This post is inspired by a recent post on Really Bad Eggs, which opened like this:
Me: . . . [I]f you sat down to play a game of, say, swashbuckling adventure, why did you create a character who desires neither swashbuckling nor adventure?
TheBigDice: What if I want to play a Jesuit priest? Someone completely genre-appropriate. But that doesn't own a sword, has never had any training in swinging from chandeliers and has no intentions of ever personally killing a man. Of course, he's got political acumen, contacts and social skills. He's a completely functional character for the setting. He just needs a reason to adventure, a push out of the door.
Me: Then he's not genre-appropriate as a swashbuckling adventurer. He's suitable as a period-appropriate npc, not a player character.
TheBigDice: See, now we're getting somewhere. What you're actually saying is, in a swashbuckling adventure, everyone must play a swashbuckling adventurer, or the campaign isn't about what you as the GM wants it to be about. So we're not actually in a true sandbox here. It's probably got aspects in common with a sandbox, but given that you're saying no to a primarily social character in favour of action oriented ones, I wouldn't say that the "open world" ethos is really on the table. If you're unwilling to accommodate a character that talks his way out of situations rather than fights his way out of them, and aren't prepared to give what [Mike 'Old Geezer' Mornard] described as an unwilling adventurer the push the character needs, then are you really giving your players the freedom that a sandbox seems to demand?
Apparently this is an argument about games and character options. However, it seems to me TheBigDice is using arguments about character limitation to criticize sandboxing, without entirely understanding what a sandbox is. Black Vulmea's post covers this, but there were a couple of aspects that I wanted to mention.
A Game of X
It seems to me that part of the issue with TheBigDice's hypothetical character is that "a game of swashbuckling adventure" is not the same as "a story of swashbuckling adventure". The Three Musketeers is a story of (amongst other things) swashbuckling adventure, and features characters who do other things than fighting. In fact, such characters can play very important roles in turning a swashbuckling adventure from a series of lovingly-described fights into an actual story. They can be major characters in a story, with great influence on events.
Take such characters in isolation though - the simple country girl, the well-meaning priest, the villainous noble, the protective mother, the thuggish landlord - and have them interact, and they will reliably not produce a swashbuckling adventure. Confront them with the typical challenges faced in a swashbuckling adventure, and they will either overcome them or not, without ever buckling swash. This, I think, is a useful hint.
Fundamentally, I think "a game of swashbuckling adventure" is not simply "a game in which swashbuckling adventure occurs", but "a game about doing swashbuckling adventure". In exactly the same way, "a game of investigative horror" requires at minimum a) "horror"; and b) "investigation" to befall the player characters. "A game of social intrigue" is about doing social intrigue: it is about being at the Duke's ball and mingling with the crowds, about turning people to your advantage or cutting them dead, about spreading gossip (or carefully not spreading it), or displaying the creations of an undiscovered milliner, or gaining the right person's hand for the waltz. If you are, instead, sneaking through the palace attic in search of the vampire's lair with lemon in hand, or battling a horde of orcs in the street just outside the palace while the nobles chat and nibble, you are not playing a game of social intrigue. At least, not if you're doing those things for more than a fraction of the time, but quite possibly if you're doing them at all.
Note this crucial line:
TheBigDice:...given that you're saying no to a primarily social character in favour of action oriented ones, I wouldn't say that the "open world" ethos is really on the table. If you're unwilling to accommodate a character that talks his way out of situations rather than fights his way out of them, and aren't prepared to give what [Mike 'Old Geezer' Mornard] described as an unwilling adventurer the push the character needs, then are you really giving your players the freedom that a sandbox seems to demand?
We're still talking about a swashbuckling game here. A Jesuit priest with zero ability to buckle swashes, but who is a primarily social character, is not competent to do swashbuckling adventure, which is the very basis of this specific game. As Black Vulmea notes, the priest is a setting-appropriate character, but not a genre-appropriate PC. And this question is irrelevant to the openness, or otherwise, of the world. You could easily run several campaigns in the same world, each in a different genre: social intrigue, swashbuckling, mystery, survival horror. In each game, different characters would be appropriate, even though the setting doesn't change. The Jesuit priest is perfect for a game about sociopolitical interactions in a small town, or even low-action mystery (think Brother Cadfael), where actual swashbucklers would be unsuitable; you could include swashbuckler-types, but they wouldn't have any swashes to buckle, so they'd be using that background while relying on social skills and family influences.
The next quote is apparently intended to be a killer argument against character limits in sandboxes, but I think it gets it very nearly right without realising it:
TheBigDice:...What you're actually saying is, in a swashbuckling adventure, everyone must play a swashbuckling adventurer, or the campaign isn't about what you as the GM wants it to be about. So we're not actually in a true sandbox here.
Not quite. What I'd personally say is, "in a swashbuckling adventure, everyone must play a swashbuckling adventurer, or the campaign isn't a swashbuckling adventure".
Okay, that's wording it very slightly too strongly, but it's close. Any PC needs to be able to engage with the core of a game, be that mechanical prowess, skill set or attitude. A Call of Cthulhu investigator who adamantly refuses to do any investigation whatsoever or engage with the situation at hand is inappropriate. A D&D wizard who refuses to learn any spell not to do with interior decoration is very unlikely to be a functional companion in a dungeon-crawling campaign (though they might do perfectly well in some other D&D-based campaign).
