[EM] criteria compliance and strategic vulnerability

[Chris Benham]

James,
I've looked over  your  "Voting Methods:definitions and criteria" page, 
and to my mind there are
quite a few things that are conspicuous by their absence. The way you 
have categorized the different
criteria may not be ideal, because some of them relate to more than one 
of  "majority rule", "defeat strength"
and "strategy".
There is no version of  Minimal Defense, or  the Plurality criterion. 
My  "No Zero-Information Strategy" and
"Weak Burial Resistance" criteria are absent. (The latter is very 
closely related to your Dominant Mutual Third
criterion.)
There are no "Gross" versions of  Condorcet and Smith.What do you think 
is the advantage of  amalgamating
Later-no-Harm and Later-no-Help into the "Later preference criterion"?
IMO, two "consistency" criteria that are of greater practical importance 
than the ones you list are "Mono-add-plump"
and "Mono-append".

Regarding the methods, I think it is a bit incongruous that you list a 
few rankings methods and Approval alongside
one rankings/ratings hybrid  (Cardinal Pairwise, which uses a cumbersome 
double ballot); but no rankings/approval
hybrid methods. I suggest you list AWP, DMC and  AM.

I have a few comments about some of the entries in your "Strategy 
definition summary" table.
I  am not a big fan of  Approval, but I wouldn't say that  
Compromising-compression was strictly "forced". That is
only true of voters who wish to approve candidates other than those they 
sincerely rank equal-first.
Likewise Burial-compression is only "forced" for voters who wish to not 
approve candidates other than those they
sincerely rank equal-last.
You don't distinguish between  ER-IRV(fractional) and  ER-IRV(whole). 
Which one does the table refer to?
In any case they both should definitely have a higher number than does 
plain IRV in the "paradoxical" row, and the
number for ER-IRV(whole) should be 4 or 5.
Maybe WV should have a higher number than Margins in the  
"compromising-compression" row, because in WV
but not Margins  the voter can have a zero-information incentive to 
equal-rank at the top.
Probably WV should have a lower number  than Margins in the 
"compromising-reversal"  row,  because sometimes
in WV  compromising-compression can be an effective "defensive strategy" 
but to achieve the same effect those voters
in Margins  have to compromise-reverse.

There can sometimes be objective measures of how badly a method fails a 
criterion. For example, "CDTT,X" (where "X"
is a Later-no-Harm method), which orders the candidates according to X 
and then elects the highest ordered in the CDTT,
strictly meets Later-no-Harm unless there are more than three candidates 
in a majority-strength top cycle (very very unlikely
to occur in practice).  These methods fail Smith(Net) and the Plurality 
criterion but meet Smith(Gross), and with all the voters
having no incentive to truncate then those criterion failures are very 
unlikely to occur in practice.
In this context, my favourite X is IRV because it is the only 
Later-no-harm method that both has no clone problem and  meets
Weak Burial Resistance.

I found some of the colours on the dark background a bit hard to read 
unless I highlighted everything (by right-clicking on
"Select All").

Chris Benham



Abbreviations... 

compromising-c: compromising-compression

compromising-r: compromising-reversal

burying-c: burying-compression

burying-r: burying-reversal

c forced: compression is forced by the ballot

nER-IRV: IRV without equal ratings allowed

ER-IRV: IRV with equal ratings allowed

WV: winning votes