Clone independence — can running copies of yourself help you win?¶
Ranked Robin's headline defense: adding near-identical candidates ("clones") shouldn't change who wins. These cases probe it — first the LH-only pair that shows teaming on paper, then the BV-backed pair (BV2142/BV2143) where LH and BetterVoting resolve the resulting Condorcet cycle differently.
The cases¶
| Case | Page | YAML |
|---|---|---|
| Before cloning: A, B, C tie in a cycle (LH-only) | page | clone_teaming_01_pre.yaml |
| Teaming: A runs clones, coin flip → win (LH-only) | page | clone_teaming_02_post.yaml |
| BV2142 — clone independence (1/2): a no-Condorcet cycle, where LH and BV part ways | lesson · page | bv2142_4gfwdq_clone_cycle_pre.yaml |
| BV2143 — clone independence (2/2): teaming succeeds on LH, fails on BetterVoting | lesson · page | bv2143_9pr3wr_teaming_fails.yaml |
Frozen BV exports sit next to their yamls; full audit mirrors in clone_independence_tabulated/. Why the divergence is possible at all: LH breaks Copeland ties by margin → lot (deterministic), BV by head-to-head → random — see RR tiebreak, LH vs BV.