To which I say: A combination lock isn’t a puzzle.
There are two ways that puzzles can appear in an RPG scenario: indirectly, as just another feature that the characters need to beat with the appropriate mechanic (skill roll, magical power, etc); or directly, as a challenge to the players.
The first doesn’t really require much discussion. Some kinds of puzzles, such as bank vault combinations, will occur naturally in the game world and if the characters want to get inside the vault they’ll have to figure out how to deploy their abilities to do so. It could be the whole focus of an adventure, say in a caper scenario, it could be a feature of a larger scenario, say a wall-safe in a room where a murder has occurred that might contain a clue, or it could be completely coincidental, as when a group of super-heroes interrupts a robbery in a bank. In any case, the exact details of the puzzles–even the solutions–don’t matter because the players aren’t expected to solve them. If there’s any solving to be done it will be at the abstract level of the characters.
The only reason to even mention the indirect kind is because scenario designers will sometimes confuse the two types, and present the details of an in-character puzzle as if it were a puzzle for the players, which leads to sucky “puzzles” as above.
Puzzles that directly challenge the players to solve them are a completely different kettle of fish. The first thing you have to decide is whether the kind of game you’re playing has any place for such direct challenges to the players at all. If the players want to approach everything from an in-character standpoint, player challenges are problematic. The player may even see the solution, but feel obligated not to point it out because the character wouldn’t get it. Now that’s a recipe for frustration right there, or even hard feelings at the table if not all the players are as dedicated to fire-walling in-character vs. out-of-character knowledge and insights.
Even if you decide you’re comfortable with direct PC challenges, you have to decide what the price of failure is: what are you willing to put up with in game if the players just can’t solve it. In old-school challenge-oriented gaming, the answer tended to be “if they can’t, they can’t.” If they don’t figure out the puzzle opening the door to the lower depths of the Liche’s tomb, and they can’t figure out a way around the door, they can’t go down there. Maybe they can come back later when they have an idea. That probably seems unduly harsh to most gamers nowadays, who may have an entire plot line riding on the party getting past that door, or just see having the players sit around the table brainstorming for an hour trying to figure out the puzzle as boring and a waste of time. You could give the players an out of letting the characters roll for it if the players don’t make progress after a certain amount of time, but not only is that likely to be a let-down, you still face the possibility that the characters will fail and the party is still stuck. At that point you could give in and give them a clue, or have a bright idea strike a party member or NPC from on high, but you have to wonder whether putting the puzzle in was a good idea in the first place. Whatever you decide, it’s best to have thought it out before-hand. With forethought, you could place clues or resources before-hand, or design an alternate (perhaps more dangerous) route so that solving the puzzle can reward the players, but failing to solve it won’t get them stuck (see “When Failure Is Not An Option“). Nothing you’re likely to decide on the spur of the moment as the players are getting frustrated and cranky will be as good as what you can design in from the start.
But supposing that’s all settled, and you’ve decided that the players will have fun with a challenge to their puzzle-solving skills and decided what to do if they get stuck: how do you design a good puzzle?
Good puzzles can be solved by reasoning. Anything that requires a brute-force approach isn’t a puzzle, it’s (at best) a time-sink. It’s possible that the puzzle can require some bit of knowledge about either the real world or the game world, though in the latter case you had better be damn sure that the players actually know it and it wasn’t buried in some twenty-page back-story that you handed to them and they never read. Let’s look at the liche’s crypt again:
Levers numbered 1-6 that need to be in an arbitrary pattern (up, up, middle, down, down) to open the door: not a puzzle.
Suppose, though, the levers weren’t numbered. Instead, each of the six levers has six positions. Next to each position is a letter. The letters, top to bottom, read: D, E, F, I, N, R.
That, at least, is a puzzle. Moreover, it’s a puzzle that a fantasy RPG geek is particularly likely to get.
Which brings me to another point: unless you’re running a straight-up challenge-oriented game where everybody will happily spend an hour or two wrestling with a juicy puzzle, good puzzles in RPGs should be simple. Complicated enough to give that little buzz of aha satisfaction when they get it, but nothing the players are going to agonize over unless you’re absolutely sure that agonizing over puzzles is how they’d prefer to spend their RPG time. Err on the side of the obvious, which at least won’t slow down the game. Also, unless you have personal knowledge that one or more of your players is quite familiar with something, trivia and pop culture references can be dangerous. Just because they seem blindingly obvious to you doesn’t mean the players will have the slightest clue, and unless you let them Google it, they won’t be able to obtain one during the course of the game, either.
Good puzzles should also relate back to the adventure at hand, if at all possible. Don’t just plop a riddle or a logic puzzle that you got off the Net in the middle of your adventure unless you can relate it back to the themes and motifs of the rest of the adventure. A riddle whose answer is “The Sun” might be good in the Temple of Ra, but seems kind of random in the Tomb of the Liche Lord. The lever puzzle might be improved by using the Liche’s original name as the key instead, if that’s something the players had good reason to know.
Good puzzles, at least in RPGs, also should reward the players immediately. If it’s a clue, make it a big honking obvious one, not something that itself further confuses the players or that they file away in their minds and might or might not remember later. And for heaven’s sake, don’t punish them for figuring it out. Yes, if you really were a diabolical Liche Lord designing a room, you might think it’s hilarious if you arrange it so that “success” in figuring out your lever puzzle releases a swarm of flesh-hungry beetles into the room, but while the Liche Lord may be a dick, you the GM are not. And unlike the Liche Lord, as GM you’ve got an interest in the players continuing to accept and look forward to your challenges.
I think the best puzzle I’ve ever put in a game, at least from the players’ point of view, was in a game of DC Heroes, where I cut up a picture of Tutankhamen’s mask into a simple jig-saw puzzle (about 7 pieces, iirc) and had the Riddler leave one piece of the puzzle behind at each crime scene, leading up to a big Museum of Antiquities heist. These were college-age players, so it’s not as if a 7-piece jigsaw puzzle was a real challenge, but by careful cutting of the pieces and arranging which order they got them I was able to keep them guessing until piece 5 or 6, and they really seemed to appreciate the whole thing out of all proportion to the difficulty of the puzzle or the amount of time I spent preparing it. And when they figured it out early (before they had the last piece(s)), I let them get to the Museum in time to be waiting when the Riddler and his crew arrived. That was a moment they really enjoyed.
In conclusion, if your players are interested in being challenged directly, and you’ve thought out the stakes and the fall-back if the players are stumped, puzzles can be one of the most satisfying things in a game. After all, it’s not the players actually swinging the sword, or flying the chopper, but it is the players themselves matching wits with the puzzle and it’s not lucky dice but the players’ own triumph when they succeed.