Candy Crush was immensely popular. It's not exactly chess.

1. As the article says, "People will be willing to go along with pretty simple and pretty familiar problems as long as the feedback is great."

2. For arbitrary n x m boards, Candy Crush (and Bejeweled, it's predecessor), has been shown to be NP-complete. That means as a general class of problem, it's officially hard. This is why I said that "A lot of very good problems seem stupidly simple, but have depths to them." If you look at some of the ones I examine in this talk: https://www.raphkoster.com/games/presentations/games-are-mat... you'll see they are often what seems very simple!

Sorry, but that's a bit of hyperbole. Candy Crush may be NP complete (in some sense), but that doesn't make it hard to play.

"Hard to play" has almost nothing to do with "the underlying game system is Hard in a complexity sense." And both are somewhat orthogonal to the complaint about "adding more stuff" -- doing so may make something more complicated, but that doesn't mean it becomes more complex mathematically.

Adding more stuff in general may very well make something harder to play, but it has more to do with how many of the added things are simultaneously visible. We can go back to the old 7+/-2 rules of thumb on that one.

Candy Crush is Hard in a math sense. So are Battleship, Minesweeper, Pipes, Othello... they are all easy to play. They are fun because they are Hard in a math sense.

Dark Souls is not hard in a math sense, meaning it is not formally complex. It is hard to play mostly because it is a reaction and timing game that presents very difficult challenges, and the ambiguity comes from mastering signals and reactions.

Civilization is complicated, and almost certainly Hard in a math sense. But it has solid onboarding and is not that hard to play.

Dwarf Fortress is complex AND complicated, and has no onboarding, and a ton of stuff in it, and is both Hard and hard. DF is built out of a ton of separate Hard game atoms.