That’s what I meant about being pragmatic: Don’t try to enter a genre which prioritizes stuff you can’t do, and turn your limitations into a style. Yes, it’s true that the graphical extravaganza game is unreachably difficult to produce for anyone who isn’t a big game studio, but those aren’t the only kind of games people play, and it’s possible for one or two people to make all the art involved in a 2D platformer.
This is a good point, and you can see it even if you restrict your examples to FPSes. It’s all about story: MIDI Maze, a game I knew as Faceball 2000, had no story. You, a 3D smiley face, explored a maze to shoot other 3D smiley faces. When you finished the maze, you got another maze until the game was over. The original Doom had barely more plot than that, but it did have a story told in text screens. Quake (with the exception of the deathmatch-only games) had barely more story than Doom, but the trend was clear: FPSes were going to have stories now, and you can’t tell someone a story if they keep dying in the first level. As per Graham Nelson, interactive fiction is a narrative at war with a crossword puzzle; well, FPSes were trending IF-ward, and the narrative was beginning to win. That pushed the difficulty down a bit.
This is all true as well, and player expectations probably influenced that too: If your first exposure to a game (or style of game) was in an arcade, you’d expect other examples of that game (or genre) to be arcade-like and, therefore, quite hard. Throughout the 1990s, arcades began to die, and the arcade tropes were never learned by younger players who never imprinted on them.