precincts can simply report the 
standard vote counts, they do not have to analyze. Analysis will be 
no more complex than required by the ranked election method used, and 
should be easier because of the limited number of ballot types involved.

This is a case where overvoting could create problems; however, there 
might be a way to analyze overvotes that would work. But I won't get 
into that here, I'm just noting that there is, with this proposal, on 
the face, a reason to discard overvotes that is rational, unlike the 
case with discarding them under standard plurality. Overvotes here, 
unlike the case with standard plurality, would create counting 
complexity even if a way is determined to use them. I'd agree with 
keeping it simple.