[EM] Explaining PR-STV

[Raph Frank]

On Fri, Aug 28, 2009 at 11:08 PM, Kathy Dopp<kathy.dopp at gmail.com> wrote:
> Ralph,
>
> I think you've forgotten some crucially important points in your
> explanation of how STV is counted. My comments below...
>
>> From: Raph Frank <raphfrk at gmail.com>
>> Subject: [EM] Explaining PR-STV
>
>> PR-STV is based on 4 main principles
>>
>> 1) Each voter gets 1 vote and they can vote for any candidate they want.
>>
>> ** All votes are equal. **
>
> To clarify, add..
>
> "All votes are equal"... *if* they are counted but some voters' never
> have their 2nd or 3rd choices counted at all or before their 2nd or
> 3rd choice candidates are eliminated

I think we have a philosophical difference here (in fact, I know we do).

You consider each individual ranking a vote.  However, I don't look at
it that way.

Each voter has exactly 1 vote.  This vote means that they can increase
by 1 the vote total for any candidate.

The ranks are not votes, they are just instructions to the counting
official on how you want your vote handled.

If a candidate is eliminated, you instruct the counter to move you
vote to the next highest candidate who is still running.

Similarly, if a candidate is elected, you instruct the counter to move
the part of your vote that they don't need to the next preference.
.

> often *not true* but it is complex to explain how in STV a candidate
> with more votes might be eliminated and a candidate with fewer votes
> win instead due to nonmonotonicity which itself is due to the unequal
> treatment of voters' votes whereby voters who support the least
> popular candidates have the most say in which candidates are
> eliminated, etc.

It is true.  You won't be eliminated unless you have the least number
of votes at that point.

>> 3) If you vote for a losing candidate, your vote is transferred to
>> your next choice
>
> This is only true in special circumstances in STV/IRV, namely your
> vote for a losing candidate is *only transferred to your next choice
> **if** either:
>
> 1. your vote for a losing candidate occurs in early rounds so that
> your next choice has not yet been eliminated, and
>

Fair enough, I should have said most preferred candidate who has not
been eliminated or elected.

> 2. your vote for a losing candidate is for a losing candidate who does
> not lose in the final elimination round, in which case your later
> choices will never be considered.

Right, there is up to 1 Droop quota of voters who don't get
represented.  However, this is much better than potentially 49% of the
voters not being represented in a single seat district.

>> This reason for this rule is is so that you can safely give your first
>> choice to your favourite even if he is a weak candidate.
>
> This is a wholly, entirely, deceptively false statement. In IRV/STV
> your first choice vote can always hurt the chances of your 2nd choice
> candidate winning.

Mostly, I don't think non-monotonicity is an issue with PR-STV (or at
least the benefits outweigh that disadvantage)

Would you prefer the simple version where the 5 candidates who
received the most votes in the first round win?  That is monotonic,
but is much less fair and gives more power to the parties.

> Yes. this would be more accurately rephrased "Be careful to vote for a
> very very weak candidate first if you do not want your later more
> popular candidates to lose."

Actually, one of the strategies for increasing the power of your vote
is to vote for a weak candidate first.

Your later more popular candidate will not be eliminated before a
weaker but preferred candidate.

>> 4) If you vote for a candidate who gets more votes than he needs, the
>> surplus is transferred to your next choice.
>
> Again, this is only true in special situations similar to those
> mentioned above for having your vote for a losing candidate
> transferred.

Ok, it is only transferred if you have indicated which candidate you
want to transfer it to.

I don't see how that is unreasonable.

> You forget to mention that often STV can not fill all the seats unless
> the quota is reduced to account for all the voters whose ballot
> choices have been expired or eliminated so that many many voters in
> STV are prohibited from participating in the final counting rounds.

I don't think voters should be required to fill out all their rankings.

If voters don't indicate a preference, then that is their choice.

Having said that, I would support decreasing the quota on the fly.
However, that is just making the method more complex for little
benefit.

> Hence any jurisdiction which has adopted STV have had to eliminate any
> requirement for majority winners, etc. since the method most often
> fails to find sufficient candidates that meet the quota.

Well, the post is about PR-STV.

I agree that in IRV, the disadvantages outweigh the benefits.
However, the sheer power that PR-STV gives to the voters outweighs any
disadvantages (and the disadvantages are lessened by it being
multiseat anyway).

It shifts power over candidate selection from party to the voters.

> totally false statement depending on the definition of "strong".

Well, in the context, I meant a candidate who easily reaches the quota
(in the example, the candidate received double the quota).

> There
> are many examples where voters in STV are only allowed to have a vote
> counted for one candidate even though they are supposed to be electing
> a multi-seat *at-large* council, or where the *strong* candidate makes
> it to the final counting round and then loses, where a *strong*
> candidate (the first choice of *all* voters in a pairwise comparison)
> is eliminated in an early round and a less *strong* candidate wins,
> etc. etc.

Well, the the single seat case, a "strong" candidate would be one who
gets a majority of first preferences.