[EM] Can ranking improve on Approval? MDDA strategy.

[Dave Ketchum]

On Tue, 18 Oct 2005 01:03:58 +0000 MIKE OSSIPOFF wrote:

Can happen that I desire electing Tom and hate Harry. Further, I am not 
sure Tom can win so, in case he loses (and only if he loses), I want to do 
what I can to make sure Dick gets in as my second choice.

> For me, as an individual voter, in the existing political system, Approval 
> would be fine. It's for other voters that I'd want the benefits of a rank 
> method. It's for the voters who, with Approval, would insincerely vote Kerry 
> equal to Nader.


You seem certain that you will never feel as I describe above, so you 
consider Approval acceptable.

Many, coming from the Plurality world, would see no reason to care or 
would see Approval as an acceptable step ahead.

MANY expect to see what I describe above in real life, and therefore 
consider Approval unacceptable.

Also, for the day when we find Tom and Dick equally acceptable, many 
variations of Condorcet and other methods permit equal ranking, leaving 
Approval no advantage in these cases.

Also, voting Kerry equal to Nader does not have to be insincerity under 
Approval - it is the sincere vote when your major desire is to be rid of Bush.

SUMMARY:  This discussion gets complex at best - can we please drop 
Approval out of it.

> 
> Of course hopefully they wouldn't keep doing that in Approval, when Nader 
> outpolls the Republican, or when they find out what a Democrat is.


This really gets nowhere - there are often trios of candidates which 
voters see fitting into separate categories.

> 
> The need for better voting systems has to do with things that we don't want 
> voters to (believe that they) have to do.
> 
> We don't want them to vote an undesirable over their favorite. That's the 
> worst, because they aren't showing any support for their favorite. 
> Fortunately it's possible to guarantee that that won't be necessary.
> 
> Less important, we don't want them to incincerely vote an undesirable equal 
> to their favorite. Because the freedom from that need isn't essential, 
> Approval & RV would be good to have. But wouldn't it be nice if, even in the 
> 1st election with the new method, voters didn't feel comelled to vote a 
> lesser-evil equal to their favorite? That's how a rank method could improve 
> on Approval. But it would have to be a good rank method. It would have to 
> meet FBC, to avoid the worst kind of LO2E giveaway.
> 
> Of course that latter benefit cannot be guaranteed, but one can still do 
> much to make people willing to vote Kerry below Nader. That's what this 
> posting is about.


Seems to be two parts  to this:
      Have to be able to vote sincerely and not get double-crossed.
      Need to be educated that the method provides for this.

> RV, by the way, would likely encourage giveaway-progressives to give a 
> little less to Kerry than to Nader. But I hope that a good rank method would 
> let them _fully_ vote Nader over Kerry.
> 
> Compare how giveaway and particular progressives would vote in Approval and 
> MDDA:
> 
> The candidates are: Nader (N), Kerry (K), and Bush (B):
> 
> Approval:
> 
> Givaway: NK
> Particular: N
> 
> MDDA:
> 
> Giveaway: N>K
> Particular: N
> 
> Obviously N>K is a lot better than N & K.
> 
> No method can completely guarantee to the giveaway voters that N>K couldn't 
> possibly let Kerry when, while N=K wouldn't. But would it take for them to 
> regret N>K?


Why are they supposed to regret N>K if that is their sincere desire?

