Richard
I suppose the simple reason why AMS (MMP) has not been programmed is
that no-one would be stupid enough to want to use AMS (MMP) for
private elections. It was the use of STV-PR in private elections
that drove the initial programming. The STV calculations are simple
enough (at Primary 7 school level), but it is the repeated sorting
of the ballot papers in ever more but smaller parcels that is both
fiddly and time-consuming, even for Gregory Method rules. (Much
more fiddly and time-consuming for WIGM rules, and impossible by
hand for Meek STV for anything other than a Mickey-Mouse demo count.)

So far as I am aware no country publishes the actual ballot data
from AMS (MMP) public elections. New Zealand has published a "Split
Vote" analysis from which it would be possible to reconstruct the
data files needed, but unhelpfully, in recent years they have
published that analysis only in percentages, not actual numbers of
ballot papers.

The full ballot data for the 2007 Scottish Parliament AMS elections
can be reconstructed from the cross-tabulations published for each
constituency by the Scotland Office last year. That is unlikely to
be repeated because we shall revert to two separate AMS ballot
papers for the next Scottish Parliament elections and the ballot
papers are likely to be counted by hand (as in 1999 and 2003).

Once you have the total numbers of votes for each candidate and each
party, it is a simple spreadsheet problem. However, the
spreadsheets would have to be unique for each set of AMS (MMP) rules
because the implementations differ in several significant respects.
For example, the are quite major differences among the versions of
AMS used in Germany for the Federal elections and for the various
Land elections. In Scotland and Wales, AMS is uniquely
regionalised, but there are differences even between these two sets
of rules. Of course, if it can done on a spreadsheet, a computer
program can be written to process the same data, and it would be
possible to include every known variation in one program. But there
is (understandably) no incentive for anyone to write such a
program - I am surprised Iain McLean should have made the quoted
remark.

James



--- In stv-voting@yahoogroups.com, "Richard Lung" <voting@...> wrote:
>
> James,
>
> My source for AMS not having been programmed, a couple or so
decades ago,
> came from a book (edited?) by Iain McClean. (his name might be
spelt
> slightly differently) You'll know who I mean, the Oxford prof. who
took over
> from David Butler.
> McClean said: By the way no program had been written for AMS yet.
> So, he seemed to think it becessary at the time. I wouldnt know.
>
> I gather from your reply that no AMS program has been written but
that
> spreadsheets will do, at least for Scottish AMS.
> However, I take it that there is no program in the sense of a
general
> procedure
> for tackling all or most of the contingencies thrown up by this
anomalous
> system
> of Additional Members as variously used thru-out the world..
> I mentioned the dual partisan count both of list votes and another
totaling
> of party votes
> from the single member system. As you know, there are other
partisan
> problems
> such as the proportion of list seats to singles; the size of the
small party
> threshold;
> over-hang seats ( or is it hang-over seats?); closed lists or open
lists
> (any number
> of varieties); dual candidature or not.
>
> An AMS program could count the votes in all these and no doubt
more ways to
> arrive at many different results from exactly the same votes cast.
> Also for any given result, AMS gives, its proportional aspect
would be
> transformed
> to an unreformed FPTP majoritarian type result, if it
> were revealed that one of the parties was a dummy for another to
boost its
> representation disproportionately - as I believe you were the
first to tell
> me.
>
> Proportional partisanship can be a deceptive duplication of
representation,
> even if
> there was no conscious attempt to make it so:
> People often get furiously at odds over quite small differences of
opinion,
> that
> make mountains out of molehills like Swift's big-enders vs little-
enders.
> Such
> doctrinal divisions on partisan lines may monopolise the political
issues,
> that
> representative elections could liberate.
>
>
> Richard.
>
> ----- Original Message -----
> From: "James Gilmour" <jgilmour@...>
> To: <stv-voting@yahoogroups.com>
> Sent: Sunday, February 15, 2009 12:10 AM
> Subject: [stv-voting] Re: Announcing OpenSTV 1.3
>
>
> --- In stv-voting@yahoogroups.com, "Richard Lung" <voting@> wrote:
> >
> > A couple of decades ago, AMS hadn't been programmed. I suppose
that
> > deficiency has long been made good.
>
> Richard
> I may have missed something, but I didn't see anything about AMS
(aka
> MMP) on the OpenSTV website - indeed, it would have been a
surprise
> to find it there as it does not use preferential ballots.
>
> But in any case, why would anyone want to write a "program" for AMS
> (MMP)? It all tumbles out with a comparatively simple spreadsheet
or
> set of spreadsheets for the regionalised version we use in
Scotland.
>
> James
>