Adversarial GM - incentives to win?

[Filip Luszczyk]

A thought occurred to me a few days ago, when I've been working out some new rules for my project. Basically, in my game GM has the following functions:

-He controls everything that the players don't (he plays the world).
-He possesses a little bit more narrative veto power than the other players (there's not much GM's fiat).
-During the conflicts he constructs challenges for players (using special pool of points), and plays the challenge against them. If he wins a conflict, his available resource pool decreases (there is some minimum) and if he loses he gets more points for the following conflicts.

I've discovered a little problem connected with the last point. I'm putting GM in a direct adversarial role against the players. I'd been expecting GM to compete with the players during conflicts, but then I realised that he lacks any incentive to try his best. Actually, if he loses, his resource pool increases (which would be irrelevant if he doesn't want to win conflicts anyway).

Now I realised, that during every playtest of the game I've been automatically assuming it is GM's responsibility to provide tough challenge in conflicts, and so it's normal for him to do his best to win with the players. Moreover, I've been somehow assuming that GM abandons this adversarial stance when out of conflicts, not trying actively to crush the PC's completely (conflicts are the core of the game and they consume most of the sessions time, but there is usually some space between them in which GM and players work out context for the next challenge).

The thing is, I want the GM to compete with the players during conflicts, and to cooperate with them other times. I don't want him to pull his punches during conflicts, as well as I don't want him to aggressively try to defeat the players. I want him to provide a fair and engaging opposition, and - what's important - to have the same kind of tactical fun as the other players have.

It occurred to me that there might be some possible problems with GM:

-Aggressive GM's who try to crush the players, stretching their adversarial role out of conflicts, are probably going to run the game dysfunctionally.
-Passive GM's, who don't try to compete with players during conflicts won't provide a fair challenge.
-GM's for whom success in competition is not a reward in itself will probably need some additional incentive.

Should I just state, that the game works under some specific assumptions, and leave it be? Or should I provide some incentive for GM? If so, what could be such an incentive? A reward system of some sorts?

Having that said, I have a strange feeling that my confusion arises from some basic misconception I'm not aware of.

[timfire]

Hi Filip,

I think for real competition you need 2 thing: First, you need to limit the GM's power. The reason that a GM in a traditional game needs to hold themselves back is because, technically(*), they can just say anything they want, including such outrages statements as, "a boulder falls on you, you're dead." But if you were to limit the GM's power mechanically, so that they are on an equal footing with the players, they can go as hard as they want and still keep things fair. You can look at Prime Time Adventures as a game that struggled with this. In that game, the GM has a fixed number of dice for the entire session, based on the PC's power (it's more complicated than that, but that's the basics).

Second, you should make it so that losing a conflict won't knock the players out of the game. If you look at DnD, the reason the GM can't go all out is because the only penalty for conflicts is damage (the only mechanically significant form of conflict in DnD is combat), and if you take too much of that, the player is knocked out of the game. I think Dogs in the Vineyard has a good example of a tacticly interesting conflict system where there are consequences for your decisions, but losing a conflict doesn't change your future effectiveness. That sort of thing may not totally be what you want, but it's worth looking at.

I think you should drop reward for losing conflicts for the GM. Or another possibility, you might enact different rewards for both winning and losing. If you provide more specifics about your game, we might be able to suggest some more mechanical solutions.
_________________________
(*) I say technically because in the real world the GM is constrained by social pressures.

[MatrixGamer]

Asking GMs to mentally shift geers is interesting. I think you are going to have to be very explicit about this in your rules and educate people about how to do this.

It reminds me about Solo Wargaming articles I read years ago. The Lone Warrior (the journal of the Solo Wargamers Association) describes lots of ways to run games. Since these were wargames the single player naturally had to play both sides. This involved a mental shift each turn. Unfortunately players often pick favorits which throw the game.

I combated this tendance in my own games by making the best Engle Matrix Game argument I could for that side on that turn. I then let the dice gods settle the matter. I detached from the process and viewed it as an interesting exercise in dice rolling and thinking on my feet, since each successful argument changed the picture of the world. I generally ruled all arguments as average (which mean they happen of a 50/50 roll) unless it was very obvious that the tide had turned against them.

