[EM] Re: why ranking should be allowed for approved candidates only

[Russ Paielli]

Araucaria Araucana araucaria.araucana-at-gmail.com |EMlist| wrote:
> Russ Paielli <6049awj02 <at> sneakemail.com> writes:
> 
>>Araucaria Araucana araucaria.araucana-at-gmail.com |EMlist| wrote:
>>
>>>On 17 Apr 2005 at 14:28 UTC-0700, Russ Paielli wrote:
>>
>>>>By only allowing the approved candidates to be approved, we can
>>>>significantly simplify the procedure for both the voter *and* the
>>>>equipment manufacturer. And we can do so at very little "cost" in
>>>>terms of voting "expressibility." If you are serious about actually
>>>>getting a new voting system adopted, I urge you to reconsider
>>>>allowing ranking of unapproved candidates.
>>>
>>>
>>>Hi Russ,
>>>
>>>The strategic ability to rank below the cutoff is what enables 
>>>DMC/RAV
>>>to discourage defection cases like this:
>>>
>>>   27: A>>B
>>>   24: B (truncates >A preference)
>>>   49: C
>>>
>>>Without that strategic disincentive, voters in this election might
>>>simply bullet vote and you end up with C.
>>
>>For the votes you show, I figure that DMC/RAV picks C. If ranking of 
>>unapproved candidates is disallowed and the 27 A>>B votes are changed to 
>>just A, then C still wins. If we start with your votes and change the 24 
>>B votes to B>A, then A wins. If we start with the A>>B votes changed to 
>>A, then change the B votes to B>A, A wins.
>>
>>I must be missing your point. According to my tallies for the four 
>>variations mentioned above, it makes no difference whether the 27 A>>B 
>>votes are changed to A only or vice versa. It is true that if the 24 B 
>>votes are "untruncated" to B>A, that gives the election to A. But so 
>>what? If A was the B voters *approved* second choice, they shouldn't be 
>>overly disappointed. What did I miss?
> 
> 
> C wins any which way because A voters put the approval cutoff above B to
> discourage defection.
> 
> If B voters don't want C to win, they must not truncate to create a cycle.  
> That is the anti-defection strategy (on A's part) I'm talking about.  It's
> sort of like a poison pill.

Ted,
I'm still not getting it. Let me lay out my calculations more explicitly 
just to be sure I'm not making any silly mistakes. I'll use "|" to 
indicate the approval cutoff (I like to be different).

27: A|B
24: B      result: C wins
49: C

27: A
24: B      result: C wins
49: C

27: A|B
24: B>A    result: A wins
49: C

27: A
24: B>A    result: A wins
49: C

Do you agree with these results?

Considering these four cases, A wins if the 24 B voters approve A as 
their second choice, or C wins if they don't. In either case, the 
outcome is not affected by whether or not the 27 A voters rank their 
unapproved candidates. Hence I don't see why this case is relevant to 
the issue of allowing ranking of unapproved candidates.

The fact that A wins when the 24 B voters approve A seems reasonable to 
me. Sure, those B voters might regret having approved A, but at least 
they'll get one of their approved candidates elected.

> DMC/RAV is the simplest summable voting method to discourage this kind of
> defection.  But it works ONLY with an approval cutoff.
> 
> 
>>>If the ballot has to be simplified, 3 approved + 2 disapproved ranks
>>>are pretty simple.  This allows a voter to rank 3 choices as
>>
>>I don't care for ballots that have arbitrary restrictions on how many 
>>candidates can be approved or disapproved or arbitrary conventions about 
>>which candidates are approved or disapproved. I also think that such 
>>arrangements will inevitably lead to confusion.
>>
>>Granted, even if we only allow the approved candidates to be ranked, we 
>>will still have some confusion, but it just seems more "natural" and 
>>intuitive to me. Think of it as a generalization of Approval voting: you 
>>only select the approved candidates, except that now you can rank them 
>>too if you wish.
>>
>>By the way, if you don't wish to rank them you can make them all equal. 
>>Then your vote will have the same effect it would have in Approval. If 
>>you think Approval is a good method, how can you complain about that?
> 
> 
> I'm well aware that equal ranking is possible.  I'm not satisfied with
> approval alone because it loses preference information.  "Just good enough"
> is the enemy of the great.

Well, if the optimal strategy in every case was to rank all the 
unapproved candidates equal and all the unapproved candidates equal, 
then DMC/RAV would be "as good as" Approval. I doubt that will always be 
the optimal strategy, however. Hence, I claim that DMC/RAV is "better 
than" Approval. How much better I don't know yet, but I'll bet it's 
significantly better.

> 
>>>    1 2 3
>>>    1 2 4
>>>    1 4 5
>>>
>>>to move up the approval cutoff.  Or as grades,
>>>
>>>   A B C
>>>   A B F
>>>   A D F
>>
>>I don't like grading schemes either. They just don't seem right to me 
>>for public elections.
> 
> 
> It isn't necessary to have limited ranks or fixed cutoffs at all.
> 
> Say that OCR is used to read filled in ordinal ranks, with large boxes
> that the voter fills in with big numbers:
> 
>         +---+  +---+  +---+
>         | 0 |  | 0 |  | 1 |
>         +---+  +---+  +---+
>         +---+  +---+  +---+
>         | 0 |  | 0 |  | 2 |
>         +---+  +---+  +---+
>         +---+  +---+  +---+
>         | 0 |  | 0 |  | 3 |
>         +---+  +---+  +---+
> 
> 
> Just like the 1040EZ hand-written tax form (at least the version from a few
> years back.
> 
> Add an extra "Minimum Rank Approved" candidate to each race, and you get the
> approval cutoff with no extra software -- just use the votes vs. MRA to fin
> the approval scores.  If no vote for MRA are entered, all ranked choices are
> approved.
> 
> Machine reading would be adequate for 95-99% of the ballots, the remainder
> could be entered by hand.
> 
> Machine assisted ballots are also easy -- you could use a PDF form for the
> election that would be printed out and not saved.  Only the paper ballot
> would be counted.

Those schemes might or might not be acceptable. I realize they seem very 
simple to you, but I think they may still be too complicated for major 
public elections. Also, I don't like the idea of requiring the voter to 
actually write a number. That's asking for trouble because the written 
number will sometimes be ambiguous.

I envision something like the Graphical Voter Interface (GVI 
http://ElectionMethods.org/GVI.htm) that I developed a while back just 
for kicks. It has a column of buttons, each about a half inch high by 3 
inches wide, with a candidate's name and party on each one. You select 
them in order of preference by simply touching them on a touchscreen (or 
clicking on them with a mouse on a conventional monitor). You can always 
backtrack, of course. You can specify equal rankings by touching a 
selected candidate a second time (GVI doesn't currently allow that, but 
it could be added).

Remember that there is little or no time for "training," so the 
interface needs to be as simple as possible -- especially for Democrats! 8^)

--Russ