The first issue was providing enough fun. Every puzzle should have some reward, and a complicated or multi-stage puzzle should provide some minor rewards for partial solutions. So once I had the puzzle structure in mind (more about that later), I could see which puzzles were going to open a lot of new game-play and which were only going to bring the player up against another puzzle — the structural equivalent of getting through one locked door to find that there’s another beyond it. Everywhere there was a puzzle without much game-play reward, I added plot material for the player to discover instead — ideally, a hint that raised more questions than it answered, something that would both reward him for getting part-way through the puzzle sequence and keep him interested in what would turn up next.
The other point had to do with managing player attention. The more time a player spends in the presence of an unsolvable puzzle (say, a door he can see from the first room but that stays locked until half-way through the game), the more importance he tends to attach to that problem. It’s a huge let-down to walk through that door and find that it leads to a broom closet with one cheap treasure in it. So the big puzzles, the puzzles the player has been taught to care about, should pay off in multiple ways at once: *both* major new game-play *and* major plot information. –Emily Short interviewed by Jim Munroe –Inside Interactive Fiction: An Interview with Emily Short (Gamasutra)