I disagree. I worked on a Rust application with a lot of necessary async and it was orders of magnitude more difficult than the same thing in Node. "Bad implementation choices" - sure, but how hard is it to realise what the good choice is and then develop that into an intuition for the correct implementation choice for next time? It's easy to say "Oh you were doing it all wrong" in hindsight, but it is anything but clear what types and signatures to use when starting out on a problem like the one explored here.