I bought Wilderness of Mirrors for one reason - because it does a really good job of modeling (in gaming) the planning stage of spy movies, caper films, etc.
In other game systems: This common element of espionage films is often a bit of a trouble in RPGs.
If there's one obvious solution (just one good way into the enemy base, for example) then there's a good chance the players will feel railroaded.
A GM who's conscious of that will come up with more than one way in - which means the GM does a lot of redundant prep work when they pick only a single approach. There's a chance that the players won't see one of the paths to success, or will value one as being much better than the other, and they'll still feel railroaded.
This results in the typical dungeon-structure showing up in spy headquarters, military camps, space stations, etc, because trying to hit the right balance is really hard - you want to keep the player's decisions meaningful, but not have to come up with a ton of work for yourself or bore everyone with a lengthy exposition dump. *
Ultimately, once you've given them the intelligence report on the site of the mission, it becomes a contest between the craftiness of the GM and the craftiness of the players. Often that contest involves the GM intentionally handcuffing themselves by leaving back doors that it's hard to believe the enemy overlooked, so that the players have a chance. If you don't, you run a greater risk the PCs will hit a wall they can't surmount.
This is going about things the wrong way - it's not supposed to be the players trying to outsmart the GM, it's supposed to be the Player Characters (who have different skillsets than their players do) trying to outsmart the NPCs. Really, James Bond shouldn't be at a disadvantage because the guy playing him doesn't know as much as the GM does about how alarm systems work. But I've ranted about that before. **
Wilderness of Mirrors avoids all of those traps. How, you ask? By not having the GM do a darned thing, and always assuming the PCs kick butt.
Between sessions, the GM only has to brainstorm a one-sentence description of the new mission. He can seriously just show up to the game with that sketchy little description, and absolutely zero additional prep work. ***
Each session starts with the planning stage, with the players collectively sketching out a loose map, and filling in details of how the defenses that have to be overcome. Each such defense or detail gets you a mission point award, which you'll use later to overcome those same defenses.
While the rules don't say this as clearly as they could, I believe you're supposed to run the planning session in-character. At least as much as possible. So it's not "Wouldn't it be cool if there were like those infrared beams, you know, set up like a web somewhere and we had to like crawl through them without touching them? Should we put that in, like near the vault room?" Instead you should be saying "Before he was captured, our Mole reported that an InfraWeb 9000 alarm system had been installed in the chamber outside the vault. We'll need to requisition the proper gear to see the infrared lines, and assign our most acrobatic agent to getting past that."
The point is, you're not motivated to come up with something really easy, or really hard - but rather to come up with something really cool that'll look and sound awesome when you circumvent it. The players should absolutely be proposing things that sound terribly difficult, but which they already have an idea on how to solve or slip past.
The difficulty to infiltrate the evil lair (or whatever the framework of the mission is) is beside the point. Your dice rolls don't determine success or failure, they determine narration rights. Success is assumed in most circumstances. If I roll poorly, I may still take out that guard - but the GM will decide his body is found by a random patrol 10 minutes later, and we have to act fast.
As with any good caper or spy plotline, the infiltration is only half the mission - getting out again is the tricky part. The setback points that the GM accumulates every 20 minutes are designed to make all hell break loose once you're inside. Expect something to go wrong, and a good chase or fight scene to break out in the second half of the night. By the same token, you might win the roll, and still choose to narrate a result similar to my guard example in the previous paragraph. Little unexpected twists get you "Trust Dice", as does betraying the group. (Gotta love the name of those Trust Dice.) Narrating a minor setback now can help arm you for controlled success at a critical juncture where it really matters.
This game is sweet.I'm very excited to try it.
*: For an idea of my own on how to avoid the exposition dump, see: Discussion Cards. That idea works well with science anomalies / mysteries and for conveying situations where the Players lack info the PCs have.
**: Pet Peeves Regarding Attributes. Since writing those words 20 months ago, I've seen a few "hippy games" that let you use mental and social traits this way. Sadly they all seem to be games set up for one-shots only, either being very gimmicky or having no XP system. (Or both.)
***: Myself, when I get around to running this in a couple of weeks, I might actually spend 20 minutes before / between sessions. I like the idea of presenting a loose structure to the map, for the players to build off. Like just drawing a coast line, or the outlines of 5 buildings that are left unlabeled for the players to define. I'll then absolutely have the PCs do the planning stage in-character.