The Simple Art of Mystery

After this weekends Detectives & Daredevils game, I was talking with the game's creator and our GM, B. Portly, and the man from Kaijuville, Steve, about running mystery-based games. This is something I've put a bit of thought into as most of my Weird Adventures games are mysteries in one way or another. Here are some things I feel like help make a mystery genre game (as opposed to game that just happens to have mystery elements or a mystery setting) successful and enjoyable:

1. Get players' buy in. To create the feel of a specific genre, everybody needs to be on the same page about what you're doing--at least if it's going to be fun for all involved.
2. Plan, but leave some blank space. You need to know the "who," the cast of possible "whos," and at least have a good idea of the "why," if you're going to be able to effectively lay clues for the PCs to uncover. There needs to be some fuzzy areas though, as the player's are going to suggest interesting details either purposefully or through their actions during play. So long as you're not changing the fundamental facts of the mystery the PCs are trying to uncover, this only enhances things.
3. The PCs always find the important clue. This one is borrowed from Robin Laws' GUMSHOE, but it can be employed in any system. If the PCs look, there going to find the critical clues. If they don't look, be on the look out for alternate ways they can discover the information. There can always be some details players' might miss, but if it's really important, don't make it hard to get.
4. Repeated interviews yield new information. As Raymond Chandler pointed out in "The Simple Art of Murder," one of the "unrealistic" things about the murder mystery is that it features a close-knit group of people. Going back to those few NPCs with new questions will get new information, because they will have thought of things since last they were interviewed or new things will have occurred as the malefactor's "plot" precedes.
7. Keep things moving. To again quote Chandler's "The Simple Art of Murder": "When in doubt, have a man come through a door with a gun in his hand." If the PCs have hit a wall or their just not connecting the dots, they shouldn't flounder too long. Their investigative actions are going to make the villains react or (related individuals with something to hide), and the reaction will often be to try to kill the PCs or throw them off the trail. Maybe the villains don't come after the PCs, but after someone else they think might give the PCs information. Their actions shouldn't be random; they should make sense, but their exact timing can be when the game needs it.
8. Everybody has got secrets. Even when someone isn't the killer/primary criminal, they may have something to hide. Hints at these provide good red herrings and discovering them gives the PCs a feeling of accomplishment while they're slowly chipping away at the big case. Be careful not to let these overwhelm the main mystery or make them too hard to discover, lest the PCs spend too much time on a tangent.
9. It's not necessary to be Sherlock Holmes. In The Maltese Falcon, Sam Spade does very little investigation. He mostly reacts to people coming after him; he thinks on his feet, keeps the other guy talking so they give up a lot of information for relatively few questions, and uses violence judiciously.

That's what I've got. Anybody else got any pearls of wisdom from their gaming table?