VERSION 1.09
NAME
HandCountMethods
Description
Documentation on Hand Counting Elections and the Vote::Count equivalents.
Instant Runoff Voting
IRV is extremely easy to handcount.
Stack the Ballots by Top Choice
Count the stacks
Elect a Majority Winner
If there is no Winner, eliminate the low choice, distribute the Ballots to the new Top Choice. Either re-count the stacks or add the number of Ballots being added to the stack in the tally.
In the event of ties the default for Vote::Count is Eliminate All.
Implemented in Vote::Count::IRV.
Condorcet
It is easy to create pairings by hand. With a larger field the work to fill out the matrix becomes extensive. To reduce the Matrix as much as possible, Hand Count Condorcet Methods should use the TCA Floor Rule (see the Floor module), unless voters will be marking all choices (in which case Floor rules have no effect).
If a method only seeks a Condorcet Winner, a significantly smaller number of pairings is needed. A 10 choice election requires 45 pairings, a method that seeks only a Condorcet Winner can find the Condorcet Winner or determine there is none in as little as 9 pairings.
Hand Count Condorcet Methods do not require a full Matrix, only the determination of the Condorcet Winner or that all choices have at least 1 loss. Use the following process for this:
General Hand Count Condorcet Process
If you count Approval first, you can prefill some of the matrix. Any choice with an Approval lower than the Top Count of the other will lose the pairing.
Elect a Majority Winner if present.
Start a sheet for each choice with a Wins and Losses Column.
Then starting with the Top Count Leader compare them to the next highest choice (that they haven't already been paired to) and pair them off, recording the result on the sheet.
Continue pairing the winner to the next choice.
The next choice will be the highest Top Count that has not yet been in a pair. If all choices have been in a pair, select the highest Top Count choice not yet paired with the current undefeated choice.
When a Condorcet Winner is found they are the Winner.
When all choices have a loss there is no Condorcet Winner.
Benham Condorcet IRV
This is the best known Condorcet Hand Count method.
Returns the winner as soon as there is a Majority Winner or one of the choices is shown to be a Condorcet Winner. Because it is not necessary to produce a full matrix this method is easier to count than other Pairwise Condorcet Methods.
Criteria
Simplicity
Benham is easy to understand and is handcountable.
Later Harm
Benham has less Later Harm effect than many Condorcet Methods, but not a lot less.
Condorcet Criteria
Meets Condorcer Winner and Loser, fails the Smith Criteria.
Consistency
In so far as Benham will always elect a Condorcet Winner if present it is more consistent than IRV, when none is present it shares the consistency weaknesses of IRV.
Benham Handouct Process
Top Count the Ballots
A Majority Winner will always also be a Condorcet Winner. Whenever Top Counting the Ballots finds one, elect them.
Determine if there is a Condorcet Winner.
If no Condorcet Winner is found, then remove the choice with the lowest Top Count. Repeat the search for a Condorcet Winner, now ignoring losses to eliminated choices. Repeat until there is a winner.
Implementation in Vote::Count
An equivalent to Benham can be implemented easily through Vote::Count::CondorcetDropping.
use Vote::Count::CondorcetDropping;
my $Benham =
Vote::Count::Method::CondorcetDropping->new(
'BallotSet' => read_ballots('myballots'),
'DropStyle' => 'all',
'DropRule' => 'topcount',
'SkipLoserDrop' => 1
);
my $Result = $Benham->RunCondorcetDropping();
Implemented in Vote::Count::Method::CondorcetDropping.
Note
The original method specified Random as a Tie Breaker, this has the advantage of making the system fully resolvable, but at the extreme Consistency expense of making it possible to get different results with the same ballots.
Your Election Rules should specify a tiebreaker, the default is Eliminate All; the modified Grand Junction Tie Breaker provides the maximum possible resolvability.
Condorcet vs IRV
Determine the IRV Winner.
Follow the General Hand Count Condorcet Process, start with the IRV Winner in your first pairing, and prefer the IRV Winner for the next pairing choice. If the IRV race was decided in a final pairing, don't forget to mark it on your sheet.
If there is no Condorcet Winner or the Condorcet Winner is the IRV Winner, elect the IRV Winner.
Repeat the Condorcet Process. Treat all ballots with the IRV Winner as the First Choice as ballots only for the IRV Winner. The pairings involving the IRV Winner will be unchanged, but other pairings will need to be rechecked (the approval count will change if you are using it to shortcut). If there is a Condorcet Winner, elect the FIRST Condorcet Winner.
This method is implemented in Vote::Count::Method::CondorcetVsIRV and explained in more detail there.
Borda
Count the number of first, second, and so on votes for the choices.
If unranked choices have a default rank other than 0, make sure to tally the values for unranked choices. (Currently Vote::Count only implements 0)
Determine the weight, unless the weighting is fixed it will change with the number of choices.
Multiply the weight of each ranking times the number of votes and total these scores for each choice.
Highest Score Wins.
Beware of variations in weighting rules.
Implemented in Vote::Count::Borda.
STAR (Score Then Automatic Runoff)
STAR is counted the same way as Borda, with two changes.
The weighting is determined by the voter on the Range Ballot, unranked choices are scored as 0.
The two highest scoring choices are then placed into an automatic runoff. Count the top choice between the runoff choices. Ballots which rank neither or rank them both at the same value are ignored.
Implemented in Vote::Count::Method::STAR.
#FOOTER
BUG TRACKER
https://github.com/brainbuz/Vote-Count/issues
AUTHOR
John Karr (BRAINBUZ) brainbuz@cpan.org
CONTRIBUTORS
Copyright 2019-2020 by John Karr (BRAINBUZ) brainbuz@cpan.org.
LICENSE
This module is released under the GNU Public License Version 3. See license file for details. For more information on this license visit http://fsf.org.
1 POD Error
The following errors were encountered while parsing the POD:
- Around line 302:
=cut found outside a pod block. Skipping to next block.