"The biggest myth in speedrunning history": A cosmic ray didn't help a Mario 64 speedrunner

I tend to think that https://www.youtube.com/watch?v=vj8DzA9y8ls&lc=Ugz4FWk8qAk5qpc_80J4AaABAg and the thread following it get my sentiments here on the nose. LunaticJ's dismissive to the point that he's becoming the machine he's raging against. Infinitesimally small odds things are happening everywhere every day in the universe and it's human beings' inability to comprehend the concept that "infinitesimally small doesn't mean never" that causes him to create the mirror image of his major complaint.

He's upset that everyone buys the Cosmic Ray as truth, but he also states several times that it's absolutely a myth rather than what should be the accepted reality, which is that it's unlikely, but it's just about as likely as any other highly unlikely occurrence all things considered. He provides no proof that any of the other possibilities are any more likely than the one that he claims to be debunking, he's just certain that that one isn't it for... reasons?

In fact, his only real claim is that a bit flip working from imprecise knowledge of the inputs in use at the time of the original incident doesn't match the original perfectly, while offering no other explanation that even comes close. Honestly the recreated bitflip looks insanely similar. Given that an N64 joystick can register something like 7600 different positions and that number isn't even "the law" since the joystick technically uses 8 bits for up/down and left/right. It's only the shape of the stick enclosure that limits it to the ~7600 number. Wear and tear on the physical components can expand or contract that number.

So combine everything above with the fact that you're transcoding the video while streaming to Twitch, so frames won't necessarily line up one to one in the recording vs what was actually happening in a frame, and then you realize that you'd have to perfectly replicate one of ~7600 joystick positions (maybe less, depending on how Mario 64 handles the analog motion curve, I've actually never looked) every frame at ~30 frames per second from a copy of the original that doesn't necessarily represent an exact frame by frame replay thanks to the nature of perceptual encoding. It's extremely unlikely that a replay attempt will ever recreate an exact 1 to 1 and it seems even more unlikely that the recreation we have being so damn close would allow for another explanation to offer a more compelling solution.

Is it right to say that it was a cosmic particle? Probably not. Is it right to say that it was a flipped bit? Probably yes. There's no 100% confirmation for either, but the bit flip is probably as close as any recognized scientific theory is to reality given what we've been able to observe. If you start from that reality and start to enumerate potential causes for a bit to flip, the previously infinitesimally small chance starts to become slightly less so and at that stage a cosmic particle is probably only slightly less likely than any of the other contenders.

TL;DR: A decent video one issue aside. He was far too certain in his dismissal of the bit flip for someone that failed to provide anything else that came within a country mile of reproducing the same effect. But he is getting north of 175k views and he got my view, so it's probably paying just fine for him to be probably wrong about this not being a flipped bit, however it happened.