So, I've talked about how careful you need to be not to have second-tier characters, to make sure every player will have enough interesting things to do. Now I'm going to talk about the exact opposite: every player has different preferences, so there need to be a variety of unique characters. I'm not saying one character is a spy and another is a cook: I'm talking about the nature of the choices they will make.
As a clear example, let's think about three players. Anne plays a boff-weapon LARP religiously, almost every weekend, even during the winter. Bob has never played a LARP before, and is a bit shy. Charlie has played LARPs and is brilliant, but he's tired as hell because this is the second day of the con and he didn't bother to sleep at all last night.
Each of these players is an asset to the game, but in very different ways. Any player who isn't going out of his way to screw up your game is an asset, because even poor players and imbeciles can be used as sticky widgets and information blockades... but that's getting ahead of myself. We're talking about characters, not play topology. Wheee! Made up words!
Anyway, each of these players is suitable to a different category of character.
Anne is likely to be quite aggressive, and is very well suited to a character with a lot of active, outgoing links. The character who knows everyone, or the character who wants to get things done. Anne's character can definitely be one that gets surrounded by complexity and stress.
Bob is going to be a timid little animal, so his character should be a bit simpler than the others in terms of motivations. Additionally, he isn't going to be aggressive, so instead of having very many outgoing links (people he wants to talk to), he should specialize in incoming links (people who want to talk to him). But, most important, he needs a character with very clear motivations, a character who can safely ignore most of the chaotic intrigue.
Charlie is normally a sparkling, brilliant player, but it's important not to underestimate how much "tired" screws up a player. Tired is generally most relevant in the beginning of the game, because as the game goes on, he'll either wake up a bit or finish falling asleep, and either way the problem is solved. So, while in the late game Charlie is happy to have outgoing connections and be surrounded by complex situations and choices, in the beginning he needs to be pulled into the game using someone like Anne and a cup of coffee.
There are a lot of different kinds of player, of course. There's players that like drama, players that like humor, players that are fine with being in a romance plot or crossdressing or being gay, and there's players that aren't. And, of course, there are the super-players that are good at everything under the sun. Not every game will worry about every kind of player: for example, in a comedy game, if there is a drama plot, it's probably a really silly one that even a newbie could get behind.
If you're designing your game around specific players (or are tweaking characters to fit them), you can usually figure out what kinds of players they are by asking the right questions and begging them to answer honestly. I generally use a multiple-choice questionnaire, including questions about how many LARPs they've played, whether they've ever done any acting, if they're comfortable with romance, whether they're funny, whether they like being confused...
Of course, if you actually know the players, these questionnaires are irrelevant. You should be able to tell what kind of capabilities someone has within ten minutes of meeting them.
The more difficult problem is when you're writing a game "for posterity". You don't know who the players will be.
In this case, you can make your job easier by issuing a few restrictions: "For experienced LARPers only" or "serious drama!" or "a great first LARP!" This will limit the kinds of players that will play.
However, whether you restrict it or not, your real job is to make the characters in the game run the range. You can then cast in much the same way - a simple questionnaire - but the players are assigned to these pre-existing characters.
This is where a lot of LARP writers get sidetracked. They start thinking of characters and writing them down, and they forget the first rule: every character must be worth playing. There should be no second-string characters.
This is not about stapling in a character for the newbie, it is not about adding in shallow, pointless characters that have no longevity or complexity. It is about making your characters have a wide range of suitability. The characters are all integral to the game, the characters are all delightfully complex with roles you can get into. It's just that some are more amenable to experienced gamers, others to new gamers. Some to shy people, some to drama queens. Some to sinister genius, some to hilarious punnery.
If you know that your audience is going to be a specific way - for example, everyone's new, or everyone's tired - you can build your characters to more exactly fit your target audience... of course, if it's ever played by people who aren't in those categories, it will fit poorly...
Anyway, something to keep in mind.