For what it's worth, if the image they used is representative of the other variable names they found, then that line of reasoning doesn't make sense - the names are just gibberish characters in Chinese, and some even have random radicals and other characters thrown in. This seems more like the type of strings you get when you take a bunch of random valid UCS-2 code points.
"Easily" is a bit of an exaggeration - the text is so tiny you wouldn't be able to read it without putting it up to your face and you'd run into the other security features first.
To be able to submit to arXiv, one needs to be endorsed by someone in the domain, which could be a potential hurdle for someone not in academia. (and I'm not about to suggest viXra as a way to get around that)
Yeah, I actually used to be a math post-graduate and have a draft of a paper with an interesting result, so I wanted to publish it but I had no idea how to get "endorsed". This was like 10 years ago and I switched to a career in programming so I doubt I'll even understand my own proof now.
Mark Rutland pointed out in the comments that the `PAC` and `AUT` instructions GCC picks to do the authentication are treated as NOPs on older processors.
