This one was also sent to me by the Center for Election Science which dives into the failure in the Alaska Congressional election as well as performing quantitative analysis of likely outcomes in other states given the level of polarization in each:
I don't 100% buy the modeling assumption of a left-right spectrum capturing Americans' political affiliations in the present day - I think there are more axes than that - but the numerical demonstration of the relative representativeness of a Condorcet candidate when one exists is compelling and I believe would extend to a multi-axis model. The IRV outcomes are much more partisan by comparison.
You might want to check out:
https://link.springer.com/article/10.1007/s10602-023-09393-1
You can get the published version free of cost:
https://drive.google.com/file/d/1dFN5Zd2z3U8-cC2eoVGV7Mj1CxVn92VQ/view
I still think my submitted version is better:
https://drive.google.com/file/d/1jIhFQfEoxSdyRz5SqEjZotbVDx4xshwM/view
Hear is a one page primer (talking points) on Precinct Summability
https://drive.google.com/file/d/1YtejO54DSOFRkHBGryS9pbKcBM7u1jTS/view
Partha Dasgupta and Eric Maskin 2004 Scientific American article: The Fairest Vote of All
https://drive.google.com/file/d/1m6qn6Y7PAQldKNeIH2Tal6AizF7XY2U4/view
Here are a couple of articles regarding the Alaska RCV election in August 2022 that suffered a similar majority failure:
https://litarvan.substack.com/p/when-mess-explodes-the-irv-election
https://thehill.com/opinion/campaign/3711206-the-flaw-in-ranked-choice-voting-rewarding-extremists/
This one was also sent to me by the Center for Election Science which dives into the failure in the Alaska Congressional election as well as performing quantitative analysis of likely outcomes in other states given the level of polarization in each:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4411173
I don't 100% buy the modeling assumption of a left-right spectrum capturing Americans' political affiliations in the present day - I think there are more axes than that - but the numerical demonstration of the relative representativeness of a Condorcet candidate when one exists is compelling and I believe would extend to a multi-axis model. The IRV outcomes are much more partisan by comparison.