Goldfish (single-winner method)
Blake Cretney
bcretney at my-dejanews.com
Thu Aug 27 21:42:48 PDT 1998
--
On Thu, 27 Aug 1998 00:40:07 Norman Petry wrote:
>Blake,
>
>You wrote:
>
>>>My previous posting turns out to be incorrect as an implementation of
>original Tideman. However, I still like it better than Goldfish 0.2, so I
>am calling this algorithm Goldfish 0.3. I also like it better than Tideman,
>and it passes the GMC test from the
>
>You're right, it's definitely not Tideman. In fact, it's much better than
>Tideman. I think what you've discovered with Goldfish 0.3 is a very
>efficient algorithm for the Schulze method.
>
That's my hope. Anyway, I thought I would provide an
example of Goldfish at work. Here's how Goldfish 0.3
handles the Tideman bad example for GMC.
100 voters.
B>A 61
A>D 62
D>B 60
B>C 1
C>D 2
C>A 59
Here's how Goldfish handles this example
A B C D
A X 62
B 61 X 1
C 59 X 2
D 60 X
I find the highest value among non-eliminated rows, that's the 62
in A,D. I star the D row to show it has been eliminated. I update
the A row by copying any greater values from D. I treat X's as
high values.
A B C D
A X 60 X
B 61 X 1
C 59 X 2
* D 60 X
A B C D
A X 60 X
B 61 X 1 The new high value is 61
C 59 X 2
* D 60 X
A B C D
* A X 60 X
B X X 1 X The high is 59. Remember that starred
C 59 X 2 rows can still be used for merges
* D 60 X they can still lose, they just can't win
A B C D
* A X 60 X
B X X 1 X The 60 is higher than the 1, so C wins
C X 60 X X
* D 60 X
A B C D
* A X 60 X
* B X X 1 X
C X 60 X X
* D 60 X
I hope this will help anyone who is having trouble with the method.
-----== Sent via Deja News, The Discussion Network ==-----
http://www.dejanews.com/ Easy access to 50,000+ discussion forums
More information about the Election-Methods
mailing list