Talk:IRV Prime: Difference between revisions
no edit summary
No edit summary |
|||
Line 150:
::: Let's say that the voter's ballot is A>B before and A>B>C after. If A or B won before and now C wins, that's LNHarm failure. If C won before but A or B wins after, that's LNHelp failure. You *could* try to devise a crafty mathematical proof, but with simple criteria like these, brute force is easier. [[User:Kristomun|Kristomun]] ([[User talk:Kristomun|talk]]) 21:05, 14 October 2021 (UTC)
:::: Thanks! I think it'd be really nice to get a mathematical proof just because I don't have a lot of faith in simulations. I thought about this a bit more & I think the key is my original proof - I think it DID prove that the method satisfies later-no-harm & condorcet, but what I was missing was later-no-help. I updated the proofs, can you take a look to see if they make sense? --[[User:Marcosb|Marcosb]] ([[User talk:Marcosb|talk]]) 05:43, 18 October 2021 (UTC)
::::: When it comes to disproofs, simulation can be just as effective as reasoning through the method. This because if you want to prove that some property X is not satisfied, all that's required is a single example. See e.g. https://cdn.paperpile.com/blog/img/lander-1966-700x394.png. If you want to produce a more general proof (like Woodall did) or find out some deeper structure (e.g. why it's impossible), then you'll probably need to reason, but not to merely establish that X is not satisfied. [[User:Kristomun|Kristomun]] ([[User talk:Kristomun|talk]]) 16:29, 19 October 2021 (UTC)
== Arrow/IIA ==
| |||