Why truncation resistance is important (RE: [EM] Re: Rob: MDDA vs BeatpathWinner)

Sun Oct 9 00:07:55 PDT 2005

On Sat, 2005-10-08 at 21:30 -0500, Paul Kislanko wrote:
> All of the gobblydegoog aside, to Rob - we don't care what you want, us
> voters want to NOT have to rank all altertantives. I want to list only the
> ones I find acceptable in the order I prefer them. Any method that
> "encourages" me to rank all alterntatives whether I know anything about them
> or care about them will just encourage me to not vote.  

Paul,

Here's the problem with that philosophy.  Let's say there are three
credible front-runners: "mediocre", "bad" and "worse".  I would hope
many voters would like to be able to express both that "bad" and "worse"
are unacceptable, while also expressing their preference of "bad" to
"worse".

The sad fact of the matter is that when there's genuine disagreement
about candidates, we have to compromise.  The only way to accurately
find the compromise is if voters are actually encouraged to state what
they'd settle for to avoid their worst outcome.

As to whether "we don't care what you want", who is the "we" you think
you are speaking for?

Rob

> > -----Original Message-----
> > From: election-methods-electorama.com-bounces at electorama.com 
> > [mailto:election-methods-electorama.com-bounces at electorama.com
> ] On Behalf Of MIKE OSSIPOFF
> > Sent: Saturday, October 08, 2005 8:50 PM
> > To: election-methods at electorama.com
> > Subject: [EM] Re: Rob: MDDA vs BeatpathWinner
> > 
> > 
> > Rob--
> > 
> > You wrote:
> > 
> > The one major objection I have [to MDDA] is the lack of
> > truncation resistance - I really hope we can find a system that
> > encourages a full ranking.  It seems that MDDA would tend to 
> > discourage
> > ranking anyone below a plausible frontrunner, if I understand it
> > correctly.
> > 
> > I reply:
> > 
> > Not necessarily. For the person who needs and wants to make 
> > use of FBC 
> > compliance, then yes, that person will equal-top-rank the 
> > acceptables, and 
> > power truncate the unacceptables. If power truncation isn't 
> > available, then 
> > s/he will truncate them or rank them in reverse order of winnability.
> > 
> > Likewise, s/he'd sometimes rank the unacceptables in reverse order of 
> > winnability in BeatpathWinner, if offensive-order-reversal 
> > deterrence isn't 
> > the primary consideration. S/he must judge if offensive 
> > order-reversal is a 
> > threat, and, based on that, whether to truncate or 
> > strategically rank. With 
> > MDDA there similarly would be a choice between truncation and 
> > strategic 
> > ranking. Truncate if you think you can best beat all the 
> > unacceptables in 
> > the Approval count. Strategically rank if you think you can 
> > best beat them 
> > all by making majority defeats against eachother. With power 
> > truncation you 
> > don't have that concern: Just power truncate them. Much simpler than 
> > BeatpathWinner or unenhanced MDDA).
> > 
> > But I'm saying that SFC compliance is for the person who 
> > doesn't need FBC, 
> > for the sincere voter. If you believe you're in a 
> > sincere-voting majority, 
> > then you can freely rank all the candidates, and you're 
> > assured that no one 
> > that your majority like less than the CW will win. That 
> > doesn't require any 
> > truncation, or any kind of strategy.
> > 
> > So it isn't that MDDA forces you to truncate. It depends on which 
> > criterion-compliance you want to use, which kind of strategy 
> > you want to 
> > use, what kind of a voter you are. The two kinds of strategy, 
> > the mix of 
> > criterion-compliances, makes MDDA more versatile than BeatpathWinner.
> > 
> > You continued:
> > 
> > I'd also really like to rename SFC to something that is more 
> > descriptive
> > of what's measured.  If a system encourages truncation, it can't
> > truthfully be called "strategy free".
> > 
> > I reply:
> > 
> > I haven't named MDDA "Strategy-Free". When Kevin devised and 
> > introduced 
> > MDDA, we agreed on the name MDDA.
> > 
> > But MDDA is indeed strategy-free for a majority with the 
> > conditions and goal 
> > described in SFC. SFC describes condidtions under which a 
> > majority can 
> > ensure that no one worse than the CW wins, without using any 
> > strategy, with 
> > complying methods. So yes, SFC is about strategy-freeness, where it's 
> > possible.
> > 
> > 
> > Such subjective titles don't
> > serve us well anyway.  "Majority Pairwise Winner" would be an accurate
> > and suitable name, but certainly not the only one.
> > 
> > I reply:
> > 
> > The words strategy-free describe SFC's guarantee. 
> > Strategy-freeness where 
> > it's possible.
> > 
> > You continued:
> > 
> > I've contemplated
> > "Majority Condorcet Winner", but using "Condorcet" in the name isn't
> > helpful, since I don't think Condorcet ever proposed anything 
> > like that.
> > 
> > I reply:
> > 
> > That Condorcet didn't propose SFC, or a name for it, doesn't 
> > mean that it 
> > wouldn't be appropriate to name SFC for its protection of CWs, or its 
> > relation to the Condorcet Criterion.
> > 
> > I sometimes refer to SFC as Condorcet's Criterion for 
> > majorities. And I 
> > sometimes refer to GSFC as the Smith Criterion for majorities.
> > 
> > I have no objection to those names in the paragraph before 
> > this one.  If 
> > others prefer them, I'd be glad to use them. Then, I'd capitalize 
> > "Majorities", so the criteria would be "Condorcet's Criterion for 
> > Majorities" and "The Smith Criterion for Majorities".
> > 
> > But I feel that "Strategy-Free Criterion" describes what 
> > makes them the 
> > pinnacle of the promise of rank-balloting.
> > 
> > Mike Ossipoff
> > 
> > _________________________________________________________________
> > Express yourself instantly with MSN Messenger! Download today 
> > - it's FREE! 
> > http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/
> > 
> > ----
> > Election-methods mailing list - see http://electorama.com/em 
> > for list info
> > 
> 
> 
> ----
> Election-methods mailing list - see http://electorama.com/em for list info