I am going to run my next 4e game like a caper, simple as that.
It actually takes surprisingly little in the way of preparation to make the transition. Characters are already highly capable and aware of their capabilities, the only real difference between the standard dungeon crawl and a caper is a simple matter of information management. All of the material is already in place, the trick is to change the players from the reactive group experiencing the dungeon to a proactive group exploiting it.
Now, I could use almost any pre-published dungeon to do this, but I will definitely be a little bit picky. First off, I need ones where the motive to go into the dungeon is clear and is more nuanced than "There's treasure there" or "The Bad Guy Lives There". Those are flat motivations, and flat motivations make for flat capers. The characters are badass professionals, not simple thieves, and there should be a reason they are directing their considerable talents to this task rather than something else. Additionally, a clear motive helps make for a better adventure because players can come back to their true goal when they consider plans. For example, if they're just after loot, then the plan is unlikely to get more sophisticated than a smash and grab. If, on the other hand, the goal is the recovery of a specific item, then the players could still smash and grab, but they might also try to trick the owner into moving it, swap in a substitute, con the owner out of it or almost anything else. A clear goal allows for specific, thoughtful action. A murky goal allows for only mess.
As such, a certain amount of urgency is usually called for. Without it, players can afford to wait for the perfect moment, and that can be dull taken to its extreme. In a caper, the characters have a LOT of information, but they shouldn't have ALL the information, so there needs to be some sort of reason why things need to happen *now*, rather than the next time the bad guys are going to town for supplies. Dungeons that also are strongly personality driven, that is ones that are the base of operations of a well fleshed out villain, also tend to be a lot more useful for a caper since they're usually designed with more flexibility of response.
But with those simple ground rules, it's a simple matter of handing the players most of the information you have about the dungeon. Give them a map, give them a breakdown of who is where and doing what. Don't give them stats or anything, but rally give them enough information to accurately describe everything in the dungeon, then let them come up with their plan.
Now, naturally, there should be a few surprises. Pick a few pieces of information to keep in reserve as surprises - tunnels that aren't on the original map, secret resources of a lieutenant, and unexpected twist or whatever. This doesn't take anything hugely sneaky or cunning, just find the things the author clearly put in as twists and then make them real twists.
And seriously, that's all there is to it.
Ok, yes, I'm skipping over a big chunk here and just handing the players the information, with the assumption that they've already successfully done the footwork and research to get it all together. On some instinctive level, it feels like I should set up some skill challenges and die rolls to make this investigation part of the play process. But that instinct is dangerous. See, the caper is not a familiar model for players, so the first time I do it, I don't want to dwell over the supporting details, I just want to get to the meat of it. Based on success or failure of this, I might take that to inform on my decisions for the NEXT caper, allowing for some rolls beforehand to tune what information the players do and don't get, but at no point do I want that to be the *focus* of play. If I really want to make players work for information, I'll make getting THAT information a caper of its own.
Anyway, that's how I'd do it.
1 - Does that sound mean? It's not mean to be but it's really what's going on in most dungeons. There's an illusion of proactivity through player directed action as they explore, but that's mostly sleight of hand. Excepting certain key decisions like when to rest, the set pieces in a dungeon might as well be coming to the players on a conveyor belt.
2 - Yes, the players might decide between right and left and that impacts the order things happen in, but that's still sleight of hand. Which does not make it a BAD thing. The dungeon is a very efficient adventure delivering device.
3 - Doing this also spares us the kind of convenient timing that guarantees the demonic ritual is underway JUST as the players bust into the room - fun once or twice, but absolutely trite after a while. If the players know that the ritual is coming and will happen in 3 days, they have a timeline to beat.
4 - Owen KC Stephens' adventures tend to excel at this.
5 - More snark, I know, but the reality is that a "twist" in the context of a dungeon is virtually meaningless. It's a disposable environment, and all the information within it is disposable too. Twists are merely unexpected. It takes some investment (as in the case of the knowledge that players are given) to give a twist some weight.