You've just bought the latest novel of your favourite author. Your favourite chair is waiting already, no-one's to disturb your private reading hour. You've been looking forward to this ALL WEEK, and now you sit down and dig into the plot. It's GOOD: Maybe even better than the last one, and you find yourself enjoying the whole book a lot. Perfect bliss ensues.
And all of a sudden... the pages of chapter 5 to 7 are glued together, chapter 9 is written backwards, and all pages with an uneven number have only empty spaces where a vowel should be. Chapter 3 reads like a summoning spell for Great Cthulhu himself! And in addition to that, from page 666 onwards, there's only blank pages.
These things don't happen when you buy books. Books require no effort to follow the plot (though it may be a complicated one), turning pages is all you need to do. When you start on page one you can rely on finding the last page eventually. Books are easy.
Adventure games, on the other hand, are similar to books in that they tell a story, and totally different in that they require the player to put some effort into FINDING the story. That means that sooner or later some gears will grind to a stop, and the player is facing a puzzle he can't solve, and the plot reaches a more or less long "pause".
An adventure game tells its tale in small chunks. It's usually a bit of exposition or background story that gets you started, then interactivity in form of puzzles takes over, and once you reach a certain "key event", you're fed some more plot. That's how adventure games do it, and there's nothing wrong with that. Some games have the plot tied to the puzzles in a better way than others, some have a cooler plot, some have a better pacing, but basically they all work the same: Plot, puzzles, key event, more puzzles, rise and repeat.
If you were to play an adventure game in a "perfect run", without ever needing to figure out a fiendish puzzle, you'd get the illusion of an interactive movie, where your actions nicely lead from cutscene to cutscene, and a satisfying plot is created by doing the proper things. If you were to record such a perfect run you would end up with a movie, or cartoon.
But adventure games have puzzles, and usually there's at least one really tough chestnut that has the player stumble. Maybe he doesn't see the puzzle itself, or he can't figure out how to solve it, or he falls victim to a red herring- either way, he will do what players do: He'll start revisiting locations to see if he has missed something, he will retry all dialogue options, he'll start using everything in his inventory on everything else. And eventually he will get a walkthrough.
Personally I have nothing against hard puzzles. I don't even mind "guess the designer's mind" or silly puzzles as long as they make some sense within the game's world. While I like the stories adventure games tell me, I equally like the way they tease me to figure out something.
Finding the key to open the door is not a puzzle at all, that's just some interactivity.
Finding a way to dispatch the door's guard by figuring out how much he needs to go to the loo, and them offering him a lot of iced tea and telling him about my last trip to the Niagara Falls, now, that's really something I can admire.
But either way, once the player hits the bottleneck such a puzzle invariably creates, a certain flaw can become very, very obvious: As a rule, adventure games are EXTREMELY static. Time advances depending on the player's actions. It will remain midnight until the hero unearths the mystic artifact from the crypt (even if it takes you days to figure out the rune code), the merchant will continue to tell you that he will sell the cool weapon any second now (but will still have it ten hours later when you finally figure out how to steal it), and the damsel in distress will keep her hold on that frail branch for however long you need to get that rope from the sunken ship TEN MILES AWAY.
As long as your player can easily dash through the game, you have good flow. As soon as he's stumbled, your nicely set up world turns into a lot of loose ends waiting for the player to pick up the thread again. You could say that one of the main purposes of an adventure game (intellectual challenge) totally contratics another main purpose (entertainment).
Now, you can hardly do anything about players being stuck. You can of course resort to only very simple puzzles, but that can become boring and still is no guarantee that no-one will ever get stuck. There are people who got stuck in PHANTASMAGORIA.
But you can do a lot to minimize the risk of everthing coming to a full halt with just a few simple tricks.
First, do not make your game a linear string of puzzles. If there is Puzzle A that leads to Puzzle B, which leads to Puzzle C and so on, you already lose a lot of that "it's a real world out there" feeling. Still, several games are linear and still very enjoyable, but one catch remains: If you're stuck on Puzzle X, you're totally stuck. There's nothing to go back to (because you solved all puzzles before), and you can't find your way forward.
Make sure there are several puzzles open at once. That way, once the player is stuck on one puzzle, he can at least turn to another one for a while. It's common practive to have puzzle overlap, too: Halfway through Puzzle A you will get a clue about Puzzle B, and maybe even an item that's needed to solve it, and so on. Personally I love the way certain games drop more and more hints about a certain puzzle that the designer realised to be a potentional bottleneck (Gabriel Knight did this to great effect). You are stuck, but you still can do other things, and the one thing you were stuck at suddenly becomes much clearer.
Second, refuse to freeze time. If your game has several NPCs that, once you're stuck, just repeat their old lines and basically just sit there waiting for you to figure out how to tame that nasty eel in the bucket... you break some of the illusion. It's very easy to make these static characters at least a bit more dynamic. Let's say there is a complicated puzzle in a certain location where there is also a wry old sailor. Make a simple counter, and have the guy come up with a few snappy comments depending on how often the player failed to figure out that ancient flag code: "Eh, when I was a kid, we knew these codes by heart after two tests!" or "Aren't you getting tired of fiddling around with these flags?" Heck, if you want you can even make the sailor cry (after 20 failed attempts): "Lad, it's BLUE RED RED BLUE!"
Alternatively, have people wander around. Add random muttering. Anything, really.
Third, balance your puzzles. You are the designer, all right, and may not see the wood for all the trees, but you should be able to guess if a certain puzzle is complicated or not (might take some practise; I was extremely surprised how much of a game stopper the "cherry bomb" in my own game OUAC turned out to be). That's where beta testers are so very useful, if possible, have your parents, non-computer-owning friends and your DOG play the game too!
Then check if you have a series of very hard puzzles all in a row. These can turn out very frustrating; try to loosen things up a little. After a very fiendish puzzle, open up a few simpler ones. Or break up a hard puzzle into several steps, each one just a tad simpler. Shifter's Box did this well: While very linear, it offered a rather different challenge with each puzzle, and some were intentionally easier to figure out.
Finally, never see a puzzle as separated from the whole game. It stands to reason that the plot connects all puzzles, and the puzzles themselves are a way to keep track of the player's "position" in the plot. One can design a lot of puzzles, restrict them to one location, and once they are solved, they just disappear from the game. A very good example is the Vulcano Puzzle in Curse Of Monkey Island, where our hero makes a volcano erupt to melt a small kettle of cheese (!!!). I mean, half the island is covered in lava afterwards! Do we see any reaction? No. But we have molten cheese now. If a puzzle has some consequences, you win big time: You can have people comment on it, enriching your game's "illusion of being a real world", and maybe you can take advantage of that change in another puzzle. If you always try to remember the whole picture, blowing up a vulcano offers a lot more than molten cheese.