When you ask the GM to be competitive - the combat system (or conflict resolution system) limits what they can do. When you ask them to step back they need a new goal to work towards. Maybe it is enjoying watching things unfold as I did in my solo games. Maybe it is making certain that the players win in the end. Maybe is it playing the good host and finding out what the players want to do next. Maybe it is making something up that flows from the last conflict that propells the players into the next conflict. Etc.

You can tell the player/GM what you want them to do. They can of course decide to ignore it but you've done your duty as a designer presenting them with a tool that works for the task you set out.

Chris Engle
Hamster Press = Engle Matrix Games

Hum... Enjoying watching events unfold as the dice gods dictate. Sure looks simulationist to me - it is just enjoying the world and its happenings.

[Filip Luszczyk]

I think for real competition you need 2 thing: First, you need to limit the GM's power.(...)

Already done. In AlternatiV every player has some amount of narrative control, thanks to the veto rules. GM has only a bit more of that power. When anyone wants to veto something, there is a vote - in a case when GM is involved, any player that abstains from voting is automatically "yes" for GM (and I've been considering dropping that rule entirely and putting GM and players on the same level - its not yet decided).

Anyway, I want conflicts to be the only real field of competition. And it is impossible for GM to "cheat" in a conflict, he is bounded by the mechanics just as other players are.

BTW, I'm not familiar with PTA, but probably my game is somewhat similar - but it's focusing mainly on tactical, competitive gameplay.

Second, you should make it so that losing a conflict won't knock the players out of the game.

That's right - it is impossible unless the player actually decides he wants to be knocked out of the game. It is also impossible to do anything serious with the elements of the story which players consider important (purchased as Threads, something similar to Relationships, but more broad and abstract).

I have no problems with those things - what I'm aiming at is actually something like much more tactical DitV. The problem that occurred to me is rather that there is actually no apparent reason for the GM to want to win conflicts. Players are awarded for trying their best to win, but right now GM is simply expected to try his best for the sake of providing demanding opposition.

I think you should drop reward for losing conflicts for the GM. Or another possibility, you might enact different rewards for both winning and losing.

The "reward for losing" is actually there for the sake of removing GM's fiat in choosing challenges. The mechanics automatically balance the difficulty of conflicts according to how well players perform. If they constantly win, things get tougher and tougher, if they start to lose, the difficulty levels itself down.

I could consider making the GM's resource pool for conflicts not variable, and simply base it on the mechanical potential of characters and the number of players. Still, I'm not sure if I want it that way - constantly facing just as difficult challenges would probably become somewhat stale.

Or, I could simply increase the pool after every conflict. But that would become an incentive for the GM to finish conflicts as soon as possible, and amass greater number of points. And after too many conflicts during one session it could be possible that the players wouldn't have chance in the final showdown. Or they would start constantly losing after some time.

In one of the old versions of the system before event conflict players awarded GM resources for creating the challenge - and then, they were rewarded as many xp if they managed to win. But I dropped the idea of xp entirely, basing player awards on something else. Anyway, it was in turn an incentive for the players to make important conflicts as easy as they could, in order to achieve story goals more easily, and to try to earn a lot of xp on conflicts less important to the story.

As for different rewards for winning and losing for the GM, I simply have no idea how could they look. GM is really constrained only during conflicts, and every conflict is mechanically entirely separate event.

If you provide more specifics about your game, we might be able to suggest some more mechanical solutions.

Here the Power 19 of my game:

http://www.indie-rpgs.com/forum/index.php?topic=19179.0

When you ask the GM to be competitive - the combat system (or conflict resolution system) limits what they can do. When you ask them to step back they need a new goal to work towards. Maybe it is enjoying watching things unfold as I did in my solo games. Maybe it is making certain that the players win in the end. Maybe is it playing the good host and finding out what the players want to do next. Maybe it is making something up that flows from the last conflict that propells the players into the next conflict. Etc.

Maybe this is the thing I lack - conflicts are the core of the game, and there's not much to do outside conflicts besides creating some story context for the next conflict. It's the same for the players - their main job outside conflicts is to look for an opportunity for the next one. Anyway, what I want is probably something similar to fluff scenes from Final Stand, where between combats there was a mandatory segment of role-playing and moving the story forward.

Still, there are practically no mechanics outside conflicts (besides opportunities for regaining the resources spent by players through role-playing, which technically could be achieved during conflicts as well).

