[EM] Tactical voting under a jury model

Colin Champion colin.champion at routemaster.app
Mon Mar 7 06:45:01 PST 2022


Kristofer - for completeness here are my other results. Compromising, as 
you imply, is much the same as sincere voting; false cycles are much 
like burial. My final condition is "exhaustive" tactical voting, in 
which all voters supporting a given candidate submit the same insincere 
ballot. A subversion is considered successful if it leads to the 
insincere voters getting their way and their candidate is worse than the 
sincere winner; and the loss is the resulting drop in valence. I'm not 
sure how close this is to Daniel's tests, but it was based on what he 
was saying at the time. Exhaustive tactical voting is again much like 
burial.
    Other parameters are the same as in my previous run.
    Like you, I was a little perturbed by Condorcet/Hare not being at 
least as good as IRV; I simply assumed (or hoped) that the conditions of 
JGA's theorem didn't hold.
    Colin

*Compromising*
            random   fptp    sptp     av   sinkhorn borda     mj coombs
           115.6767  4.2340 17.5522  3.4865  2.8701  2.8441    - 3.4740
          condorcet benham   btr    nanson minimax minisum    rp river  
schulze   asm
              -     3.4419  3.3958  3.3815  3.3559  3.3530  3.3659 
3.3858  3.5267  3.5209
condorcet+ random   fptp    sptp     av    borda
            5.9255  3.3738  3.7981  3.4409  3.3058
  copeland+ random  fptpf   fptpr    sptp    avf     avr    bordaf 
bordar minimaxfminimaxr
            3.4407  3.3690  3.3868  3.4452  3.4296  3.4302  3.3063 
3.3388  3.3530  3.3647
     smith+ random  fptpf   fptpr    sptp    avf     avr    bordaf 
bordar minimaxfminimaxrtideman   q&dc
            3.5267  3.3720  3.3890  3.6909  3.4409  3.4382  3.3058 
3.3259  3.3559  3.3559  3.4344  3.5345

*False cycles*
            random   fptp    sptp     av   sinkhorn borda     mj coombs
           115.6767  4.2340 17.5522  3.4932 10.2714 10.0818    - 26.3221
          condorcet benham   btr    nanson minimax minisum    rp river  
schulze   asm
              -     3.5519  5.4141  8.5122  8.6953  8.5354  8.7293 
8.7462 12.4334  9.2892
condorcet+ random   fptp    sptp     av    borda
           11.1509  5.2152  6.1789  3.5529  7.3317
  copeland+ random  fptpf   fptpr    sptp    avf     avr    bordaf 
bordar minimaxfminimaxr
           10.6530  5.1834  5.3717  6.0292  3.7368  3.7725  7.1250 
10.5974  8.3103  8.3840
     smith+ random  fptpf   fptpr    sptp    avf     avr    bordaf 
bordar minimaxfminimaxrtideman   q&dc
           12.4334  5.2357  5.4327  6.1769  3.5529  3.5543  7.3305 
10.8659  8.7099  8.7099  4.1473 12.2860

*Exhaustive*
            random   fptp    sptp     av   sinkhorn borda     mj coombs
           115.6767  4.2340 17.5522  3.4934 29.7429 28.3067    - 47.0813
          condorcet benham   btr    nanson minimax minisum    rp river  
schulze   asm
              -     3.5522  5.4275 11.0898 10.0524 10.7085 11.5937 
11.6862 15.6366  9.9130
condorcet+ random   fptp    sptp     av    borda
           18.3528  5.2430 13.4912  3.5582 20.3795
  copeland+ random  fptpf   fptpr    sptp    avf     avr    bordaf 
bordar minimaxfminimaxr
           15.7254 10.6274 10.7833 15.1104 10.3921 10.5389 19.4887 
13.4124 12.2174 12.2592
     smith+ random  fptpf   fptpr    sptp    avf     avr    bordaf 
bordar minimaxfminimaxrtideman   q&dc
           15.6366  5.3535  6.3920 13.5263  3.5589  3.5666 20.3782 
17.8307 10.0525 10.0525  4.1848 19.7032



More information about the Election-Methods mailing list