[EM] Election Calculator additions: equal rankings IRV

[Brian Olson]

On Jun 23, 2004, at 12:42 PM, Kevin Venzke wrote:

> --- Brian Olson <bql at bolson.org> a écrit :
>> Whee, a bit more hacking and now Condorcet is implemented with both
>> Winning-Votes beatpath and Margins beatpath.
>>
>> A recent example can be seen as:
>>
>> *49 A
>> *24 B
>> *27 C>B
>>
>> http://bolson.org:8080/v/et?vrr=-clist&if=-
>> cname&cand=3&data=*49+A%0D%0A*24+B%0D%0A*27+C%3EB
>
> I followed this link.  There are an awful lot of methods supported.
>
> The Bucklin result is very strange:
>
> Bucklin: B 75.5, C 63.5, A 61
>
> I have to admit I didn't think ER-Bucklin(fractional) could even be 
> defined.

Bucklin can be expressed as:
In round N add the Nth ranking votes to the tallies for the choices. 
Starting at 1, repeat until a choice's tally is greater than 1/2 the 
number of voters. The highest tally is the winner.

I added to this:
If a voter has not marked a Nth rank choice, distribute 1 vote across 
the choices which the voter has not ranked.

I think this modification deters truncation.
The voter who just votes A as 1st choice out of A,B,C, in the second 
round 1 vote is split between B and C.  It is also flexible enough to 
vote A=1st, D=4th. Then in the second and third rounds that vote is 
split between B and C but not D.

I added Bucklin truncative so you can compare it to Bucklin 
distributive. In that example it didn't change the winner, but did 
switch 2nd and 3rd.

Brian Olson
http://bolson.org/