I also think two of the principles you laid out - ease of implementation and transparency/explicitness of code are the hottest takes in Hare. I can see why people like or dislike those choices, and it's compelling to think about, but I'm not sure that the explicit choices to add or not add features from PL research reflect those.
For example the paragraph on GC - I think there's a ton of reasons not to use a garbage collector in a language like Hare (for example, you have to write a GC in some language and not the one that is GC'd) but I'm not sure the reasoning you laid out is among them (concurrent and hard-realtime GCs are thing, whether removing the burden of manual memory management is more/less transparent to the intent of a programmer is very debatable).
> for example, you have to write a GC in some language and not the one that is GC'd
This is not true.
Honestly, I kind of expect them to walk back on the borrow checker promise for that reason alone.