Sunday, 31 August 2008


One big problem I experience again and again with GMing is managing to genuinely surprise the players. How can you throw them into a real state of confusion and fright, as opposed to just fake, in-character shock? In other words, how do you get around the fact that the players will have read most or all of the supplements for a given game, and are highly unlikely not to have heard of that special trick you have up your sleeve?

To illustrate: a friend once told me a story about a D&D game he was involved in. He and the other players were walking along a road in a forest and came across an old woman apparently caught in a snare, who begged them to help her. They immediately assumed the worst, threw a sack over her head, and beat her to death. And sure enough, it turned out that she had been a wolfwere. The look on the DM's face must have been priceless.

A funny story, but annoying and unrealistic in the extreme, and one that sums up the problem perfectly: the players are suspicious not due to any in-game reason; they are acting and reacting based purely on the supposition that the DM must always be up to something. Verisimilitude goes out of the window. I call it The Verisimilitudinous Wolfwere Problem Brought About By Too Much Player Knowledge, or TVWPBABTMPK, for short.

One way to get around it The Wolfwere Problem to cheat, put on your best poker face, and bluff that no, actually, the wolfwere wasn't a wolfwere at all, and oh look, you guys just killed the local almighty archmage's dear old grandmother... But there's something deeply unsatisfactory about gamey tricks like that - it feels like reloading a save in a PC game - and besides, my poker face has never been good enough to pull it off anyway.

The better solution is to get creative, by playing around with the 'standards' and well and truly messing with your players' expectations. I'll be putting some ideas up tomorrow on how to do this - put your own in the comments, why don't you?


  1. The best advice on this matter I've seen was in Feng Shui. There was a rule in there which said that if any player started learning the monster stats, then just bulk up the creature in question, give it a couple of new powers, and let the players stew.

    The discussion over at rpgnet about the Mystara monster manual suggests something similar; there are plenty of strange variants in there, like the minotaur that causes disease, things that can really surprise the players.

    That's the kind of thing I would do; add little twists to existing creatures. What happens if an ogre gets lycanthropy? Wouldn't it be a shock if that goblin happened to have quaffed a fire-breathing potion just before the players came in.

    Oddly enough, it's not a new idea. I think we've become used to using the texts as is, and a lot of the innovation of the early days doesn't happen; take a look at the scenarios in early White Dwarfs, where you'd barely see a canonical monster ripped straight from the manual.

    Of course, not every monster can be an exception, otherwise the effect is lost.

    And that's just monsters. I'm not sure how you'd add surprise in other parts of the game. A Dusk Till Dawn genre-switch halfway through the scenario is always good, as is putting the players through a plot which isn't standard for the game genre, like a murder mystery in a fantasy game.

  2. What's wrong with actually having it just be a poor old woman? In a situation where the players are going to assume the GM is up to no good, it's win-win - if they assume it's a monster and kill it, they've just slaughtered an old woman for no reason. If they don't, off she goes on her way with no real effort on the party's part, and it will drive the players nuts wondering when she's going to come back to haunt them.

  3. Which expectations are we messing with?

    I mean, you seem to want to be able to use game events to affect the players. Meaning, you're disregarding the characters' presumed feelings, and treating them as tools with connections to the players. But then when the players also treat their characters as tools with connections to them, that's a problem?

    If I was trying to surprise players in that situation, I would definitely try to subvert the expectations of the genre. One easy way to do this is to put them into another genre with well-known forms, like Kelvin said. Police procedurals alone can give you a hundred plots, a years worth of sessions after the players agree to be deputized by the local constabulary. Make them try to find out who killed a guy. Make them try to track down drug smugglers. Make them go undercover with the local crime syndicate, or in the prison (or rather, the dungeon).

    The other easy way is to entirely reverse some expectation. You know, send them to rescue a baby dragon from a princess' castle, or whatever.

    I remember this one article from Pyramid.

    It started off by saying that this worked best with a game that had a medieval setting, but which was not D&D. I believe Dark Ages Vampire was given as an example.

    The characters are led by rumor to a tomb or tower or whatever. In it they find a large bare room. In the middle is a graven stone orb. It radiates magic. Eventually, someone will touch it. It will fizzle. They will shrug and go back to the town.

    The next day, there's some problem. Farmer Engilmer's daughter Flora is missing. Someone saw orcs. There's an encampment to the west. The players go, there's orcs, they kill them, rescue flora, hailed as heroes in the town, carouse, rest, recover.

    Something else happens. Giant wolf spiders have been killing all the cattle near the Big Cave. Please help us, in they go, kill spiders, hailed as heroes, refill potions, carouse.

    This continues as long as you can make them do it.

    But then eventually, something will seem off. First, names or places might repeat. Farmer Flower's son George will be kidnapped by kobolds living in the Big Cave. Second, isn't it a little odd that there's always something happening in this little town? Players will get suspicious. They may try to leave. Villagers will stop them, telling them about the wargs killing cattle near the yadda yadda, or that the bridge is out (did we cross a bridge to get here?), or that snows are going to fall soon (what month is it?). Villagers insist they're not _trying_ to keep the party hear, though they are very glad that's how things turned out.