Actually, there are two levels of play in AlternatiV - competitive conflicts form one level, and out of conflict cooperation is another one. The second level is there to provide context for conflicts that are more engaging (I think I'm aiming at something similar to jrpg's structure - lot of tactical fights, between which there is some story development). I have a feeling, that something doesn't fit there - maybe it's connected with that "shifting of geers" thing.

Hum... Enjoying watching events unfold as the dice gods dictate. Sure looks simulationist to me - it is just enjoying the world and its happenings.

Hmm... recently I discovered, that when I GM traditional mainstream games I have a strong tendency for simulationism - while normally I'm more of a gamist.

[Callan S.]

Hi Philip, welcome to the forge!

Hope this post is useful - it's a short and direct one rather than long, since it may not be applicable to your game.

Now I realised, that during every playtest of the game I've been automatically assuming it is GM's responsibility to provide tough challenge in conflicts

Who decides whether the challenge is tough or not? The players would. So the GM is tasked with something he has no descision power about, as he can not decide what 'tough' is. Only the players can. It's like he's been tasked with providing a player their favorite colour, without having been told by the player what their favorite colour is.

Any situation which involves person A deciding what person B will enjoy (when really only person B knows what they like), will be fraught with issues.

One option is to rearrange who gets to decides (what tough is).

[Filip Luszczyk]

Who decides whether the challenge is tough or not? The players would.

Well, not exactly in this system. Difficulty of conflicts is always mathematically measurable, since there is only one way to win a conflict, and it is impossible to achieve outside the boundaries of the abstract rules. One thing that comprises difficulty and can't be measured objectively is GM's will to win. If GM doesn't try his best, conflict is automatically less difficult for the players to win. That's the reason I'm wandering whether I should provide some strong incentive for GM (and what kind of incentive for that matter) - it's just not everyone would automatically take the same assumption as I've taken in playtests. And right now without GM taking this assumption the game doesn't function as it was intended to.

BTW, for the sake of clarity - you might have understood my use of the word 'challenge' incorrectly. I didn't precisely mean 'challenge' in GNS sense in this context. In AlternatiV 'a challenge' is simply the opposition in the conflict, mechanically constructed and played by GM. Maybe I should have used different term, still I didn't translated the rules yet and consequently I didn't defined every term from my game in English.

Any situation which involves person A deciding what person B will enjoy (when really only person B knows what they like), will be fraught with issues.

In this case game aims to produce certain kind of enjoyment - fun is expected to arise mainly from tactical thinking during conflicts. One doesn't have too many ways of enjoying chess - it's either playing to win, or playing for the sake of tactical thinking (which requires the player to aim for winning anyway). Chess is a game designed with certain assumptions about the players. So if someone would prefer enjoying chess by, lets say, creating elaborate stories about the figures life, it is probably not the game he seeks. Sure, one can play chess that way, but that's a drift more 'playing with chess' than 'playing chess'.

So, this statement of yours sounded a bit strange to me - why should person B reach for a game that isn't suited for their taste?

[Thunder_God]

Refresh the GM's pool after every Conflict, if he loses give him X dice, say 1. If he wins give him 5X dice, or something to that effect.

Give the GM a pool, but give him a variable limit.
If starting pool=20, initial conflict max is 5, he can bid up to 5 dice. If the characters win a conflict his limit goes up by 1, they lose it goes down by 1, to a minimum of 1.

[Filip Luszczyk]

Right now GM's pool refreshes after every conflict (just as most of the player's resource pools). Its starting value depends on the number of the players (10 + 5 per player). After every conflict he gets 1 point per player on the winning side (without upper limit), or loses 1 point per player on the defeated side (with a lowest limit of 5 per player). Besides of that the only two ways of increasing the pool was purchasing new Threads by the players (which adds a number of points, depending on the number of starting level Threads the player has at the moment) and bidding Threads in conflicts by them (which adds 5 points per Thread, but only for the duration of that conflict, and gives players chance to increase the level of the Thread or to resolve it - which in turn is the only way of increasing traits and gaining new abilities). I've been playing with the exact numbers, but I was rather sure about the formulas.

I don't want the points from the GM's pool to be bid on conflicts - they represent total mechanical power of the opposition at a given time. If GM spends these points, it is possible he will run out of them, and then a significant number of conflicts would become devoid of any difficulty. And I'm not for random increases of the resources available.

