When are "Real World Failure Rates" Useful?


Continue from


never said you could.

there are three differences in my method from traditional Stv, not one. (the other two are using the correct quota instead of an unbalanced quota, and using a full ordering (all permutations)).

in the single winner case my method reduces to a borda count, which doesn’t have condorcet failures.

in the multi winner case my method also doesn’t have condorcet failures, while multi winner borda would because it doesn’t transfer unused votes.


Re: "in the single winner case my method reduces to Borda count, which doesn’t have condorcet failures"
Uh yes it does…?
5 A>B>C
3 A>C>B
7 B>C>A
Condorcet winner is A.
Borda: A = 16, B = 19, C = 10.


I got corrected already on another thread. My mistake. Borda Count has condorcet failures. (and thus by implication so does my method.) thanks for the correction.


Anyway, Condorcet isn’t particularly applicable for proportional representation methods. In fact in some cases it’s not possible to both elect a Condorcet winner and maintain proportionality. See: https://www.rangevoting.org/WoodallP5.html (I know he used Droop quota, but the example he gave could easily be modified for Hare.)


Condorcet philosophy does indeed apply to multi-winner elections as well as single-winner elections. The only difference is that in multi-winner elections, the goal is electing the “Condorcet” election outcome rather then any individual Condorcet winner, where different possible election outcomes are compared pairwise to other possible election outcomes.

Two methods that do this are Schulze-STV and CPO STV:

When comparing the outcome in which A wins a seat and B wins a seat, to the outcome in which A wins a seat and C wins a seat, you compare how “much” voters prefer outcome AB to AC. However, I put “much” in quotation marks because in order to maintain proportionality, voters who expressed more support for A don’t have their preference of AB to AC counted as strongly because if the outcome came down to AB and AC, voters who supported A are better off, and proportionality emphasizes giving representation to those who are insufficiently represented.


Incidentially, this approach could be used to extend DAC to a multiwinner system by (1) changing the order in which sets are considered to the same as DAC and (2) eliminating candidates from contention for the current seat the way DAC would (but you would reintroduce them when you fill the next seat). Each time candidates are eliminated this way, transfer their votes to the next highest choice remaining.