In the case of deep learning, all current state-of-the-art models can be expressed as nothing more than matrix multiplication, addition, and clipping. That's all you need. It's so simple: multiply, add, clip. Multiply, add, clip.
But the model isn't transparent. It's clear what each step is doing (multiply! add! clip!), but it's not clear to the human why that's necessary.
I eventually wrote my own editor, which is effectively just a text-mode window manager wrapped around a simplified nano clone which uses the control key bindings I'm familiar with from non-terminal environments. I'd be reluctant to recommend it to anyone else, since it's such a specific implementation of my own personal taste, and it has a handful of minor problems I've never bothered to fix that might cause someone else significant irritation; but it suits me. It's a very comfortable feeling to know all that can be known about the tool; I never have to waste a moment trying to remember how any of it works, and can direct my full attention through the tool to the work I'm actually trying to do.
While there are thousands of ways programs can parse arguments (and that's great!) there is only one argument passing mechanism in Unix (plus environment variables for implicit static things and file descriptors for streams).
Unix doesn't care how you parse your arguments, and that's why it is lean, and why it has been successful as a development platform.
Donald Knuth: "I define UNIX as 30 definitions of regular expressions living under one roof."
Nothing beats the simplicity of Sublime and the good'ol Terminal.
Finally, I've come to appreciate using a mouse with my editor. Until I completely got rid of using most keyboard shortcuts (other than the most basic ones), I didn't realize how much cognitive taxes I was paying to retain and context-switch among different families of keyboard shortcuts. Mousing is so much more uniform.
Not sure what they stand for for sure. E is probably editor. B might be buffer?
Personally, I wish sam had even less complexity and features than it currently does and thus be more understandable and hackable. Also, never going back to a normal terminal again, "dumb" terminals are awesome.
And actually, this experience has lead to a philosophy of software where I think the "dumb"er a piece of software the better and more reliable/performant it is likely to be. So I'm also hugely skeptical of the current Deep AI efforts leading to anywhere good.