Anyway, there is something in this variable limit idea, maybe I'll use it somehow. Maybe I'll put some restrictions on allocating the points depending on players performance, and increase the pool size after every conflict by some small amount, and a little more if GM wins. Still, I need to think of some way for the players to actually decrease the size of the pool. OK, you've put my mind on the track again it seems ;)

[MatrixGamer]

After a private message I see that you want to use conflict resolution but add more options into the system than is now used in other games.

My understanding of conflict resolution now is that the players or GM set up a situation in which a conflict will happen. They set what is at stake. It could be something like this. "Bob is going to go up and talk to the supermodel. The stakes are how far he gets with her. Will he get a date or have his face crushed into the sidewalk?" Players role play the situation to gain advantage to "win" the conflict. A single die roll is made (or a bunch of dice are rolled at the same time) and they win or lose. Either way the outcome is narrated and play moves on from the consequences of what happened.

There are four steps in the process. 1. The scene is set up and stakes are determined. 2. The players role play to jocky for position. 3. The player rolls. 4. Someone narrates the outcome.

There are no mechanical tactics in this mechanism. The GM could set up a scene of their own design but the player can balk from it if they don't like it so in effect the players set up scenes. Players jocking for position are just trying to impress the GM (or other players). There are different ways to do this but that is really about social skills rather than part of the written rules of play. The single die roll is a one time event - no room at all for tactics. The real action is what get's narrated after the roll. The options people are using there are what set up the next scene. At which point in the process are you looking to add in choice?

I'm interested to see how this basic mechanism can be changed. It is a very simple tool right now - a rock, instead of a screw driver.

I use secondary choices in Engle Matrix Games. The players make arguments about what happens next and the referee/GM decides how likely they are to happen. If they happen though, the referee has a variety of tools to use to increase or decrease tension in the game.

They can speed up play by asking all the players to argue about one topic - a big dice rolling contest ensues and one argument wins. Play can move through major points quickly. It is a good tool for wraping up the game.

They can slow play down by allowing the other players to make counter-arguments to the argument just put forward. This is resolved in a dice rolling contest just like above. Play can also be slowed by calling for a second round of arguments. The referee can view an argument as starting a conflict. A second argument determins the outcome (similar to conflict resolution described above). Arguments can also slap negative statuses on characters. The referee can give them a second argument to deal with the trouble before it really hurts them.

These tools are like a throddle on a train engine. They can add intensity of drain it off. What they don't do though is put the referee in competition with the other players. The referee never controls what happens in arguments, they only decide if arguments should have a chance in hell of happening.

Are these kinds of tools like what you're thinking of? If not (and I suspect they arn't) can you telll us more about what options you want to add in?

The risk I see is that options could water down conflict resolution and turn it back into task resolution. The EMG tools don't do that since they really stand outside of the events of the arguments and instead affect the players real time playing rather than anything happening in the game.

Chris Engle
Hamster Press = Engle Matrix Games

[Filip Luszczyk]

There are no mechanical tactics in this mechanism.(...)

The way I'm looking at DitV conflict resolution (which is probably the most mechanically complex conflict resolution I'm familiar with):

There is a number of options available in DitV - you can either block/dodge, counter, take the blow (or simply surrender), you can support other players, you can roll your traits to increase available dice resources, or keep them unrolled till the last moment, as a "wild card" and so on. And there is a bit of resource management. There's nor much of it, and the openness of rolling the dice and leaving them at the table can easily change it to more of a counting contest, but there certainly are some 'tactical' choices. It always occurred to me that DitV is a system with some gamist potential.

I'm probably aiming at something similar, but I specifically focus on actually making the conflicts full of options and decisions, carefull choosing from which could give you an advantage against the opposition. Of course, my game works on different priorities than DitV - things that are most important there are completely secondary in AlternatiV. First of all, it aims to engage the players in intense thinking.

I should probably add, that the game mechanics went through more than 10 versions throughout the last year, starting from task resolution. The general structure of current version is quite satisfactory for me, though I still work out a number of details.

Are these kinds of tools like what you're thinking of? If not (and I suspect they aren't) can you tell us more about what options you want to add in?

