I'm a mediocre hobby programmer, and for me, the hardest part is coming up with the concrete specifics of what to implement.

I'm actually half decent now (for a hobbyist), but I remember that when I first moved from coding along with worked exercises in textbooks, to actually trying to solve real life problems, it got exponentially more difficult. The textbooks say "use this tool to achieve this goal", and even if you're bad at using that particular tool, you can put all your effort into doing it well. However when you're having to make the bigger decisions for yourself, you have to choose what tool to use, and I found that super hard — especially when having to choose between an approach that I felt confident with that wasn't quite the right tool for the job, and a new tool that was completely new to me (and if I am unfamiliar with a new approach, am I sure that it's the right tool?).

At one point, when I was deep in a personal project, I had a period of a few weeks where I had no reliable access to internet, and so I did most of my coding using paper and pen. It forced me to think more deeply about the implementation I was using, rather than just diving straight into the code