This is especially hard if you enjoy building voices, which is not something I care about much. I like mostly selfcontained modules that give me a lot of controls for timbre so I can change a voice in a patch somewhat dynamically. Still working towards my ideal on this front, but you can definitely build around versitility in eurorack if that is important to you.
I’m personally in the generative camp, so for me composing is practice, patching, and looking for ways to refine the results to better match my musical goals. When I feel intentional, I have controllers for that, but that control always feeds into my performance patch as an alternate way to interact with my modular. So I can either augment what’s being generated, play on top of, or in place of what the system is already doing. I cannot, however, easily write a song outside of maybe using Ansible Earthsea? But I didn’t build my system to do that.
That is to say, what always reaffirms itself in modular for me is that having clear goals that you reevaluate as you build your system is maybe the most important part of the process. It’s not an instrument you pickup and practice.
You have to select the tools, deteminte how they interact, and decide how you want to compose with it, and that’s just a kind of base-level “So you want to modular?” level of inquiry, which says nothing about constraints implicit to those tools or your goals. I can absoltely see why people get out of modular after falling in deep. I love it, but it is decidely not for everyone.
