[EM] MDDA & MDDAsc

[Michael Ossipoff]

I'd like to add this:

I've been saying that you can't choose between the members of your
strong-top set without increasing the probability of electing from your
strong bottom-set.

For example, in MDDA & MDDAsc, you can add an additional level of
protection for your strong top-set if you equal-top-rank them, instead of
ranking them in order of preference.

But, remarkably, if you rank them in order of preference, you're still
giving them just as much protection as you would be giving them by
approving them all in ordinary approval:

 If one of them is the CWs who, alone, has no majority-disqualification,
then burial or truncation against hir will result in an approval-count that
will be won by someone in a majority's approval-set.  ...and, by approving
all of your strong top-set (even not at top), you're fully voting for them
to be in that winning majority approval set.

If a majority's preferences are like yours, then one of your ranked &
approved candidates will win.

Sure, if some who share your approvals rank and approve lower than you do,
they could give the approval-count away to someone in your strong
bottom-set. That's true in Approval too, and, in Condorcet, ranking too low
can likewise allow burial to succeed. So that risk isn't new with MDDA &
MDDAsc.

You can do something to avoid that risk, by equal-top-ranking your strong
top-set. If your majority do so too, that makes it impossible for burial or
symmetrically-completed truncation to majority-disqualify them.

In MDDAsc:

Your top-ranked candidates are fully protected against everyone else.

Your ranked & approved candidates are fully protected against your unranked
unapproved candidates.

Among your middle (not top or bottom) ranked candidates, of course your
higher-ranked ones are favored over your lower-ranked ones by the fact that
your ranking influences the matter of which one doesn't get
majority-disqualifed. But you aren't giving them any protection, against
eachother, from truncation or burial.

So, only partially, are you protecting your middle-ranked candidates
against eachother. But that's ok, because the important thing is the
protection of your ranked & approved candidates against your unranked &
unapproved candidates.

In MDDA without symmetrical completion:

There's full truncation-proofness.

So your middle-ranked candidates are truncation-protected from eachother.

Of course that truncation-proofness applies between _all_ pairs of
candidates too. But, in MDDAsc, your're already fully protecting your
top-ranked candidates against all others, and youre already fully
protecting your ranked & approved candidates against your unreanked &
unapproved ones. So the full truncation-proofness makes the most difference
for protecting your higher-ranked middle-ranked candidates from your
lower-ranked middle-ranked candidates.

...something that isn't as important as protecting top-ranked, or
protecting ranked against unranked.

So MDDA & MDDAsc both qualify as the methods that best deliver on the
promise of rank-balloting, bring its best properties.

Though, speaking for myself, I prefer full truncation-proofness to
Mono-Add-Plump, that will ultimately be a matter for the method-adopting
public to decide. MDDAsc makes it possible to ensure that MDDA's advantages
are available in a surely publicly acceptable proposal.

Michael Ossipoff

On Fri, Dec 9, 2016 at 11:46 AM, Michael Ossipoff <email9648742 at gmail.com>
wrote:

>
> I said this, buried in a long discussion-post, but I'd like to also say it
> in a post labeled with this topic:
>
> It seems to me that MDDA & MDDAsc are the methods that best deliver on the
> advantages & properties that can be available with rank-balloting.
>
> Those properties include FBC, avoiding chicken-dilemma, and solid
> protection of a majority's ranked candidates against their unranked ones.
>
> Their protection of candidates ranked & approved by a majority,  against
> other candidates is positive and solid, in contrast to the way burial &
> truncation can sometimes succeed in MMPOsc or the way burial could succeed
> in IC,MMPO.
>
> There's a tradeoff with those MMPO methods, because, with them, burial and
> truncation are deterred as well as foiled, because, though those offensive
> strategies can sometimes succeed, they're penalized when they fail.
>
> But that deterrence, too, is a bit iffy, because, if the CWs is in
> someone's strong bottom-set, then that person has little or no reason to
> not risk burial.  ...and the perpetual burial fiasco ensues, with always a
> chance that the burial could succeed..
>
> That's why I say that MDDA's & MDDAsc's protection is more solid and
> positive.
>
> Before, I'd felt that compulsory approval of all ranked candidates in MDDA
> was necessary in order to thwart burial. Before Forest Simmons pointed out
> that that isn't so, that hadn't occurred to me, and I hadn't heard anyone
> else say that, or the anti-chicken-dilemma value of _optional_ approval in
> MDDA.
>
> But now that it's mentioned, of course a majority's approval-set are
> protected just as well in MDDA with optional approval, as they would be in
> ordinary Approval.
>
> Whether or not he preferrers of your unapproved candidates approve your
> approved ones, the approvals that they get from a majority will make them
> beat that majority's unapproved candidates.
>
> It would be nice, but not necessary, to have the added burial-resistance
> that compulsory approval would briing, but that compulsory approval would
> lose the chicken-dilemma protection that MDDA can have.
>
> Of course MDDA, MMPO, & IC,MMPO are all from Kevin Venzke, but the fact
> that MDDA doesn't need compulsory approval of ranked candidates, and the
> fact that, without that compulsory approval, MDDA provides a way to avoid
> chicken-dilemmma--Those are things that I first heard from Forest Simmons.
> ...along with the use of symmetrical completion to meet Mono-Add-Plump.
>
> So, not just MDDAsc, but also MDDA with the option to not approve all
> ranked candidates, for chicken-dilemma protection, is probably something
> that was a new suggestion from Forest.
>
> Michael Ossipoff
>
>
>
>
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20161209/eae9b88a/attachment-0001.htm>