Here's article on STV as proposed for Portland:
Here's a criticism of the examples:
In each, the process shown is one or two rounds more than needed to find the winner. you don't need quota to win. To win at the end you just have to be the most popular among last standing when the number of candidates is thinned to number of remaining seats.
Transfer of surplus votes does not have to use fractional votes as done in the examples, but can just be whole votes, as done in Ireland and Malta for last hundred years.
So these examples actually shows STV as more complicated than it has to be.the man thing is each voter cats just one vote and multiple members are are elected - no one group can take all the seats as happens under FPTP and Plurality block voting.
Most -- 29 or so of the 36 voters in these examples -- see the candidate marked as first or second choice elected. so a high rate of effective votes, while under First past the post as much as 82 percent of votes cast can be ignored when the single member is elected.
It is good to know the basic mechanics of STV; Performing pretend votes is cool (or I think so anyway)
But the main thing is the proportionality and fairness of the vote under STV - with or without party lines, under STV most voters see at least one of their preferred choices elected - a high proportion of effective votes - the general reason why we need to switch to PR.
These examples are not great for that.
Nowhere is the number of effective votes calculated - in fact it is noted in one example that the voter sees two of his three marked preferences elected. The goal of PR is not to have more than one choice elected for each voter but to have one elected for as many voters as possible.
Examples go one or two rounds more than required.
Example 1: only two rounds were required. with one elected in round 1 (no surplus to transfer) and one eliminated in round 2, there were only three candidate remaining for two remaining empty seats. the least popular of the three is eliminated and the two remaining are declared elected to fill the last two seats. Easy parcheesi! you don't actually need quota to be elected.
so you don't need the Round 3 shown.
Example 2. pretty much the same
Round 2 one eliminated and votes transferred electing one other.
Three remaining and two empty seats. The least popular of the three is eliminated and the two remaining are declared elected to fill the last two seats. Easy sneezy! you don't actually need quota to be elected.
Round 3 and 4 as shown are not necessary so avoiding the ultra-detailed Gregory fractional vote transfer shown.
Tie complicates things slightly but (generally) the candidate with the most first preferences is awarded the advance in case of tie.
Fractional vote transfer not necessary in any case - transfer of whole vote based on next marked preference is considered good enough in many systems.
Example 3. pretty much the same
Round 2 one eliminated and votes transferred electing two others.
Two remaining and one empty seat. The least popular of the three is eliminated and the one remaining is declared elected to fill the last seat. Easy meezy! you don't actually need quota to be elected.
so you don't need the Round 3 shown.
effective votes measured would be something like this:
Example 1 29 elect either first or second choice
24 elect their first preference,
5 elect their second choice,
7 do not elect their first choice (and in real life would not be used to elect anyone)
(in the un-necessary transfer shown, it is shown that each of those voters sees their second choice elected. so due to extra transfer, all voters see their first or second choice elected.
example 2: one wolf vote did not go to elect first or second choice
seven Hansel votes did not elect first choice.
(in the un-necessary transfer shown, it is shown that each of the Hansel voters sees their second choice elected; the wolf/Hansel vote is shown to be wolf/hansel/Goose so in the end it helps elect the Goose.)
so 25 voters see their first choice elected.
9.8 voters see their second choice elected.
1.09 voters (wolf/hansel/Goose) did not see first or second choice elected; they did see their third choice elected. This is done in the un-necessary round 4 transfer.
without that Round 4 transfer, the stats would be :
25 see first choice elected
2.8 see their second choice elected
8.1 not used to elect anyone.
Example 3. 23 voters see their first preference elected
13 (7 Hansel; 6 Gretel) see their second preference elected.
(the 7 Hansel votes in real life would not elect anyone. and would not be transferred.
but even without the Round 3 transfer of the Hansel votes.
29 out of 36 voters would see their first or second preference elected.
so them's good PR.
(In these examples, no votes were exhausted.)
note that with only five candidate and three seats there were not many rounds, so not much recourse to go to third or later preferences and transfer of previously-transferred votes.
Under Gregory system these can be complicated as it entails fractions of fractions but under whole vote surplus transfers even that is fairly easy math.
and note than in each example the candidates who were most popular in round 1 were elected in the end. Thus, final result is shown to be same as result under SNTV (if votes cast same) excepting ties. (Ties actually are seldom seen in real life elections.)
The transfers in STV elections sometimes don't change the order produced by single voting in MMD, and seldom change as much as 30 percent of the seat winners in a district.
In the examples they did not change any front runners, but did resolve the Goose/Hansel tie in example 3 and Goose/Hansel/Gretel tie in example 1. (ties do not happen in real life except once in a century). in real life, transfers do ensure that parties (not shown here) are represented proportionately (such as in many cases ensuring that surplus votes are transferred from one candidate of a party to another of same party)) but almost always it is the most popular candidates of that party in Round 1 who are elected in the end, even if some initially-popular candidates of other parties are denied a seat. Party proportionality and high number of effective votes trumps round 1 popularity but this is determined by voters' secondary preferences transfer when candidates are eliminated and when surplus votes are transferred - if voters mark their ballot that way - parties actually have no direct influence in STV. But due to single voting in MMD, already in round 1, seats are distributed among parties (if parties are used) and already half or more of votes are already distributed among the candidates who would be successful in the end.
so the demands of party proportionality and high number of effective votes seldom is so demanding that the initial ordering of popularity of candidates is affected much by transfers.
In these three examples, the transfers did not change the initial popularity at all (except resolving ties:
example 1 - two of the three in the tie were elected; one was not.
example 3 - Goose was elected; Hansel was not.)
But despite the almost zero effect of transfers, the result was fair and high number of votes saw their vote used to elect first or second choice.
unlike FPTP elections where two thirds of votes in a district are often ignored, and as many as 82 percent have been ignored.
=============================
Comments