Mr. Eppley wrote: If it doesn't take much time, how about doing these 3-candidate random simulations: *** 2. Every voter ranks all 3 candidates. Some voters rank two candidates as equal second choices (e.g., A > B=C) and the rest of the voters have no equal rankings. (There are 9 possible orderings.) ----- D- did 40 simulations. Results- 36 Condorcet winners, 3 circular ties, 1 with 2 different winners and tie votes for remaining pairing. Average winning percent- 58.7 (range 50.1 to 74.6) -- 33.0 percent of voters voted a A>B=C type vote (range 17.9 to 50.9 percent of voters), 67.0 percent voted a ABC type vote (range 49.1 to 82.1 percent of voters (reverse of the other range since the vote for 2 option is not in this simulation). Such 33-67 split is highly suspect for a real election.
