[EM] My Bucklin multiwinner method turned more sequential

[Ted Stern]

Hi Kristofer,

I am very interested in PR multiwinner methods, especially those that
use ER-Bucklin.

However, I have a hard time following your logic.

Would it be possible to work out a relatively simple example using a 3
winner election, a Droop-like quota of 25% (just to make things easy),
and two factions, one with 3 candidates and 55% of the vote (thus
winning two seats), and another with two candidates and 45% of the
vote (thus winning one seat)?

Alternatively, you could use Warren Smith's 'real world' example with
9 seats and an 'Easy' Droop-like quota of 4 votes (10% of 39 votes =
3.9, plus 10% of one vote), to compare to other methods that can work
with range ballots.

  http://rangevoting.org/June2011RealWorldRRVvotes.txt

My implementation of Bucklin Transferable Vote finds the following
winners for that example:

  {106,102,109,101,103,108,105,110,116}

Ted

On 10 Feb 2012 14:05:36 -0800, Kristofer Munsterhjelm wrote:
>
> On 02/10/2012 11:02 PM, Kristofer Munsterhjelm wrote:
>> We can add a candidate C_k to PC if there exists a subset (coalition)
>> that supports at least k+1 candidates, where k is the cardinality of the
>> intersection of PC and that coalition, and that coalition also contains
>> C_k.
>
> Oops, seems I reused a letter there. This should be:
>
> "We can add a candidate C_i to PC if there exists a subset (coalition)
> that supports at least k+1 candidates, where k is the cardinality of
> the intersection of PC and that coalition, and that coalition also
> contains C_i."
>
> I.e. the k in C_k had nothing to do with the cardinality of the
> intersection.
>
> ----
> Election-Methods mailing list - see http://electorama.com/em for list info
>

-- 
araucaria dot araucana at gmail dot com