Looney Labs Icehouse Mailing list Archive

Re: [Icehouse] IGDC rankings idea

  • From"Carlton Noles" <carlton.noles@xxxxxxxxx>
  • DateWed, 20 Feb 2008 17:38:18 -0500
In my conceited dogmatic opinion, the only way to get better results is with a larger sample (of voters). Any reasonable system will produce decent results that way. Otherwise the smaller the number of voter the greater impact each voter has. Plain and simple. Weighted averages of some sort or another may help offset this  somewhat but in general little changes.

On Wed, Feb 20, 2008 at 5:28 PM, Doug Orleans <dougorleans@xxxxxxxxx> wrote:
Dale Sheldon writes:
 > On Wed, 20 Feb 2008, Jeff Zeitlin wrote:
 > > Suppose the IGDC says 'ten points per choice on the ballot'.  In the
 > > example above, each voter can share 40 points among the four choices.
 > > So, the first voter goes with Z=20, I=12, M=6, S=2.  The second voter
 > > goes with S=12, M=11, I=9, Z=8.
 > These sorts of methods, if you presume strategic voting, tend toward every
 > voter giving each entry either the maximum or minimum number of votes,
 > (i.e. break down as approval voting); and you didn't even list a maximum
 > other than the number of points, so this will break down to plurality
 > voting (i.e. each voter gives all points to a single candidate.)

This may be naive, but I think the fact that the votes are public will
help to discourage strategic voting.  A truly exaggerated assignment
of points (or skewed ratings using range voting, i.e. all 10s and 0s)
will stick out like a sore thumb.

Icehouse mailing list

Carlton "Kermit" Noles
"Joy Multiplies when it is shared among friends but grief diminishes at every division. That is life"
-Drizzt Do'Urden
Want your own GMail Account?
Ask Me for an invite.