[EM] Sticks-in-the-mud

[Diana Galletly]

On Fri, 13 Aug 2004, Adam Tarr wrote:

> I'll be "quoting" the text of the link from here out.
>
> >Condorcet, however, would tend to produce consensual results, in that the
> >conclusion tends to be support for the proposition which has the broadest
> >support, even though other propositions may have majority support.
>
> I find this statement hilarious, because it is a garbled version of a
> specious criticism.  It is as if someone played the children's game
> "telephone" with Don Davison on one end, and this was the statement that
> came out the other end.

The horror is that the only evidence they based all this on is ... my
own :-/

This is the document I submitted to this review committee:

http://www.chiark.greenend.org.uk/~galletly/university/evidence.pdf

(and I know it's awfully heavily laboured and repetitive, but I knew that
the committee was full of people who had no knowledge of voting methods;
indeed there was only one mathematician, who became an administratot about 30
years ago)

> I would challenge them to produce a single example where a method that has
> majority first-place support loses in Condorcet voting.  They won't be able
> to, because it is impossible.

They don't understand what they're talking about. Really, they don't. I
spent an hour trying to talk to them in person and their concerns were
not about fair voting systems, but about how much it would upset people
to change system ...

> >Condorcet rules
> >
> >For the voter action is the same as for STV. The votes are counted
> >differently, however, with the successive exclusion of least popular
> >pair-wise combinations,
>
> This betrays an ignorance to the nature of Condorcet methods.  I assume
> this is a somewhat garbled description of sequential dropping.  Not only
> are they ignoring other methods such as Ranked Pairs, but they are ignoring
> the more fundamental formation of a pairwise matrix, and determination of
> whether a Condorcet winner exists.

I did explain this to them; however, as stated above, I'd have needed months
to explain stuff. CpSSD was what I proposed as resolution should one end up
having a cycle.

> This statement is meaningless without defining "separate support".  It is
> possible for the candidate with the most first place support to lose, but
> this is true of every method except FPTP.

Worse, they think that "broadest support" isn't something one might want
in a self-governed academic community!?!  The administration actually *want*
"warring" factions at each other's throats? Ummmm.

> >It can be complicated to count.
>
> Again, betraying an ignorance to the counting methods.  For a non-trivial
> number of ballots, counting a Condorcet method election is far easier than
> counting a IRV election.

Something I explained to them when they met me.  None of them were even
vaguely computer-literate...

> >  It was put to the Review Committee, that STV is particularly suitable in
> > voting in elections for representative bodies, such as the University
> > Council, where it is desirable to represent different strands of opinion,
>
> Here I agree.  Condorcet IS a bad method for filling a legislature, unless
> you use one of the more complicated multi-winner variants.  It is far
> superior to STV in single-winner elections.
>
> >and that Condorcet is particularly useful in voting on policy matters
> >where it is desirable that the result should be one of broad consensus,
> >rather than of representing the largest single body of opinion.
>
> They keep saying this.  It remains a terribly ignorant statement.  STV does
> not reliably represent the "largest single body of opinion".  If that is
> really what they want, they want plurality.

I fear this is somewhat my fault.  I was arguing against IRV here.  I may
not have worded my statements appropriately.  If you read my PDF file,
perhaps you could be kind enough to tell me what I ought to have said instead?
(It will come up for public debate this autumn).

> >  does indeed seem applicable to University circumstances, it is not
> > necessarily the case that the best policy or legislative solution would
> > be achieved consensually,
>
> How should policies solutions be achieved, then?  Erratically?

... this is where they attempt to come up with their own thoughts, rather
than misrepresenting mine ...

> >  and the case for adopting Condorcet rules is not, therefore, made out on
> > that account.
>
> I can see why you're unimpressed, Diana.  They seem to think they
> understand Condorcet and STV, but in reality they're obviously ignorant to
> the differences between these methods.

Given that I'm maybe one of three people in a community of 15000 who really
understands ... (and was the only one of those three to stand up to be counted
at this point in time)

... yeah, I'm unimpressed.

Diana.