To take an extreme example, imagine a four-player swashbuckling campaign in which every single player creates a pacifist Jesuit priest. No swashbuckling is going to happen. Is this still a swashbuckling adventure? Of course not. It may be a perfectly functional political game in the same setting - and the players' chocies may indicate that that's what they're rather be playing - but you cannot describe it as a swashbuckling adventure.
This is not simply a case of grumpy GMs complaining about their precious game being spoilt. Players commit themselves to a particular type of game, and everyone is expecting a particular type of experience (though of course, games can change over time by mutual agreement). An inappropriate character is very likely to undermine the game experience for everyone. A character who refuses to join in with the rest of the party's activities must be either dragged along (awkward and hard to maintain), left behind (boring for them, and causing split attention) or allowed to limit the party's activities to those they prefer. If four people have signed up for a game of swashbuckling adventure, but find themselves stuck trailing around behind a priest intent on using "political acumen, contacts and social skills" to solve every situation, they are not getting what they wanted, and may well not be enjoying it. This is especially likely because characters who have been created as swashbucklers or semi-swashbucklers, in accordance with the genre, probably aren't much good at politicking. What this leads to is one player controlling the game and hogging the limelight, consciously or otherwise.
Another point in this particular case is that action games tend to feature physical danger, and anyone lacking the ability to defend themselves is at risk. Even if the player decides to tag along with the swashbucklers, therefore, they will not be able to do their fair share of the work. They will need defending from enemies, placing other characters in unnecessary danger. They are more likely to be injured, and need healing, or to need help to overcome physical obstacles - the priest described above won't be leaping from burning buildings onto a horse, or vaulting aside from a runaway coach. While the stereotypical wizard can end up in a similar position in fantasy games, they compensate with an array of useful abilities that can defeat enemies, protect allies, and accomplish extraordinary feats within the range of likely situations faced by the party. In contrast, the priest's abilities are mostly relevant to situations the rest of the party have no particular reason to get involved in. It's the equivalent of putting a well-honed Fighter into our social game: her astounding prowess with a double-headed axe and ability to shrug off poison are of precisely zero use when trading bons mots with the Duke, while her complete inability to flirt, dance or discuss poetry will be a crippling social burden for her friends.
Taking a more coherent example: A feeble decadent aristocrat, a psychic Inquisitor and a manly manly Space Marine are all perfectly canonical and appropriate characters for an RPG set in the Warhammer 40,000 universe, but each is fitted to completely different genres. The aristocrat thrives in political games, the Inquisitor is ideally-suited to rooting out mysteries, and the Space Marine is a perfect soldier. However, the first two will be trampled underfoot in the first rush of Tyranid termagants, and the aristo can't sneak or banish worth a damn, while the Space Marine has no small talk and stands out like a sore thumb when you're trying to blend into a slum on the trail of depraved cultists. Each game type has not only particular things you need to try and succeed at, but its own specific dangers you need to be able to survive, be they physical, mental or social.
Sand, Sink and Genre-Specificity
The other issue I want to briefly mention is that I don't think TheBigIdea's notion of a 'sandbox' matches mine very closely, and even Black Vulmea didn't quite cover this. BV says:
Put simply, a sandbox game-world isn't the same thing as a kitchen-sink setting.
Sandbox games, in my view, are those where players can go where they want (within reason) and determine for themselves what's important and what objectives they want to set. This generally calls for a reasonably close match in skill-sets between PCs (social types, adventuring types, detective types), so that they are looking for similar types of situation to interact with. In contrast, a kitchen-sink setting is about the content of the gameworld: it's something that incorporates a disparate set of cultures, technologies, fantasy archetypes, mythologies and tropes. A game can be sandbox, kitchen-sink, both or neither.
But neither sandboxiness or kitchen-sinking is really the limiting factor in character choice. As far as I can see, the thing that sets the boundaries for what sorts of characters are appropriate (and will contribute to the game's success, rather than detracting from it) is fundamentally how genre-specific a game you want to run.
You can have a fantastically kitchen-sink sandbox game where PCs head from the Hobbit village to the Dalek headquarters in their elven steam mech to unleash Great Cthulhu on their enemies, having determined the entire course of the plot without the GM's intervention; but if that's the kind of thing you're always getting up to, then you need characters who are action-based. Mixing Wolverine, Superman and Mr Bennett won't work. Or you can have a very controlled game where the GM establishes each chunk of narrative and the players follow up on it, aristocrats dealing with the latest French threat to their social influence in London whenever it arises. You need social characters who are in a position to have social influence, so a street urchin is not an appropriate character, and nor is Captain Hornblower.
It's possible for a game to accommodate a very wide range of characters, but I think there are two main occasions when that's readily workable. The first is when a genre isn't really established until after characters are created, and then the game is planned out to accommodate whatever you've got; I haven't seen or heard of this happening, but I can imagine it. The other, more likely one, is where you're basically using the same campaign to run a variety of genres, or mashing multiple genres together, rather than having a specific game for each genre. So maybe one week you're swashbuckling, but the next is a tense political battle for control of key resources in the region. In this kind of mashup game, different characters will dominate in different segments, but as long as everyone's happy with that (and with sometimes helping out the weaker characters of the moment) then it could go perfectly well. That being said, I imagine it would still work best with genres that are relatively close together and there's at least some overlap in skills: police procedural and political thriller, say, rather than Ten Things I Hate About You and Enter the Dragon.