Manually counting IRV (was RE: Truncated preferences ok for Condorcet (was Re: [EM] Ignorance))

[James Gilmour]

Steve wrote:
> > Steve Eppley had written:
> > > That's true, except there's no need to count the ballots in 
> > > each pile.  It's quicker to compare the heights of each 
> > > pile by eye to see which is shortest, unless the shortest 
> > > piles are very similar to each other in height, in which 
> > > case you might want to count the shortest piles if the 
> > > group is concerned with accuracy.

I replied: 
> > Now I understand why there are problems with public elections in
> > the USA!! Nothing other than full counting of every bundle of
> > ballot papers would ever be accepted here in the UK. Electoral
> > Reform Society procedure requires a sort and a check sort followed
> > by a count and check count on each bundle (usually 50 or 100
> > papers, depending on the size of the election). James 

Steve wrote:: 
> Sure, there will always be people who stand to gain after a 
> close election is recounted, since the result might change, 
> and they push for such scrutiny and the rest go along, 
> lacking an understanding of why it's not important. (It's 
> like that here in the US too; my comment doesn't explain 
> the problems here.  Ballots aren't counted manually in the 
> US except in recounts after close elections, and in those 
> cases they are supposed to count accurately; otherwise why 
> recount?) 

My reference to "count and check count" was not to a 'recount'  -  that's something completely
different, when a result is challenged.  Nearly all votes in public elections in the UK are recorded
on pre-printed ballot papers and nearly all ballot papers are sorted and counted by hand.


> But that doesn't mean accuracy is important, yet James 
> hasn't bothered to try to explain why he thinks it is.
> Particularly with a method as screwed-up as IRV. :-)

I nearly fell off my chair when I read this statement.  I am amazed that anyone should think it
necessary to explain why "accuracy" is important in a public election, no matter what the voting
system (most of our UK elections are still FPTP).  Of course, the level of accuracy for which you
must aim should, ideally, be defined.  In practice, in UK public elections it is defined what you
get by manually counting every ballot paper and all of the candidates and their agents being
satisfied with the result.  If there is a challenge which the Returning Officer considers
reasonable, the papers are counted again, and sometimes again if the first recount reveals a "large"
discrepancy from the original count.

I just cannot imagine an election in which the Returning Officer said "She's the winner  -  because
I think there are more papers in her pile than in all the other piles put together".  I suppose you
could weigh the piles of papers, but I am fairly sure there would be objections based on the
possibility of variations in the weight of sequences of ballot papers (quite apart from the
possibilities for doctoring the papers to increase their weight).  Height of pile would certainly
not be acceptable because of the variations in packing density once the papers had been handled.
I'm still gob-smacked!!


> (I like STV prop rep, though.  STV fans might want to check 
> out Tideman's variation of STV, described in one of his 
> unpublished papers available online at his website at 
> Virginia Tech.)

At least we can agree on the merits of STV-PR!!  That's what we are campaigning for right now for
local government (council) elections in Scotland (see www.fairsharevoting.org).  Yes, and it is
almost certain that all those ballot papers will be sorted and counted by hand for the first STV-PR
elections in 2007.  Computerisation is NOT part of our campaign.  Practical campaigners climb one
hill at a time.  So we won't be using Tideman's variation nor Meek's version of STV.
James