    Eventually, players will make a concerted effort to leave, or to find out what's keeping them there. They might even go back to the orb. Either way, there's a flash, and they're in the room, except all they find they can barely move from soreness and hunger, their weapons are rusted, their food is rotted, and the dungeon between them and the exit has been repopulated by wandering monsters in the obvious weeks that have passed.

    Heh. I kinda thought that was a cool idea. I don't know how well it would work with D&D, though. Wouldn't the players just assume that the DM wasn't all that creative, shrug and keep killing monsters?

  4. Kelvin: I think the problem with the genre switch is that it might piss some people off to find themselves suddenly involved in fantasy cops and robbers, when what they really want to do is heroic high fantasy. With most groups it wouldn't be a problem, but you never know. Explaining it beforehand would spoil the surprise, but it might not be a pleasant surprise for some.

    Patrick: The old double bluff, eh? Or you could triple bluff, so that the old woman who is really just an old woman is actually a wolfwere... So the players start thinking that the tricky DM is just being tricky when in fact he is being even more tricky than they ever suspected.

    Nick: Wouldn't the players just assume that the DM wasn't all that creative, shrug and keep killing monsters?

    That was my immediate thought. It's funny how player expectations in D&D are so...well...fucked up, in many respects.

  5. But if the players want to play to the genre of heroic high fantasy, then they wouldn't kill random women and stuff them in sacks, right? Because heroes in fantasy help people and right wrongs, right?

    It seems that D&D is its own genre. Even back in the old editions, the characters are primarily defined by fighting, not examining crime scenes or diagnosing rare diseases. It might just be impossible to make a character in the rules that is a policeman. I guess in 3E there was the Expert class?

  6. The old lady is a wolfwere, but she's a neutral good wolfwere...

  7. ...but she's being possessed by a fiend.

    (I wonder how far we can make the chain go?)

  8. Eberron actually had a prestige class that revolved around being a film noir detective. They were a major part of the setting and everything. But that's Eberron.

    The only advice I have on the issue is to start building scenarios that only work if the player's make standard D&D assumptions. (Or whatever.) I guarantee, they will immediately start thinking of every other possibility, and if someone suggests the obvious answer, the others will declare it to be far too conventional for such a twisted and sadistic DM.

  9. The reason that players always assume it's some plot to trick them, or some adventure hook, is because it always IS.

    If the players occasionally meet things that aren't adventure hooks, combat encounters, or plot devices, they will stop seeing everything as such. The trick to keep them from beating the random traveler to death for no reason is to have more random travelers that aren't secretly polymorphed terrasques.

    That being said, random encounters that aren't combat, plot hooks, or otherwise notable things can seriously slow down plot progression. It's a choice to make at least.

  10. Expectations are created. If the old woman is always a wolfwere, yeah, the players are going to act on that. They learn the rules of your world from you.

    I run very investigative games. My players never assume the old woman is a wolfwere. Instead, they assume she's a spy, or a dragon in disguise, or possibly even a goddess, because that's what happens in my games. In any event, they'll almost certainly engage her in conversation with an eye towards discerning her true identity and how she fits into the plot. They'll also assume that she has more details about the local area and the conflict they're currently engaged in.

    Of course, this doesn't help me surprise them in this way because they're also cautious, just in case she turns out to be a wolfwere assassin sent by their enemies. Sigh...

    - Brian

  11. Also, go here, scroll down to JMS Speaks, and read the second entry, the one that starts "By the last few episodes..."

    Same idea.

    - Brian

  12. I think GMs need to talk with their players more in advance about what their expectations are. If the players think the game is about trying to outwit the GM, and the GM thinks that it's about telling stories, that's a recipe for hard feelings all around as they treat the GM's story-hooks as traps. And vice-versa if the players just grab onto what they think are story hooks and the GM keeps going "Gotcha!" and wiping them out. You need to get on the same page, particularly because what's kosher for one approach is forbidden in the other.

    If the players are playing against the GM, then it's cheating of the worst sort (at least from their point of view) for the GM to retroactively change the encounter to punish them for their correct suspicions. On the other hand, it's expected that the GM will try to thwart them by having encounters be a mixed bag: sometimes it's a wolf-were, sometimes an innocent, sometimes a goddess in disguise. But, if you run that sort of game then you have to provide clues the players can pick up on as to which sort of encounter it is. Otherwise you're asking them to play Russian Roulette without even knowing how many chambers are loaded, and nobody is going to agree to do that for very long. Also, it has to be clues for the players, not for the characters. Having success or failure depend on whether the character makes the roll to notice and interpret the clue is just Russian Roulette again. The deal has to be that if the players ask the right questions ("What's she wearing? She's dressed in plain homespun, but she is wearing a small wooden religious symbol: it's the open hand of Ylana, Goddess of Mercy") they get the clues--and no extended double-think of the Vizini the Sicilian variety unless the players are particularly good at it and enjoy it.

    On the other hand, if the players and GM agree that they're not adversaries, then it can be fine to (for instance) only bother to play out encounters that are part of the plot; the players agree not to use the meta-game knowledge that the encounter must be significant or they wouldn't be having it and the GM agrees not to waste their time with a lot of meaningless activity designed to thwart their meta-game reasoning.