[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Splitting Aye/Nay from vote tallying (Was: Constitutional amendment: Condorcet/Clone Proof SSD vote tallying)

Anthony Towns wrote:
> On Mon, May 26, 2003 at 08:45:31PM -0400, Andrew Pimlott wrote:
> > I _think_ the same basic strategy applies:  Rank the non-default
> > options sincerely, then insert the default option after your "lesser
> > of evils" option.
> That doesn't work. Suppose there are three options, and everyone does this.

i can see where the basic flaw is now, and why, no matter what we do, we
cannot fix this problem via vote-tallying.

we are making our votes do double duty. we are simultaneously trying to
rank in order our preferences, and say ``Aye'' or ``Nay'' to any
particular option.

this is affecting the ``quorum'' and super majority portions.

i believe it was anthony towns that pointed out that failure to get the
required seconds prevents an option from ever appearing on the ballot. i
hereby state that that is a straw-man argument: the requirements to
appear on the ballot are orthogonal to the vote tallying mechanism.

  example: in order to be president of the USA, you have to be 1) a
  natural borne citizen, 2) 35 years of age (i do not know if that is at
  time of inauguration or at time of election).
  to appear on the ballot, you have to have a certain number of people's
  signatures backing you. this number is really high, and varies state
  to state. to speed things up, political parties have already gotten
  the required signatures, and apply those signatures to a single
  candidate (usually, last year the arizona libertarian candidate was
  different from the other 49 states.  fun, huh? bet you were unaware of
  that little election anomaly ;)

i feel certain that no one would try to argue that being 34 years of age
is a fault of the election tallying method (plurality or condorcet), or
that anyone is going to assert that every natural-borne citizen that is
35yrs or greater MUST appear on the ballot. that would be ludicrous.

so: the need to get seconds (normally, 5) is in no way, shape, or form a
reflection of the vote tallying method.

the problem: our vote tallying method is doing double duty.
solution: split it out.
how: i do not know, yet, but one possibility follows. i am making this
up as i go along, so comments welcome.

for the ``quorum'' requirement: this is easy.
require X number of seconds with each anti-second counting against the
totally number of seconds.

perhaps that could be applied to super majority options also: require
that the number of seconds not only be really high, but also has to beat
the number of nay's by that ratio.

option a: no super majority requirement, requires 10 seconds
option b: 2:1 super majority requirement, requires 20 seconds
option c: 3:2 super majority requirement, requires 15 seconds
option d: 2:1 super majority requirement, requires 20 seconds

option a gets 15 seconds, 6 rejects. aye-nay<10; stricken from ballot
option b gets 40 seconds, 20 rejects. aye:nay ratio >= 2:1, aye-nay>=20;
   appears on ballot
option c: gets 20 seconds, 15 rejects. aye:nay ratio < 3:2. stricken
   from ballot
option d: gets 15 seconds, 5 rejects. aye:nay ratio >= 2:1, aye<20;
   stricken from ballot

so the ballot would look like this:

Option B
Option C
Default Option

the vote can then be run, and counted in a pure condorcet/cloneproof SSD
method. under such a scenario, i am uncertain if a default null-option
would be required, but it certainly won't hurt anything.

it is true that a single voter, if the only voter, could decide the
entire course. however, with the high number of seconds required (i'm
assuming that the number of seconds required would be increased)
sufficient buy-in has already been established.

i also used the base buy-in value, and scaled it per the super majority
requirements. in our current system, at no time may a 2:1 or 3:2
super majority be required. these numbers are simply illustrative.

this takes the burden of super-majority and buy-in OFF OF the vote, and
puts it on the appear-on-ballot process.

the interesting case would be the buy-in requirements in a small voting
populous, such as for the tech committee. a buy-in of two would still
sound reasonable: one person submits, another seconds, a second
seconds. if the CTTE is 4, then if the fourth person rejects, *poof*
goes the option. if the fourth person abstains or seconds, then that
option would appear. for 8 members, it could go on with additional
seconds and rejections.

this takes out all possibilities of strategic voting, and applies the
strategy to the proposal stage. at least the vote tallying method
remains untainted  :/

i do not think our current method of deterring R would quite work:
3/2*N**.5 (where N is the Number of Developers). that would require a
supreme effort of people to second a proposal. the proposal we are
discussing has only gotten around 8 seconds, that is not even Q.

5 is probably too low. .5*N**.5 could possibly be too high. where does
the middle ground lie?

or should the above be scrapped, perhaps for another, better, method
that separates out the aye/nay stage from the vote-tallying stage?


Love your freedom? Thank a vet.  http://www.gisearch.com/thankavet/
You are welcome.

Reply to: