[EM] range voting - strength of feeling

[Warren Smith]

>Gilmour:
If we have a fixed range scale, say 1 = least preferred, 1000 = most preferred, it is perfectly obvi
ous that every voter
will mark 1 for his/her least preferred candidate, 1000 for his/her most preferred candidate, and sc
ore the others
somewhere between the extremes.  Then all voters will contribute equally.  No problem.

But that was NOT what was actually suggested.  It was suggested that the scores used by the voters c
ould (or should)
reflect the differences in their strengths of feeling, ie without obvious reference to any fixed sca
le.  So one voter
might range 1 to 10, another 1 to 100, and another 1 to 1000.  It was then suggested that these abso
lute differences
should be taken forward into the counting method and that that would maximise the social utility of 
the result.  Well it
might - but it would be undemocratic.


--WDS:
JG, you have misunderstood me.
I have never advocated that "the scores used by the voters could (or should)
reflect the differences in their strengths of feeling, ie
without obvious reference to any fixed scale."

I have always advocated using a fixed scale such as 0-to-99.

I agree if any real number, with no constraints, were permitted as a vote,
the resulting system would be completely ridiculous for any human use
(although it might be ok for robot use, and it in fact is what human judges
of boxing matches do)

So with this massive misunderstanding cleared up (and I apologize if I were responsible
for it) I think we are presumably now in agreement.
wds