[EM] To MIKE OSSIPOFF, re: Approval, CR, & IRV

[Bart Ingles]

Bill Lewis Clark wrote:
> 
> In any event, from yours and other posts (as well as some additional web
> research) I think I understand some of the assumptions required to
> demonstrate the strategic equivalence of AV and CR:
> 
> (1) (Nearly) ALL voters under AV must use their optimal strategy.
> (2) (Nearly) ALL voters under CR must use their optimal strategy.

I believe this is implied by the term "strategic equivalence".


> and possibly (unless this is already covered in assumption #1:)
> 
> (3) (Practically) NO voters under AV can use randomized or weak optimal
> strategies.

Not really.  If randomized strategy were optimal, then it would be
reasonable to use this as a basis for comparison.


> If any significant portion of the voters deviate from their optimal
> strategies, then the optimal strategy CHANGES for those trying to follow
> their own optimal strategies.  This is an important point, because it
> completely destroys the equivalence between AV and CR.
> 
> ...and since it seems *very* likely that (in real-world situations) a
> significant number of voters will deviate from their optimal strategies,
> in practice AV and CR are NOT strategically equivalent.

But once you know how those non-optimal voters will deviate, your
optimal strategy is still the same for AV and CR:  give either the
maximum or minimal approval to each candidate.


> Kucinich and Dean both support IRV in particular -- which I know you hate,
> Mike -- but I still think that *any* election system change is a step in
> the right direction.

Interesting premise:
1. Any election system change is a step in the right direction.
2. Abolishing elections is an election system change.
3. Therefore, abolishing elections is a step in the right direction. 
;-)


> IRV certainly has its problems (and I didn't address those points in your
> post, precisely because I pretty much agree with you there) but given the
> existing political climate (no "big" third-parties) I think it would still
> do better than plurality.

Except that it would help maintain the situation of no big third parties
(i.e. the "Duverger effect").


> If Condorcet or Approval (or whatever else everyone thinks is better) had
> as much popular support as IRV, I'd be singing a different tune -- but as
> things stand, I think supporters of *any* alternative voting system would
> be best served by letting IRV proponents pave the way.

Speaking for myself (as a former IRV advocate), I'm not really
interested in expending energy to defeat IRV.  But since I consider it a
"non-reform", I have no qualms about criticizing it in order to promote
a better system, or to educate the public about election systems in
general.

Bart