What you write about looks interesting, but I don't think that's the thing I'm thinking of. I'm afraid, that what I propose isn't as original and something similar must have been done in some game out there ;)

My game has some CCG gameplay qualities. It uses cards to provide choices for the players, and dice to resolve abstract units of the conflict. There is a number of basic abstract actions and reactions available, but they are more similar to DitV's raises and sees than to a traditional task resolution. For example if your opposition in the conflict (what I've been referring to as a challenge) is an NPC you fight with, you probably use "attack" action to physically attack him, and if your opposition is e.g. harsh natural environment you need to endure in order to attain your goal (mechanically treated the same way any other opposition would be treated), your "attacks" will be connected with a descriptions of how you try to build a fire, hunt for animal to get its fur, seek for berries and so on. On the other hand, when engaged in a fight you could just as well "attack" by trying to calm the enemy with your words, or anything that would further your goal.

The structure looks like this:

1.Players and GM decide on the stakes (and there are some additional options for players here, e.g. connected with bidding the levels of their Threads in order to increase them).

2.GM uses his pool of points to build 'a challenge' - he allocates it between an abstract kind of "hit points", "energy", and hand size. Players face 'a challenge' that is played by a GM as one character, no matter what kind of opposition it is, and no matter if there are more than one NPC's involved. (As for "hit points", in previous versions I tried variations on Fallout, but finally I decided to keep a mechanical equivalent of hp - these values are totally abstract anyway, and measure a character's will to struggle with opposition rather than health; there is option of surrendering before you run out of "hp", or you can remain in the conflict as long as you wish, but that requires a mechanical sacrifice).

3.Everyone is dealt a hand of cards, and they play chosen cards face down as "action cards". After everyone played his action cards, they are revealed, and conflict proceeds from the best card to the worse. Every card represents a moment when cameras are focused on the character, or on the opposition - one can make some concrete step towards winning the conflict and attaining stakes, by taking an abstract "action" (an rough equivalent of a raise in DitV). There is a number of basic actions, allowing for dealing "damage" to the opposition, discarding its cards, recovering from "damage" and the like. Color and value of an "action card" matter. "Actions" are resolved by rolling a small pool of dice, which in turn can be modified, rerolled and so forth by different effects available for players and GM. (Still, since they represent abstract "steps" towards the goal of the conflict, just as DitV raises, I consider it to be conflict resolution rather than task resolution)

4.The cards that remain on hand after playing "action cards" can be used for defence (again, color and value matter), and they can be played for "special effects" that modify actions and reactions or allow for things like e.g changing the stakes in the middle of a turn. Every card is connected with at least one special effect - cards from 2 to 10 can only be added to the roll, while other cards can be used for things like "stealing the dice" from the opponent and so on. There are Edges that allow for using special effects of cards without discarding them, but using them costs resources (working name for the resource is Potential, but that can be changed depending on the genre and setting, e.g. you can just as easily have Mana or Chi or Battery Level or anything).

5.When everyone on the opposing side has surrendered (possibly because of running out of "hit points" and not deciding to sacrifice more of himself), the remaining side wins the conflict and their stakes are applied.

Basically, conflicts are played in a bit similar fashion to traditional combat from a traditional game - still, every situation that someone finds interesting and important enough to spend 15 minutes to resolve in detail can be dealt with using the same abstract conflict mechanics. There are stakes and specific "actions" are totally abstract, narratively interpreted in whatever way the players like. Still, it takes more than a single roll to resolve a conflict, and the very act of resolving them is a focus of the game.

The risk I see is that options could water down conflict resolution and turn it back into task resolution. The EMG tools don't do that since they really stand outside of the events of the arguments and instead affect the players real time playing rather than anything happening in the game.

Actually, I tried this game with players that were accustomed to task resolution, unable to accept and enjoy typical conflict resolution mechanics - and most of them had easily grasped the concept and enjoyed the gameplay.

I understand your argument of 'watering' conflict resolution - but I wouldn't call it a task resolution. At least, it is highly nontraditional because of the stakes and general abstraction of things. Anyway, as long as the game works fine and is enjoyable to play, I doesn't really matter for me whether it's exactly a task or conflict resolution. Whatever it turns out to be, I think it furthers my design goals.

[Thunder_God]

Look at This system tidbit from DevP on Story-Games, it may suit what you want.

