In message , Roger Mills
writes
On 03/05/2011 12:52, Robin wrote:
The downside to Condorcet, of course, is that there are multiple
rounds of voting - with all the hassle, disruption and expense
involved.

Is that right? I thought it could work by using exactly the same ballot
paper as AV but with the votes counted differently to give a result in
one round in all but a minority (around 1 in 10?) of elections. See eg
http://www.accuratedemocracy.com/c_tally.htm or
http://en.wikipedia.org/wiki/Condorcet_criterion

But I could well be wrong. Or of course outvoted



It looks, from the references, as if you can do it in one go. I must
admit that I'd not heard of this method. Its flaw would seem to be that
it doesn't attempt to measure by *how much* you prefer one candidate
over another. Supposing one voter puts A 1st and B 10th, and another
voter puts A 5th and B 6th, both voters are deemed simply to prefer A
to B, with no qualification.

If you're going to do something like that, wouldn't it be better to
weight each vote according to the preference level, and to declare the
winner as being the person with the highest weighted total?

Isn't that exactly what AV does?
The only thing to remember is that the lowest number carries the
greatest weight.
--
Ian

If you're going to do something like that, wouldn't it be better to
weight each vote according to the preference level, and to declare
the winner as being the person with the highest weighted total?

Isn't that exactly what AV does?

Not quite, if I understand what Roger means. I wonder if he might be
looking more for some form of "range voting" (q.v.) But it's decades
since I had to brief on this stuff and there's a lot more research on it
now and much less brainpower in my head. Or it might me a combination
of AV and range voting, although how that might work is definitely
beyond my brainpower.



--
Robin
PM may be sent to rbw0{at}hotmail{dot}com

On 03/05/2011 16:23, Ian Jackson wrote:
In message , Roger Mills
writes


If you're going to do something like that, wouldn't it be better to
weight each vote according to the preference level, and to declare the
winner as being the person with the highest weighted total?

Isn't that exactly what AV does?
The only thing to remember is that the lowest number carries the
greatest weight.

No! AV gives *equal* weight to all votes - which is the one feature of
it which is perhaps less than ideal. When the second (and subsequent)
preferences of eliminated candidates are re-distributed to the remaining
candidates, they're just added to the existing first preferences. It
could be argued that maybe they should count for less - but they don't
currently count for any *more* - just for the same.
--
Cheers,
Roger
____________
Please reply to Newsgroup. Whilst email address is valid, it is seldom
checked.

On 03/05/2011 16:44, Robin wrote:
If you're going to do something like that, wouldn't it be better to
weight each vote according to the preference level, and to declare
the winner as being the person with the highest weighted total?

Isn't that exactly what AV does?

Not quite, if I understand what Roger means. I wonder if he might be
looking more for some form of "range voting" (q.v.) But it's decades
since I had to brief on this stuff and there's a lot more research on it
now and much less brainpower in my head. Or it might me a combination
of AV and range voting, although how that might work is definitely
beyond my brainpower.



I've not come across Range Voting. What I had in mind was something like
this: When my voting paper is analysed, my first choice gets 1 point, my
second choice half a point, my third choice a third (or maybe a quarter,
depending on the exact algorithm) of a point, etc. Then all the whole
and fractional points for each candidate would be added up - and the one
with the highest aggregate would be declared the winner. So you'd have
to get a lot more lower preference votes to counter another candidate's
first preference vote. In theory, someone could win without getting
*any* first preference votes, provided they got virtually all of the
second preference votes. I'm sure lots of people would object on the
basis that it's no longer strictly one person, one vote.

I'm not seriously suggesting that we adopt this method - but it's an
interesting idea (to me, at any rate!).
--
Cheers,
Roger
____________
Please reply to Newsgroup. Whilst email address is valid, it is seldom
checked.

On 03/05/11 20:14, Roger Mills wrote:
I've not come across Range Voting. What I had in mind was something like
this: When my voting paper is analysed, my first choice gets 1 point, my
second choice half a point, my third choice a third (or maybe a quarter,
depending on the exact algorithm) of a point, etc. Then all the whole and
fractional points for each candidate would be added up - and the one with
the highest aggregate would be declared the winner. So you'd have to get a
lot more lower preference votes to counter another candidate's first
preference vote. In theory, someone could win without getting *any* first
preference votes, provided they got virtually all of the second preference
votes. I'm sure lots of people would object on the basis that it's no
longer strictly one person, one vote.

I'm not seriously suggesting that we adopt this method - but it's an
interesting idea (to me, at any rate!).


There is not perfect method, so the choice of system depends on which
features matter most. PR and suchlike are fine for producing committees and
compromises, not so good for creating clarity about who is responsible.


--
djc

Roger Mills wrote:
On 03/05/2011 16:23, Ian Jackson wrote:
In message , Roger Mills
writes


If you're going to do something like that, wouldn't it be better to
weight each vote according to the preference level, and to declare the
winner as being the person with the highest weighted total?

Isn't that exactly what AV does?
The only thing to remember is that the lowest number carries the
greatest weight.

No! AV gives *equal* weight to all votes - which is the one feature of
it which is perhaps less than ideal. When the second (and subsequent)
preferences of eliminated candidates are re-distributed to the remaining
candidates, they're just added to the existing first preferences. It
could be argued that maybe they should count for less - but they don't
currently count for any *more* - just for the same.

Mark each candidate out of 10, then add up all the scores?

--
Tciao for Now!

John.

On 03/05/2011 21:09, John Williamson wrote:
Roger Mills wrote:
On 03/05/2011 16:23, Ian Jackson wrote:
In message , Roger Mills
writes


If you're going to do something like that, wouldn't it be better to
weight each vote according to the preference level, and to declare the
winner as being the person with the highest weighted total?

Isn't that exactly what AV does?
The only thing to remember is that the lowest number carries the
greatest weight.

No! AV gives *equal* weight to all votes - which is the one feature of
it which is perhaps less than ideal. When the second (and subsequent)
preferences of eliminated candidates are re-distributed to the
remaining candidates, they're just added to the existing first
preferences. It could be argued that maybe they should count for less
- but they don't currently count for any *more* - just for the same.

Mark each candidate out of 10, then add up all the scores?

That would work, as long as one could put zero against any number of
candidates.

--
Old Codger
e-mail use reply to field

What matters in politics is not what happens, but what you can make
people believe has happened. [Janet Daley 27/8/2003]

On 03/05/2011 21:09, John Williamson wrote:


Mark each candidate out of 10, then add up all the scores?


Why not - or something akin to that?!
--
Cheers,
Roger
____________
Please reply to Newsgroup. Whilst email address is valid, it is seldom
checked.

On Tue, 03 May 2011 20:14:52 +0100 Roger Mills wrote :
I've not come across Range Voting. What I had in mind was something like
this: When my voting paper is analysed, my first choice gets 1 point, my
second choice half a point, my third choice a third (or maybe a quarter,
depending on the exact algorithm) of a point, etc. Then all the whole
and fractional points for each candidate would be added up - and the one
with the highest aggregate would be declared the winner.

As in the Eurovision song contest (save that they assign 1..8,10,12)? g

--
Tony Bryer, Greentram: 'Software to build on' Melbourne, Australia
www.superbeam.co.uk www.eurobeam.co.uk www.greentram.com