[EM] 3ballot - revolutionary new protocol for secure secret ballot elections

[Abd ul-Rahman Lomax]

At 12:54 PM 10/1/2006, Brian Olson wrote:
>The down side is the strategy arguments about casting an honest ballot vs
>casting a ballot more likely to get you some of what you want. Straight
>ratings does not promote honest voting, but instead promotes saturating
>your ballot to the min and max of whatever scale you're using. There's
>loss of information there, and it doesn't then find the best
>winner and maximize social utility.

This is a common argument, and quite misleading. It is based on 
conflating two concepts that really don't belong together.

If a voter fails to vote, for at least one candidate, the maximum 
rating, and for at least one candidate, the minimum rating (which 
might be a blank, depending on the details of the implementation), 
the voter has essentially cast a weak vote. The easiest way to see 
this is to consider that the voter has one vote to cast. Standard 
voting methods allow the voter to cast a full vote or a no vote, 
nothing in between. Range allows a range of votes to be cast, and, 
while they may be cast as ratings, say from 0 to 9, they are actually 
quite equivalent to fractional votes, in the range of 0 to 1. (For 
Range 10, i.e., 0 to 9, one divides the range vote by 9 to get the 
fractional vote.)

This truth is then extended beyond its proper application to assume 
that voters, to not cast weak votes, should vote for *all* candidates 
0 or 1. Essentially, this reduces to Approval Voting. Not a bad idea, 
actually, but not as fine-grained as Range supporters want to see.

If voters do this, that is, make black and white decisions, *they* 
have chosen to remove the more detailed information that would be 
expressed if they placed some candidates in the middle. Now, Mr. 
Olson assumes that there is some strategic advantage in voting black 
and white, that somehow voters will be encouraged to do this.

Actually, there is no strategic advantage to voting the extremes for 
all candidates. The most you can contribute to the victory of a 
candidate is one vote. You do this for the candidate you most wish to 
win. If you devalue all other candidates, giving them no votes, you 
must believe that they are all equally poor choices, and you must not 
have any preference between them. If you do, you have shot yourself 
in the foot by failing to rank them appropriately.

Yes, you reduce the possibility of your favorite winning if you give 
other than a zero vote to all other candidates. But you gain the 
possibility of improving the election outcome in case your favorite 
does not win.

Warren's research indicates that people will vote honestly. While his 
research was limited, to be sure, and I think he draws conclusions 
from it that are perhaps unwarranted, his thesis is nevertheless 
quite reasonable, that people will not bullet-vote. I believe that 
Range Voting actually can be shown, though I'd leave it to Warren, to 
reward honest voting by producing optimal election outcomes.

To review, Mr. Olson wrote this:

>The down side is the strategy arguments about casting an honest ballot vs
>casting a ballot more likely to get you some of what you want. Straight
>ratings does not promote honest voting, but instead promotes saturating
>your ballot to the min and max of whatever scale you're using.

In Range, contrary to what Mr. Olson implies, casting an honest 
ballot is the procedure most "likely to get you some of what you 
want." The only caveat is that voters should understand that they 
devalue their vote if they fail to give one of the candidates, at 
least, a zero, and one of the candidates a full vote. If they want to 
do that, fine. It is a partial or complete abstention. Most won't 
want this, I think.