[Filip Luszczyk]

Oh, that's interesting. It's funny that I've been playing with similar idea lately in connection with adding dice mechanics to Amber DRPG, though mine didn't work the way I wanted and I left the idea for a time. Anyway, that's not exactly what I need in this particular project.

[Callan S.]

BTW, for the sake of clarity - you might have understood my use of the word 'challenge' incorrectly. I didn't precisely mean 'challenge' in GNS sense in this context. In AlternatiV 'a challenge' is simply the opposition in the conflict, mechanically constructed and played by GM. Maybe I should have used different term, still I didn't translated the rules yet and consequently I didn't defined every term from my game in English.

As I said, my post might not apply to your game.

Are you sure you mean the GM is an adversary and should try his best? Could you mean something like that he should simply be unsympathetic?

Imagine a forrest fire. It could set ablaze a whole town, yet miraculously leave a small hut outside of town untouched (going all around it). Was the fire an aversary and trying it's best? No, it just acted as a fire always does.

Now imagine a GM playing a forrest fire and he decides whether that town burns down (which destroys the players plans). Are you sure you want competative, where most likely the GM declares the fires burn the town down so he wins? Or do you want unsympathetic, where the GM declares the fire burns down the town, simply because that's what the fire would do, with a complete disinterest in what the players goals were? So when he leaves their little hut on the hill, it wasn't GM mercy/sympathy, it was just how fires work.

[Filip Luszczyk]

Are you sure you mean the GM is an adversary and should try his best? Could you mean something like that he should simply be unsympathetic?

No, I definitely want it the other way. "Unsympathetic" is the last thing I want the GM to be in this game. Competitive it may be, I want it to be fun competition, not totally aggressive struggle for crushing the other side at every possible field. GM's job is to provide opposition during conflicts - and too cooperate with the players out of conflicts. Out of conflicts they basically work together in order to establish context for the next conflict that would be interesting for everyone involved. Then, they negotiate the stakes, and the group as a whole must be comfortable with them (it is possible to veto any given choice for a stake, so the players could easily vote the GM out if he proposed something totally unacceptable). Then they play out the conflict and this is the moment when GM assumes adversarial role. The main reason for the complexity of conflict mechanics is that they are to provide fun with tactical competition. It is a game within a game. During conflict it is everyones job, players as well as GM's, to try their best in order to win. Then, the winners stakes happen (and there are additional mechanical benefits or hindraces for winning/losing conflict - mainly you can increase the level of any Threads you bid during the stakes stage, and thus bring yourself closer to improving your Traits or purchasing new Edges).

BTW, I considered what you've written in your first post again - and I think that something similar to what you had in mind is present in AlternatiV. I recall reading some topic here at the Forge about player's choosing what is a challenge for them in gamist play - it was yours, wasn't it? I actually find it very insightfull. In AlternatiV players actually can chose in what matter they want to be challenged - that's the whole purpose of this "setting the context for the next conflict" stuff. Also, it is the player who decides to bid his Threads in conflicts - unless someone else finds it implausible at the moment and vetoes, player has the narrative power to decide when and how the things he before pointed out as important for him (the Threads) are being involved in the story and depending on the result of the conflict resolved or not.

It's just this is something that works around the conflicts, not during them. During the conflict it's only win or lose. Conflict is a part of the game in which players actually address the challenge they've chosen for themselves before.

Imagine a forrest fire. It could set ablaze a whole town, yet miraculously leave a small hut outside of town untouched (going all around it). Was the fire an adversary and trying it's best? No, it just acted as a fire always does.

Now imagine a GM playing a forrest fire and he decides whether that town burns down (which destroys the players plans). Are you sure you want competitive, where most likely the GM declares the fires burn the town down so he wins? Or do you want unsympathetic, where the GM declares the fire burns down the town, simply because that's what the fire would do, with a complete disinterest in what the players goals were? So when he leaves their little hut on the hill, it wasn't GM mercy/sympathy, it was just how fires work.

Right, but that's what is going to be decided during the stage of setting the stakes. E.g. stakes could be set in a following way:

GM: If I win, the forrest fire is going to burn the town to the ground, but you will somehow survive. [he doesn't have to state how they will survive; this can be left to decide after the conflict, depending on haw was it narrated, if only the sole fact of surviving matters - the town is going to burn anyway]
Players: If we win, our characters are going to stop the forrest fire, without any real damage to the town.

