[EM] Decision juries - notes and responses

Craig Carey research at ijs.co.nz
Sat Aug 3 12:55:39 PDT 2002

At 02\08\03 21:02 +1000 Saturday, Peter Maxwell wrote:
 >> found that one site, "Innovations in Democracy", has a number of
 >> interesting links.
 >Sure does, though there is one closely related idea that it (apparently)
 >misses: "Demarchy; random selection of decision makers" at

A page linked to from that demarchy.html page, says:

 >Random selection is also called the lot system, the jury system or sortition. >Demarchy can also be called statistical democracy.


Approval gets so many winners wrong (different from (USA's and) the CVD's IRV
that it be categorised as being a random method.


PS. My comment about Hadamard matrices is somewhat irrelevant, but the
matrices do have as many "+1"s and "-1"s, corresponding in a clear way
to how in 1 winner Approval elections with 50 candidates, it would be
maximally empowering (or minimally risky) to fill in about half of the
checkboxes. E.g. if green plants are being designed and 10 changes are
made to their genetics (maybe providing a resistance to insects), then
less than 2**10 (1024) plants are needed, if the theory of fractional
factorial designs in statistics are used (which uses Hadamard matrices).
Possibly one enhancement might neutralise another. In a 3 winner
Approval election where voters mainly wanted candidate A and B, they
could easily split into two similarly sized groups, those that voted
for candidate A but not candidate B, and those that voted for
candidate B and not candidate A. For an individual to vote for both
could the vote to lose its power. 10 candidate checkboxy Approval
evidently has a vote-splitting problem that I was not seeing complained


Also, Approval maybe should fail a test of monotonicity, since this
  can happen:

    A wins with the added paper (ABCDEF)
    A loses with the added paper (ABCDEFG)

Unless someone objects, I'll say that monotonicity is not defined for
methods that allow that to occur. No method is passed by a rule that
is not defined. Thus Approval does not pass. Anyway, monotonicity seems
to be an idea that is unimportant in the theoretical maths of
preferential voting (it does not allow the trailing right hand side
preference to be disordered. Whoever designed it in that way, i.e. to
limit changes on the right (all the way to having a non-essential
idea), may have also expected similar restrictions for the leading
preferences on the left. Anyway, Approval is said to be not a
preferential method. Mononoticity may, in any case, be undefined for
methods that are not preferential voting methods (specifically, the
Approval method).

Craig Carey
Voting science: http://groups.yahoo.com/group/politicians-and-polytopes

For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em

More information about the Election-Methods mailing list