You’re right most isn’t the right word, but let’s not stand on semantics. If fully half of voters aren’t conforming to expected behavior then you cannot expect the thought experiment about the effects to stand.

You gave me an empty simulation. I’m not doing your work for you, proving a negative is impossible. And a no majority result (Or in the US nobody reaches 270 votes in the electoral college), means Congress votes by state. We have constitutional back ups.

You are using exit polling. We have actual data about how people vote under approval voting. With things not accounted for in any of your math models, simulations, or polls; such as politicians using a one vote message during their campaign. Which is tactical voting. Ironically your next source identifies approval voting as the most susceptible to tactical voting, even above our current model.

The higher dimension election modelling source isn’t bad. But I can see why you just dropped it in there without saying anything. You didn’t read it and you just hoped the wall of text would add authority to what you were saying. Unfortunately for you he actually shows how weak approval voting is once actual human behavior is even attempted to be modeled.

The spoiler rate is not relevant. which is probably why you quickly switch to the stupid voter myth. However RCV is in use in 62 jurisdictions right now and that’s not actually a problem, except with bad actors who are going to go after any system we use. About spoilers, every election, poll, referendum, or ballot is subject to changing the measurement based on what is being measured. Put simply the choices available will always change how people vote. Imagine a world in which Trump was caught red-handed giving nuclear secrets to Putin in Mar-a-lago and he’s locked up for life with everyone agreeing he shouldn’t be on the ballot. How does that change Biden’s chances? Imagine actual Jesus is on the Ballot, how does that change Trump’s chances? These are extreme theoreticals used to prove a point but let’s go back to sophomore year poli-sci classes.

One of the things you learn around that time is how this works. Usually they two examples, and hypothetical bad faith one and a real good faith one. In the first, a city council has an empty lot and they want commercial development on it. But they need voter approval for what to do with it. So the first thing they do is a private poll to see what people want. It comes back 20 percent commercial; 20 percent homeless shelter; and 60 percent kid’s park. So on the referendum they put a choice of commercial development and a homeless shelter. They then sit back and let NIMBYism do it’s thing. In the real life good faith example they turn the page back to 1992 and Ross Perot.

So yeah, complaining about spoilers is just making noise and hoping people don’t actually figure it out. It’s really only a problem in bad faith scenarios like the GOP running people with nearly the exact same last name as the Democrat in the race.

The French study you linked does not prove you have to vote only for your preferred candidate or your safety candidate. That’s also an either/or fallacy, especially when discussing systems expressly meant to handle multiple choices. It is also, yet again, an exit poll. What have we observed about non binding polls? Lets ask the musician you so kindly linked. Oh yeah they’re likely to involve people who are more political, more conversant with the systems being tested, and less likely to vote tactically because there’s no pressure to win in an experiment.

I can see why you think we won’t be having a productive conversation after this. Several of your sources don’t claim what you say they do, in fact they pretty clearly tell you not to use their data in the way you are using it. Source bombing a conversation is a troll tactic.