[EM] The war on burial, um, pushover

[Kristofer Munsterhjelm]

Here are some stats I just calculated.

It seems like Kevin Venzke's "No-elimination IRV" is ever so slightly 
more strategy resistant than ordinary IRV. It might not be worth it, 
complexity wise, but it's still interesting, as it makes it (as I know 
it) the record holder for strategy resistance under Condorcet, 
resolvability, and UD.

I'm using his type one method. 50k elections, 32k tests per election, 
impartial culture:

Here's Schwartz,IRV:

Ties: 0

Burial, no compromise:  3473    0.06946
Compromise, no burial:  12450   0.249
Burial and compromise:  41      0.00082
Two-sided:              3       6e-05
Other coalition strats: 7720    0.1544
==========================================
Manipulable elections:  23687   0.47374

and Schwartz//IRV:

Ties: 0

Burial, no compromise:  3450    0.069
Compromise, no burial:  10949   0.21898
Burial and compromise:  1432    0.02864
Two-sided:              8       0.00016
Other coalition strats: 7754    0.15508
==========================================
Manipulable elections:  23593   0.47186

And here's Schwartz,No-elimination IRV:

Ties: 1 (2e-05)

Burial, no compromise:  3463    0.0692614
Compromise, no burial:  12568   0.251365
Burial and compromise:  12      0.000240005
Two-sided:              6       0.000120002
Other coalition strats: 6557    0.131143
==========================================
Manipulable elections:  22606   0.452129

and Schwartz//No-elimination IRV:

Ties: 0

Burial, no compromise:  3433    0.06866
Compromise, no burial:  11175   0.2235
Burial and compromise:  1285    0.0257
Two-sided:              3       6e-05
Other coalition strats: 6452    0.12904
==========================================
Manipulable elections:  22348   0.44696

The higher strategy resistance mainly comes out of a lower "Other" 
incentive; hence the modification is most likely reducing pushover.

-km