I have nothing of value to say. I've never run a play by post AGame, only AIM AGames.
I guess I'll just tell you what problems I've seen in other PBPAGames.
Mod unreliability. This is a huge one. Heroics died because M3D was constantly breaking his promises about when he would update things, which made it hard to maintain the interest (or generally giving a ****) necessary to be around when he finally got around to it.
Mod inactivity. This is similar to the above but not the same. Mods need to be very very active for a PBPAGame or it's going to feel like you're putting your actions in the mail and waiting for a return letter, which is again going to cripple interest. M3D had too much going on to run Heroics and thus the game was doomed to fail even if he did update whenever possible.
Design. If it's not run on AIM, you can't have any competitive elements whatsoever or it's just not going to balance or be fair. Glyph played a PBP pokemon AGame in the DR. It was terribly designed, because even when making conversation with shopkeepers or NPCs to get information -- useful information that obviously no one else would be aware of but him -- everything was played out in the thread. You can try to circumvent this by allowing PM actions like Eor did for The Thing but then it just becomes a cluster**** of when to do things by PM and when to do them in the thread. To muddy. Doesn't work. Also even in a cooperative game you probably need everyone to be in the same area at all times, or separated into groups by thread ala Heroics, because the thing I mentioned about Glyph's situation applies just as much to cooperative situations... if Eor is in the process of getting mugged and Tom is in his group but physically separated in the game world Tom shouldn't know that it's happening. So make something simple. Like everybody's at a bar at zombies attack or some **** I don't know.
Tell people a ton of information about the flavor of the game and gameplay before you make them sign up or you're going to get a ton of people who sign up and end up not all that interested.
Also ZV's game was a fun little thing but freedom is important in an AGame. Multiple choice gets boring as hell, particularly when reasoning out actions becomes based on some arbitrary and esoteric personal metric instead of logic and reason.
Use random.org for some things, but sparingly. Only for what can't be reasoned out by logic and needs to have random chance in effect. I played a non-PBP AGame that was overly reliant on randomization and it was awful. For a good example, I used random.org to resolve things like gunshots/gunfights in Batman (did he get hit by any bullets? Did it strike armor plating or a seam?) but strictly logic for things like hand to hand combat. It's very easy to make a new metric for each situation by just thinking quickly. For example when people are in a gunfight with assault rifles and whatever I use random.org on a 1-100 thing, where the values where you get shot and how critical it is depend on the skill of the character. A terrible marksman will only get hit values from around 80-100 and only critical hits at like 94+ if they're shooting each other at a reasonable distance, while an excellent marksman is going to get at least 50-100 for hit values and a much wider critical hit umbrella if the gun is accurate/well-sighted/maintenanced/etc. It's hard to think of what your values are going to be quickly when you have to weigh so many factors but since you're running this by post that won't be an issue at all. And again remember not to get bogged down by randomization. If logic works, just use it. If the Joker is two feet away from a beat cop with a shotgun in an open area with no cover the beat cop is going to get hit. If Hush is right behind the Joker with a gun to the back of his head Joker is going to die. Etc etc.
That's all I've got. Macman has run some PBP games so he's probably better to talk to. Don't think I'm going to play btw