Concept art is good for grabbing the feel of something. I kind of like Dwarf Fortress's approach to features - they write a story, describe the kind of feel they're trying to grasp with the mechanics.
What I do myself before starting any project these days is to write up an elevator pitch to brainstorm out what the game is all about and an example of play. It's to make sure that throughout the design process, I haven't lost the feel of what inspired the game.
Also things like visuals and audio work much better when you iterate. If you have no idea where you're going, you'll get your music out of tune with the tone of your games, your art palettes won't grasp the feel, and so on. If you plan too much and don't prototype, you might not realize that your well researched artwork doesn't even go well with the rest of the game. Iteration is the middle ground.
What do you mean by more complex systems? Dwarf Fortress is a great example of how to approach a mega epic project. It's kind of hard to follow at this stage of development, but you put up placeholders everywhere. They set up a simplistic attribute system before replacing it with a complex one. They put up simple body mechanics at first and have now replaced it with such fine details that someone's ribs can be broken in such a way that it penetrates into their heart.
This is a pretty damn epic list. But it's well organized and they know what to develop and what to placeholder for later.
For a more theoretical option, there are plenty of game design books. I'm out of date, so I can't really recommend, but I do like the Mission/Space approach to level design. Levels are split into mission (story/approach) and space (physics/moving around) and the different approaches are mapped.
But as you go more complex, you'd require more education to be able to pull it off convincingly.