It’s generally a lot more control-able with a terminal. I can remote into any of my machines, including IOT stuff, and have full control over any of their settings, like, say, volume or display brightness or whatever. With GUI it’s like what, RDP/VNC/Teamviewer in? Gonna be painful over mobile connection… Apps? The developer just went bankrupt and now it’s dead because the cloud server is down. And I haven’t even started on automation…
Also, changing multiple behaviours in one place is also nice. Say, I want to remove the volume osd and control how I want to manipulate windows. It’s either an array of small disconnected utilities(in this case, HideVolumeOSD + AltDrag), or huge RAM hogger utilities like DesktopFusion, or, I can just edit it in my WM config with just a couple of lines. Things like adjusting volume based on window position (to have a background and foreground media displays) is completely out of the picture on Windows and are a breeze on Linux.
In short, lots of benefits. The downside, I guess, is a complete disarray of components. Like in case of volume again, I have pipewire daemon pretending to be pulseaudio which is middleware for alsa… and all of their corresponding utilities work but control the same thing, so it’s incredibly confusing which ones to use. Also webapps for some reason can control their own volume in the system mixer(?) So there’s at least 4 ways to adjust just one slider. And it can create confusion when multiple things interact with different interfaces - I’m still to figure out which fucking thing keeps setting the grp:alt_shift_toggle option in my keyboard layout.