[EM] Tideman & Schwartz set

[MIKE OSSIPOFF]

Markus said:

I suggest that --when you consider the Schwartz criterion important--
then you should calculate the Schwartz set and choose the TBRC in
such a manner (1) that all Schwartz winners are ranked ahead of all
other candidates in the TBRC and (2) that the other rankings are
filled by random ballots so that independence from clones is met.

I reply:

Ok, now I understand why you said to choose the TBRC after the
election and before the count. And sure, by making it part of the
count rule that we determine the TBRC that time in a way that ensures
that the Schwartz set is ranked over all else, we can make RP always
choose from the Schwartz set even in small committees.

But if we base the TBRC on the Schwartz set, it's more accurate to
say that we make the TBRC _during_ the count, rather than before
the count, since the Schwartz set is determined during the count.

Another thing that's been made clear to me, by the Blake quotation
that you posted is why the TBRC is important to RP advocates: It's
an attempt to save RP in small committees, to try to ease its big
implementation problems in small committees. It doesn't seem to me
that it eases the problem much, judging by that quoted passage.
It still looks as if RP is unavoidably very awkward in small committees.

Much better to use CSSD, which doesn't need any of those elaborate
efforts, even in small committees. RP can't really be recommended
for small committees.

One could still use a TBRC, most likely the committee chairman's
ballot, if one wanted to, to solve ties in postal and e-mail ballotings,
where the voters can't be present to verify the legitimacy of
randomizing procedures in RB. But in CSSD, a tiebreaker is something
that can be needed once, at the end of the count, not here and
there during the count.

There's nothing wrong with the simple
procedure, for CSSD, of simultaneously dropping several equal
defeats instead of using a tiebreaker. I know you suggested looking
at losing-votes when 2 defeats have the same winning-votes, and
that makes sense, but it complicates the program unnecessarily.
Simplicity of the algorithm seems more important.

Mike Ossipoff




_________________________________________________________________
MSN Photos is the easiest way to share and print your photos: 
http://photos.msn.com/support/worldwide.aspx