[EM] To Bill Lewis Clark re: stepping-stone

[James Gilmour]

> > So, when (if ever) does IRV perform worse than Plurality?

What definition are we using for "perform worse"?  Are we considering only single office elections?
Or are we also considering single-district elections to representative assemblies and councils?

If you look at the IRV results from State and Federal elections in Australia, and make the
assumption (not always valid)  that first preferences under IRV would have been single Xs for the
same candidates under Plurality (FPTP), I think you will find examples where the overall results
would have been the same, would have been more proportional, and would have been less proportional.
If overall proportionality is part of your criterion, you should dump both Plurality and IRV (and
Condorcet).

Incidentally, Scotland has six significant political parties contesting its national elections.  In
local government, it is seven in some regions.
James