There's a bit of misdirection involved in all RPGs. The core misdirection is that the PCs need to feel that they might not win each confrontation, but end up winning nearly all of them. Success shouldn't be automatic, of course, but it should feel a lot more unlikely than it actually is. This misdirection falls on the GM a lot of the time (for example, to play up how fierce an ogre is when the PCs can easily defeat it, by the numbers) and the game system a lot of the time (to provide, for example, Hero Points or Fate Points that allow players to recover from bad rolls and turn failure into success), but some of this falls on the adventure designer, as well. Here are a few ways to deceive the PCs in a way that makes them feel successful, even when the odds aren't really as stacked against them as it seems. 

High Roll Wins. This example is something I see quite often. The PCs have to notice something--a mugging in a nearby alley, for example--and they all make Perception checks. The highest roll successfully notices the clue, no matter the result of this roll. I used to deride this as poor design--after all, if the PCs literally cannot fail to notice something, why not just dispense with the rolls and pick someone to notice it? But then I remembered how much players like feeling that their rolls matter. So requiring a roll and revealing something to the player that rolled the highest is saying, "good job, you JUST succeeded at a thing," which feels good for the player. So there's a place for this kind of trickery in adventure design, I think.

A Sandbox that Isn't. I mentioned this in my last blog. PCs like to have a lot of choice, and "sandbox" style adventures give that to them. They feel like they can go anywhere in the world and face any kind of challenge. But if some of the encounters would be much too easy for them and some much too hard, what should an adventure designer do to avoid boredom or overwhelming PC death, respectively? One solution is to scale the PCs' challenge based on where they are. If they hit the orc caves while still low level, there's just a few orc warriors. If they hit the orc caves at a much higher level, there are orc warlords and death priests and so on. I witnessed this myself playing Mass Effect. Near the end of the game, I went to a planet with exceptionally tough mercenaries who carried fantastic gear. Neat! On my second playthrough, I went to this planet very early on, hoping to sneak in and steal some of the fantastic gear. I failed, started a fight--and won pretty easy. Then, it turned out, the gear wasn't so great. The game had scaled to "level appropriate" enemies and gear behind the scenes. If I hadn't been playing through it twice--and few tabletop RPG players ever play through the same adventure twice--I wouldn't have ever noticed that the "sandbox" wasn't really a sandbox at all.  

Mystery Solved! A much more rare example of design is in a mystery adventure. This adventure puts several options in front of the PCs--someone's been murdered, for example, and the PCs must discover which suspect is the murderer--and has the PCs do legwork to solve it. After a few encounters, which normally depend on the suspect being investigated or the ritual being pursued or what-have-you, the PCs are informed enough to make a decision. And they're right! Success, praise, and XP all around! Sometimes this type of adventure is designed so that the PCs are always right, no matter who they pick: if they say the scarred sailor did it, then she did! If they instead think the shifty watch captain did it, then he did! (And the sailor would, in that case, be innocent.) This seems the most like "cheating" if the PCs discover it, and it can lead to a bad taste if revealed. But solutions like this appear often in mystery fiction, and it has to do with the scope of the narrative. A mystery novel has the investigator hit just as many dead-ends as needed to discover the culprit by the last page. An investigator in a half-hour mystery show wraps it up in that time, while an investigator in a three-hour movie takes six times as long to wrap up a similar mystery. If you're trying to solve a mystery encounter in a single evening of gaming, there isn't time to have the PCs hit several dead-ends (although they can and should hit dead-ends, and be wrong, early in the session). At some point, they have to be successful, and this design ensures that the success is paced with the medium.

I once asked a non-gaming friend about this sort of "heroes are always correct when they get around to making an accusation" storytelling, and she was initially very put off by it. But the more she thought about it, the more it made narrative sense to her, and she had a lot of good insights about how to pull it off. Most importantly, the heroes shouldn't ever know about it or it spoils the victory. She also pointed out that the final culprit should be believable from the beginning and, ideally, the threads that pointed to the other suspects should be quickly revealed as false (or, better, as planted by the final culprit to throw off suspicion). These are good insights to make this final type of misdirection really work!