Dark Horse

A “Dark Horse” voting strategy in ranked-choice voting systems which attempt to find “Consensus” candidates involves voters insincerely pretending that a candidate universally agreed to be the worst is instead a “consensus” candidate, in hopes that the resulting “Dark Horse” consensus candidate will beat their less-preferred frontrunner so that their most-preferred frontrunner has a greater chance of victory.

The problem is that this is risky: If too many people employ this strategy, the “fake” consensus candidate becomes the “real” consensus candidate – and the universally-agreed worst candidate becomes the victor.

Instant Runoff Voting avoids this pathology by ignoring all later preferences for candidates not yet eliminated; by the time a “Dark Horse” candidate would be able to eliminate your second-favorite-frontunner, your first-favorite-frontrunner would have been eliminated.

[[Open Questions]] Exactly how should we define the “Dark Horse” scenario to make it less of a fatal flaw?

• Should it only count true “Dark Horse”s – artificially inflating the ranking candidates not a favorite, as opposed to artificially inflating another frontrunner?
• It definitely should not count if there exists a non-“Dark Horse” strategy (on the side of either the voters motivated to push a Dark Horse or the side defending against the Dark Horse) which makes Dark Horse suboptimal.