> 
> If N>K completes a majority against Kerry, then it must be that Nader has 
> majority, which means that Nader also has a majority against Bush. So the 
> giveaway progressives have nothing to lose by voting N>K.  Unless there's 
> offensive order-reversal by the Bush voters. But offensive order-reversal is 
> very well deterred by the fact that if they Bush voters ranked Nader, they'd 
> be giving an Approval vote to Nader. The giveaway progressives have nothing 
> to fear by voting N>K.
> 
> N>K is an improvement over N & K. MDDA, then, does improve on Approval.
> 
> So  MDDA meets that more important requirement, FBC, and also does ok with 
> the 2nd desideratum, that voters not be strategically forced to vote 
> Kerry=Nader.
> 
> Of course, whether the method is Approval, RV, Condorcet or MDDA, the fact 
> that those progressives who are particular refuse to help the Democrook 
> means that the Democrook might not be able to get a majority against the 
> Republican, and might thereby lose to him. That's ok, because he doesn't 
> deserve being helped to win.
> 
> MDDA and Condorcet both share the fact that offensive order-reversal could 
> defeat a CW. But both deter offensive order-reversalvery well. In MDDA that 
> deterrence is automatic.
> 
> With Approval, RV, Condorcet and MDDA, it's necessary to vote down to the CW 
> if one wants to help him/her win.
> 
> In MDDA and Condorcet a Kerry>Bush majority is enforced. That's what makes 
> those rank methods work.
> 
> In MDDA majority is enforced by SFC. All that's needed for that enforcement 
> is that there isn't a majority-strength circular tie, or large-scale 
> offensvie order-reversal. Both of those two problems are unlikely, and so 
> MDDA's SFC majority-enforcement is very good.
> 
> Condorcet has added enforcements. In addition to SFC, it has GSFC, MMC, 
> Condorcet's Criterion and the Smith Criterion.
> 
> GSFC makes a difference if there is a majority-strength circular tie. MMC, 
> the Mutual Majority Criterion, helps when the premise conditions of SFC and 
> GSFC don't apply, but there is a mutual majority. Condorcet's Criterion and 
> the Smith Criterion are helpful if Compromise>Worst isn't a majority defeat, 
> but everyone, including your opponents, votes sincerely.
> 
> It might seem that Condorcet has better majority-enforcement, because it has 
> GSFC, MMC, CC, & SC. But majority circular ties and offensive order-reversal 
> are unlikely, meaning that nothing more than SFC is needed. A mutual 
> majority is a fortuitous special case, unlike the general and plausible 
> premise conditions of SFC. And the assumption that your opponents will vote 
> sincerely isn't too realistic either.
> 
> So SFC is much better than those other criteria, and MDDA's 
> majority-enforcement is effectively just as good as that of Condorcet.
> 
> Strategy in MDDA? Unless a candidate is too disgusting to vote for, such as 
> Kerry, Dean, or other Democrats, rank at least down to the expected CW. If 
> the CW is in doubt, then overshoot, just to be on the safe side. If you 
> aren't sure, you could even rank all the way down to your 2nd to last 
> choice.
> 
> Same in Condorcet, of course, except that you want to try to not rank past 
> the CW if there's a danger of offensive order-reversal.
> 
> In MDDA, the approval-count has two purposes (or maybe more): 1) It's a 
> simple solution for when there are more than one undisqualified candidate; 
> and 2) It deters offensive order-reversal.


For a couple near ties the one ahead can win.

For a true tie, lot us good enough.

When can you be sure strategy will put you ahead - when opponents have 
equal rights?

> 
> So, to answer the question in the subject line: Yes, a good rank method can 
> improve on Approval and RV.
> 
> I emphasize that I'm speaking from the position that we look at bare 
> preferences, which is what most of us are used to. Looking at it that way, 
> these good rank methods are majoritarian and Approval isn't. But of course 
> bare preferences aren't the only way to look at it. RV's ranges, and the 
> emphatic preferences that Approval counts are valid too.
> 
> My preference for MDDA over RV and Approval in public elections results from 
> the standard of bare preferences as oppposed to emphatic preferences and 
> ratings. That's an individual subjective matter, and I'm not saying that 
> Approval and RV aren't as good in an objective sense.
> 
> Now, what if all the progressives made better judgements about who is good 
> enough to vote over their favorite? Then FBC wouldn't be so necessary. Then, 
> for public political elections, SSD's slightly augmented 
> majority-enforcement, with GSFC, MMC, Smith Criterion and Condorcet 
> Criterion, would make it the best method. (By the standard of bare, or 
> ordinary, preferences).
> 
> I consider SSD to be the ideal best for public political elections, when 
> there isn't a serious lesser-of-2-evils problem, because progrssives 
> understand who is and isn't good enough to vote over their favorite, or 
> because the greater-evils have been voted out of the political system.
> 
> But in the meantime I consider MDDA the best, disregarding winnability.
> 
> Considering winnability, RV and Approval are the best, with RV seeming more 
> winnable than Approval, and somewhat better results-wise than Approval
> 
> Mike Ossipoff

-- 
  davek at clarityconnect.com    people.clarityconnect.com/webpages3/davek
  Dave Ketchum   108 Halstead Ave, Owego, NY  13827-1708   607-687-5026
            Do to no one what you would not want done to you.
                  If you want peace, work for justice.