[EM] IRV vs Plurality

[Abd ul-Rahman Lomax]

At 08:51 AM 1/27/2010, Jameson Quinn wrote:
>2010/1/26 Abd ul-Rahman Lomax 
><<mailto:abd at lomaxdesign.com>abd at lomaxdesign.com>
>
>The typical error is in assuming some "strategic" faction which 
>votes sensibly, when everyone else votes in a way that they will 
>regret if they discover the result they cause.
>
>You can't study history for two minutes without finding significant 
>groups of people (that is, ~5% fractions, not "everyone else") 
>behaving in ways they come to regret later. This is unavoidable 
>human nature, and acknowledging it is no error.

Now, is the voting system supposed to protect them from their stupidity?

We were not talking about a minor faction voting that way, but half 
of the two largest, with 20% of the vote, in the example Mr. Quinn 
created, the total faction being 40%. The opposing 40% vote 
*normally!*, all of them. The vote cast by this 20% faction 
effectively says "I have very little preference between Bore and 
Cush." (It's 1/5 vote!) So if the voting system interprets their vote 
that way, we are supposed to blame the system?

There is supposedly a "Moral issue" here, the idea that we are not 
supposed to "reward dishonesty." But a vote is never dishonest, it is 
an action, tossing a weight or weights in a balance or balances, 
intending to swing the balances some way. It's not a sentiment, and 
it is not a statement under some binding rule of honesty that could 
be defined. Reversing preference is more arguable, perhaps, but this 
still applies. Moral responsibility lies in the consequences of the 
actions, or at least in what we could anticipate from them. Hence, if 
the 20% believes that their vote was foolish, the responsibility for 
that lies with them.

Look, I would not dump full-blown Range on an electorate. I'm 
proposing Bucklin, which would allow ranking in a way that makes it 
roughly equivalent to rating, with the three ranks meaning -- and it 
might say that right on the ballot -- "Favorite," "Preferred," and 
"Also Approved." I prefer to see this as the primary in a runoff 
system, so that "Also Approved" has a very clear preference meaning: 
"I prefer to see these candidates elected to a runoff being held." 
Later-No-Harm is important to you? Fine. Just vote for your favorite. 
If your favorite gets a majority, done. If not, then, if your 
favorite is a leader by the criteria used to determine runoff 
candidates, you will presumably vote for your favorite again. If your 
favorite doesn't make it, you will still have your option to cast a 
vote indicating preference again.

This isn't *difficult.* But it leads, later, after there is more 
experience, to using a Range Ballot to accomplish the same thing, 
only with a bit more subtlety, and the Range ballot would start with 
only one more rank: "Preferred to the worst." The original ballot can 
be analyzed as Range 4, with a vote of 1 missing, the values are 4, 
3, 2, 0. The extended ballot adds the 1. Initially, that might not 
even be used to determine a winner, but would be used to collect 
preference information from voters who don't support a winner. It 
might then be used to make somewhat better choices for runoffs. It 
would never be considered approval of such a candidate, indeed, it is 
disapproval. Following basic democratic procedures, a winner would 
never be declared in a primary with less than a majority, but if, 
after study of such elections and runoffs, it is determined that 
there is no significant risk of reversal in a runoff with a less 
stringent standard, that's fine with me; compromises are made in the 
name of efficiency when the loss in efficiency is great enough to 
warrant the loss in full confidence.

Note that in the election described, by the votes, there would be a 
majority winner if we assume that 80% is approval....