A concrete counterexample: plantura.garden is a large, reputable German-language gardening magazine / brand, and probably exactly the kind of legitimate site one would expect on .garden.
So while the abuse numbers may well justify treating newly registered / low-reputation .garden domains with suspicion, blanket-blocking the entire TLD seems like it would create real collateral damage.
Nice. Check out https://github.com/meodai/color-names for a rather large (29888!) collection of hand picked colors in case you want to expand your project.
No. Their only purpose is to act as a kind of receipt: "look, I actually really did waste all this energy".
The solution to the puzzle itself is irrelevant - it's an operation that would be extremely simple but that's made very, very hard on purpose. And the sole purpose is to align incentives by using lots of energy.
Buy Me a Coffee tells exactly the kind of story that DLT (blockchain) startups often tell - connect creators with their fan base etc., just the "let's cut out the middlemen" is obviously missing.
I totally get why you'd go for Stripe to build a mature, scalable product open to as many users as possible. But just out of curiosity: did you spend some time thinking about DLT to cut out middlemen, did you actively decide against it, are you experimenting with it - what's your stance?
So while the abuse numbers may well justify treating newly registered / low-reputation .garden domains with suspicion, blanket-blocking the entire TLD seems like it would create real collateral damage.