The other way would be:

GM: If I win, the fire is going to burn the town so that nothing remains.
Players: If we win, our characters are going to stop the forrest fire, without any real damage to the town. [or they could even decide that they actually want it to burn only some specific locations - e.g. the church dedicated to their rival faith]

In either case it is impossible for the fire to burn the characters with the time - the game specifically disallows death stakes for players and important NPC, unless the controlling player consents to such a stake. What's more, if one of the players had a Thread like "I was brought up in a small hut on the hill in town X" (in which case he is interested in developing a story somehow connected with the hut), then without that players consent it wouldn't be possible for the fire to burn the hut. What's more, players could simply veto the whole "burns the town" thing, because they can. So:

GM: If I win, the fire is going to burn the town so that nothing, including your characters, remains.
Players: Veto! We won't consent for the possibility of our characters dying here and now!
GM: Okay, so if I win you survive, but the whole town completely burns to the ground.
Player 1: Veto! I've purchased a Thread connected with a hut in this town. It must survive.
GM: Okay, so if I win you and the hut survive, but the rest of the town goes with fire.
Player 2: Veto! I'm interesting in facing the priests from that rival church later. It simply can't burn right now!
GM: Do you want to purchase it as a Thread?
Player 2: No, but we can vote it out [lets say other players support him] 3 to 1, so we veto the burning of the church.
GM: Okay, so if I win you survive, and the church and the hut also survive. The fire will burn everything else in the town to the ground. And you know what? It will also spread to the forest, endangering that elven settlement you know about. Are you content with such a possibility of the story development?
Players: Yes, we are. And if we win, our characters are going to stop the fire before any real damage is done.

And then they play the conflict out, and the story takes one of the possible directions with which everyone is comfortable. So not only I don't want the GM to be "unsympathetic" - it's simply impossible in the system.

One more thing I realised is that I stated something wrong in my previous post:

Every card represents a moment when cameras are focused on the character, or on the opposition - one can make some concrete step towards winning the conflict and attaining stakes, by taking an abstract "action" (an rough equivalent of a raise in DitV).

It's not exactly that the cameras must be specifically focused on characters. They are focused on something happening, and the player decides what exactly it is. Most often, it will be something that character does. But it can be something totally out of characters control. It even doesn't necessarily have to influence the conflict in a way beneficial for the character. The mechanics, not the description decide on the result of a conflict (though interesting and adequate narration can give small mechanicall bonus to the player). So it is possible for a player to describe that his character trips during a fight and still he can use an "attack" action and "damage" his opposition. His description could be vetoed by other players or by GM if the majority of the group finds it implausible - but in this case mechanical decision can't be vetoed and the player can simply adjust his narration so that it pleases everyone else.

What's more, in the example with the fire it would be perfectly legal for the players to chose a stake like:

Players: If we win, our characters will get trapped by the flames, but then an intense rain is going to fall and the whole town will be saved without any significant damage.

And then they could focus the narration of their attacks on the darkening of the sky, gathering of the clouds, animals running from the fire, and doesn't mention anything about the character's actions. They are involved in the case, and that's sufficient. Of course, in a given group such an approach could be vetoed if large enough part of the group finds it implausible.

Oh my. Didn't I just describe Universalis? :)

Anyway, I've found this thread very helpfull - it already helped me to improve some of my assumptions and to clarify some things that were constantly going around in my head but didn't surface ;)

[Paul Strack]

I am tackling very similar issues in a game I am working on.

As for different rewards for winning and losing for the GM, I simply have no idea how could they look. GM is really constrained only during conflicts, and every conflict is mechanically entirely separate event.

This is how I addressed the problem in my game. Each conflict is part of something larger (which I call a plotline). Winning an individual conflict gives the winner Victory Points towards completing the plotline, plus narrative rights on the resolution of the conflict. Losing a conflict gives the gives the defeated party Story Points which give bonuses in future conflicts. This is a self-balancing mechanism that keeps the progress of both sides roughly equal. A plotline ends after one side accumulates enough Victory Point, giving the winner narrative rights over a major event in the story.

Since I don't know the details of your game, I don't know how applicable this idea is to what you are trying to accomplish, but adding some mechanical factor that spans multiple conflicts could help resolve your dilemma.