Many authors—perhaps joking, perhaps not—have commented that their characters have a life of their own, and from a certain perspective they aren’t wrong; but no amount of characters working against their author can equal the tendency of the standard party of player characters to not stay together. Perhaps the answer is to rely on the fact the players want to even when their characters don’t
The glib answer is, of course, to ask the players to work together; however, as the many discussions of why a particular character stayed with/wasn’t killed by the others that cluster around most major franchises show, keeping the group together by allowing nothing to separate them damages the very exploration of character we seek from roleplaying games.
A more common suggestion is to provide an in-world challenge that they all face: start the game with all the characters locked in the cages of a brutal invader, have them all summoned to the Emperor’s court because his scryer has named them as the heroes needed to defeat a particular doom; and these can work in the short term. But, just as in the real world we are pulled between benefits for the group and benefits for ourselves (or cannot even agree on what those are), so do inciting incidents lose their unifying power over time. Once the characters have escaped the cages, who decides whether one character’s desire to start a resistance is more important finding another’s family, and when that decision gets revisited? Assuming the characters even all accept the prophecy is real, how plausible is it that they’d all put aside their current plans to work together?
Unlike characters in drafts of novels, player characters can’t just be rewritten if they don’t fit the world. Player characters are each, rightly, defined by their individuality not the convenience of the GM; so even the closest of players out-of-character will end up in heated disputes in-character.
But what those players will also do is stay friends despite supporting opposing sportsball teams, give each other free advice in their specialisms rather than charge for it, cancel their plans to help someone move house despite not having seen each other for ages, and generally not keep track of exactly who is “ahead” on doing favours.
Which might reveal the answer to a question that has plagued most GMs at least once: how do I give my players freedom to make characters they like and keep them together as a group? Before starting the action in the first session, generate the same sort of links between characters that real people have.
For example, in the Legend of the Five Rings game I mentioned last week, my character spent several months in the same training class as another characters several years prior to the start of the campaign, publicly attempted to save another character’s trusted mentor from an attacker, and has an uncle who is married to another character’s aunt; and each of the other characters had similar links. So, when we coincidentally ended up in the same market place as a wounded man ran in, we all had good reason to ignore any prejudices we might have against members of a particular group because this wasn’t a samurai from the family my character’s family had been feuding with for generations, this was Bob who played that awesome prank on sensei.
They won’t stop us disagreeing over what we should focus on next, or stop us arguing about ethical choices, but they will hopefully mean we feel interested in helping each other rather than intellectually deciding our characters would because they are “good” people, fear a superior’s wrath if they don’t, or want a favour later.