Skip to content

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 the clones reshaping the finalist set, then the BV-backed pair (BV2142/BV2143) that pinned the same election on a live BetterVoting count. Both engines elect C after the cloning; the interesting part is that the clones still changed the outcome, just not in their own faction's favour.

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, and the win goes to C (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): the clones change the outcome, and hand it to C lesson · page bv2143_9pr3wr_teaming_fails.yaml

Frozen BV exports sit next to their yamls; full audit mirrors in clone_independence_tabulated/. These pages read as an LH-vs-BetterVoting divergence until 2026-08-19, when the engine's tiebreak ladder was corrected: Ranked Robin resolves a two-way tie on the finalists' own head-to-head (its 1st Degree) before it looks at margins over the field (its 2nd Degree), and the engine had only the second. See degrees of ties and RR tiebreak, LH vs BV.