[EM] Borda-Elimination

[Kevin Venzke]

Forest, and anyone interested,

I made an interesting but still possibly discouraging discovery.  Check out
these results for Borda-Elimination (standard 5-faction, 9-candidate setup):

matches        16246
sdindec        507           3.12
be<>cw         0             0       <-- !!!
be=cw          10142         62.42
unanimous      1537          9.46
be=borda       496           3.05
be=sd          1721          10.59
borda=sd       1266          7.79
no matches     577           3.55

Not even once!  I'm in disbelief.

Of course in these simulations I can't check vulnerability to strategy and clones
(probably significant).  But even with such limitations, it is amazing to find a 
non-Condorcet method that so consistently picks the CW.

It would be interesting to study those cases where "BE" does not agree with RP.
I wonder if BE's result in such cases is just erratic and senseless, or if there's
a reasoning behind it.


Kevin Venzke
stepjak at yahoo.fr


___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en français !
Yahoo! Mail : http://fr.mail.yahoo.com