You round the corner and see a square room, perhaps thirty feet by thirty feet. A great door on the opposite side of the room is bound in iron encrusted in runes that glow faintly. There's not apparent means to open it, but on the floor is an intricate pattern of glowing lines, with half a dozen short pillars of stone resting at the intersections that looks like this drawing
Oh, you have got to be kidding me.
Wait, what? It's an ancient elven...
It's a freaking puzzle. Because ancient elves hated doorknobs.
Hey, this was a hiding place for great secrets. They had to put in safeguards!
Safeguards that any idiot with enough patience could get past. Yeah, some geniuses. Seriously, can't we just skip past his and get back to the fun stuff?
Puzzles are tricky. As a GM they always seem like such a great idea. They introduce a challenge for the players and provide a nice challenge to fight scenes. They offer an opportunity to introduce depth to a setting without And Tolkien had one at the gates of Moria, so it's got to be good, right?
In practice, they never quite deliver. Either one of your players immediately sees the solution and they breeze past it (possibly leaving everyone but the puzzle guy feeling frustrated) or the puzzle stumps them, and they're stuck and frustrated until you drop a hint or one of them goes off and googles a solution. As something that can bring a game to a dead stop, it's just a bad idea.
The instinct is to just stop with puzzles altogether. That was certainly my solution for a long time, but there's a trick that let's you get all the good things out of a puzzle *without* the problems, and it's this: don't put puzzles in a place where they can stop play. Put them right out in the open where players can see them and poke at them while they still proceed with the adventure.
For purposes of this, a puzzle is any element of play which needs to be figured out to reveal something new. A well constructed puzzle has a 'trick' to it. Once you know the trick (or tricks) the answer becomes obvious, but without the trick, it's frustrating. Like a bird-eye view (or the right-hand trick) makes a maze easy to solve, or how a cipher becomes crystal clear once you have the key.
While this certainly includes most classic logic puzzles, that definition applies to a wide range of in play elements, including mysteries and clues as well as setting secrets. While their narrative may differ greatly, their structure is very similar, and the tricks that work for one can inform on another.
Take, for example, the murder mystery. It is a puzzle (who? why? How?) but rather than stop play, it _drives_ play. Players are aware of that puzzle and they work their way around it over the course of their investigation, prodding and probing until some new insight is revealed. Most GMs are comfortable with this model, but does it really translate to more traditional puzzle boxes and strange languages?
It turns out it does, and if you want a fantastic example of how to do so, take a look at some of the shows that have JJ Abrams' name on them. The man's nuts for mysteries - maybe too nuts - but he provides a fantastic template for how to drop a mystery dead center in the middle of things without it stopping events. The difference (as with the murder mystery) is that the puzzles are created to drive play, not as mere one-offs in the course of a story.
How do you do this? First and foremost, you put the puzzle right out in the open. Not only do the characters know it's there, but so do other people who also have an interest in the solution (or more precisely, what the solution reveals). Something as simple as a locked door becomes a plot when no one can open it, there's something behind it, and there's a small cast of characters invested in the fate of the door.
Second, you don't need a solution to resolve anything. Specifically, this means that the solution is not necessary to resolving whatever other plots are on hand. If a session or an arc ends with a puzzle unsolved, that's fine. Unsolved puzzles have the opportunity to *grow*, with hints and callbacks springing up in other places. If that door is never opened, it just means its all the more interesting if the players find another such door somewhere else, only this time it looks like they're on the other side.
Lastly, don't be afraid to make puzzles that are too hard or too easy. Too hard just means that you can pay them out longer. Too easy means that they can help move the players on to the next thing with a sense of quick accomplishment. As long as you don't make a big deal out of either extreme, they both remain playable. It is only when this gets you flustered that it's going to throw things for a loop.
Puzzles really are a fantastic part of gaming. I know some people still love them for all their dungeon crawling goodness, and more power to them. But that's not the only way you can use them in your game. They have a lot to offer, if you'll just give them a try.
1 - If you haven't, check out Abrams' TEDtalk about The Mystery Box.
2 - Using elements like this to build organically can really give your setting that sense of continuity that excites fans of the most convoluted of programs. Callbacks are made of win.