[EM] Re: Condorcet package

[Ted Stern]

On 23 Feb 2005 at 20:25 PST, Kevin Venzke wrote:
> Dave,
>  --- Dave Ketchum <davek at clarityconnect.com> a écrit : 
>> Try:
>> 5 a>b
>> 2 b>a
>> 6 a=b
>> By not counting the = we have 5:2
>> With counting them    we have 8:5
>> 
>> The > and < determined that a gets 3 more votes than b - to me, strength of
>> a's win over b.
>
> It's called "Margins" when defeat strength is measured as the vote
> difference.  With WV, the defeat strength is the number of votes for the
> winning side; in this case, it would be 5 votes.
>
>> I have the = making a have a total of 8 votes, useful (I think) in deciding
>> how to untangle cycles, should that be needed.
>
> A potential problem is that it becomes unclear how many people actually
> voted a>b.
>
> Kevin Venzke
[CC: Steve Eppley, Markus Schulze, Rob LeGrand, election-methods list]

Hi Dave,

As Kevin has pointed out, by counting a=b as a>b + b>a, winning votes (wv)
acquires some characteristics of relative margins (rm).  With your tabulation
rule, the margin doesn't change, but the winning vote totals do.

IMO, 'a=b' indicates equal preference, equivalent to abstention from the
pairwise election.  In other words, it's a vote to consider that particular
contest AFTER all other considerations.  That should be true whether one ranks
defeats using wv or rm.  

I have noticed this tabulation disagreement before, e.g. Rob LeGrand's voting
calculator (http://cec.wustl.edu/~rhl1/rbvote/calc.html).  I've looked around
briefly for an authoritative differentiation between the two tabulation rules,
but aside from discussion on this list I haven't found one, hence the CC's.

It would be nice to get this misunderstanding cleared up.  Any method
description or software that uses a non-standard interpretation of equal rank
tabulation should clearly indicate that it is doing so.  If it is confusing to
e-m list readers, imagine how confusing it would be to voters!

Ted
-- 
Send real replies to
	ted stern at u dot washington dot edu

Frango ut patefaciam -- I break so that I may reveal