In the United States probably. The author doesn't sound too old, and says that he/she got access to the internet much later. Which means the author is probably from the Middle east or China or South Asia. The work culture there is not as open as in the west. Daily job means monotonous routine work. What excited him about the prime numbers problem was the novel aspect of it, and the fact that he was in control. Later, in his/her day job, that was no longer the case -- someone else was making the decisions. Until these nuances trickle into daily jobs, most people will sooner than later burnout.
A precursor to the compilers class is the Formal Automata one. This is very useful in understanding fundamental computation models (what is computation, why, how can we do it etc..). I was recently reading  to refresh some concepts and found it to be more engaging than some of the older books on the topic.
Anecdotally yes, within a few percent, depending on what you measure. Safe code tends not to perform as well as unsafe code, but this is also true in C++—most applications just tend to err (and I mean err) on the side of performance.
The difference is that in Rust you can write unsafe implementation of safe interface. In C++ you cannot specify safe interface. (You can and should document memory management assumptions in C++, but C++ compiler does not enforce such documentations.)