election_title: "Successive elimination Ex.12 — sincere: the A>B>C>D voter gets their last choice"
scenario_description: |-
  The SINCERE half of Felsenthal's successive-elimination truncation example. Source: Dan S. Felsenthal (2010), Appendix A4, Example 12.
  Six voters, four candidates: 1×(A>B>C>D), 1×(C>B>A>D), 2×(C>D>B>A), 2×(D>A>B>C). Under the agenda B vs C, winner vs D, winner vs A: round 1 ties 3:3 and B survives on the earlier-letter convention; D beats B 4:2; D beats A 4:2. D is elected — the dead last choice of the voter who ranked A>B>C>D.
  That voter can do better by revealing LESS. In succ_elim_ex12_truncated.yaml they name only A, dropping out of every round A does not contest. Their absence from round 1 lets C beat B, C then beats D, and the final round ties A against C with A surviving — so the truncating voter gets their FIRST choice instead of their last, by saying less. The truncation paradox.
  What makes the mechanism visible here is that truncation is not a lie: the voter still ranks A top and says nothing false. They simply decline to help resolve contests among candidates they did not rank, and under an agenda procedure declining to participate in a round is a lever.
  Labels are Felsenthal's own, capitalized, so the case can be read beside the paper's table.
  Successive elimination exists in neither the LH engine nor BetterVoting, so this file is tabulated as Ranked Robin, which elects C by a margin tiebreak inside a Copeland tie. Run the procedure with tools_adam/pref_voting_tabulation_engine/successive_elimination_report.py --agenda B,C,D,A.
paradoxes: [truncation, condorcet-cycle]
voting_method: RankedRobin
num_winners: 1
ballots: |-
  1:A>B>C>D
  1:C>B>A>D
  2:C>D>B>A
  2:D>A>B>C
expected_winners:
  - C

# file: succ_elim_ex12_sincere.yaml