[EM] Re: IRV is summable (a little)

Buddha Buck bmbuck at 14850.com
Tue Sep 4 21:33:09 PDT 2001

DEMOREP1 at aol.com writes:

> D- As usual I ask --- 
> Does ANY choice have a YES majority  (46 YES) ???

Impossible to tell from the ballots.  IRV doesn't provide that
information.  As far as IRV is concerned, that information is not
necessary or useful.  As such, I feel that the question is irrelevant
in a discussion of the mechanics (not merits) of IRV.

IRV may not meet your criteria for a good voting system, and it may
not have the details (such as absolute ratings) that you desire.
These are reasonable criticisms of the merits of IRV.  

However, we were not discussing the merits of IRV, we were discussing
a potential alternative mechanic for IRV.  The original poster posted
an alternative method for computing IRV winners which has the slight
benefit of requiring O(2^N) bits of data rather than O(n!) as usually
claimed.  I think it is an interesting result I'd like to discuss

I find your objection to the merits of IRV off topic in a thread about
the mechanics (or alternative mechanics) of IRV.

To address your comments on the merits, I actually know of one use of
IRV that does use absolute ratings (of a sort) to determine the
winner: The Science Fiction Achievement Awards, awarded by the World
Science Fiction Society.  Their voting proceedure is a nomination
period (where any work deserving recognition may be nominated for an
award) followed by a balloting period, each ballot containing the top
5 nominated works in each category, plus a "No Award" option.  In
order to win, a candidate work must defeat the "No Award" option by a
majority.  For instance, in the most recent award for "Novel", the
voting report in part read:

# Best Novel  --- 885 ballots counted 
# first place: 
# Harry Potter and the Goblet of Fire  326  328  344  396  455 
# Calculating God                      168  169  186  218  301 
# The Sky Road                         127  129  160  199 
# A Storm of Swords                    145  145  155 
# Midnight Robber                       94   94 
# No Award                              25
# Apply the No Award test: 619 ballots rank Harry Potter and the
# Goblet of Fire higher than No Award 91 ballots rank No Award higher
# than Harry Potter and the Goblet of Fire Harry Potter and the Goblet
# of Fire is confirmed as winner...

(See http://www.netaxs.com/~phil2001/hugos/2001novel.html for full

My interpretation of this is that if more people felt that no award
would be a better indication of the merits of this year's achievements
in Science Fiction novels than Harry Potter, then no award would have
been given.

Does this method meet your desire for absolute yes/no ratings?


More information about the Election-Methods mailing list