Or, AFAIK, you can build Turing-complete stuff. However, I prefer to avoid adding games so complex/advanced.

Science has to be its main part of mechanics, rather than some subpart, as:

"Any sufficiently advanced game is indistinguishable from a science-based game."

(Saw somewhere on HN "Any sufficiently advanced text editor is indistinguishable from a dedicated app." Yes, I know it's from https://en.wikipedia.org/wiki/Clarke%27s_three_laws.)

I don't understand neither your objection to complexity, nor how to determine "the main part of mechanics" in a sufficiently complex game.

What's the main part of the mechanics in DF? I play it for 7+ years and I have no idea. Most likely, there isn't any.

Still I learned much - directly and by piqued interest - about geology - way before attempting to deliver magma to the surface. Years, in fact.

I feel that such focus on "main part of mechanics" is wrong.

It's like you'd accept a pool simulator because it's "main part" is the combination of perfectly elastic collision and conservation of momentum and reject KSP because "it's too complex".

Of course it's complex. How do you expect to learn anything if it's not?

> Saw somewhere on HN "Any sufficiently advanced text editor is indistinguishable from a dedicated app."

You mean, "any sufficiently advanced text editor is indistinguishable from an operating system"? :).

And the joke isn't that sentence - the real joke is that such a text editor can offer better UX than your regular operating system...