Compared to a lot of esoteric ruminations on game design, Ernest Adams' little essay on 'the dao of game design' is remarkably grounded and to the point. The question of how to make a game that players — not just the designer — would want to play is an obvious one; perhaps more obvious is the fact that a designer needs to know what they're trying to convey (though, considering the muddled end products we're sometimes presented with, perhaps some designers don't pay enough attention to the 'message'):
In my teaching I have tended to emphasise know thy player more than know thyself, because I feel too many students come to the process with two false preconceptions: First, that game design is a primarily expressive process in which their own desires should dominate; and second, that they are themselves the ideal player for their game.
This is fair enough if they themselves are the only people who will ever play their game, but most designers want other people to play their game as well, and that means thinking about what will entertain them.
Know thy player is doubly important when the game is one that the designer would not choose to play herself — a game for small children, say. When you make games for someone very unlike yourself, you can't rely on your own instincts. You have to study your audience.
Simple and to the point — though, really, many things should be when you get to their core. It's just a matter of paying close attention and recognising the traps you might fall in before you do.
The Designer's Notebook: The Tao of Game Design [Gamasutra]