How difficult is a task?

[Filip Luszczyk]

In many trad games the in-game difficulty of various tasks is represented mechanically via target numbers, modifiers or other means. Outside a few detailed subsystems (e.g. combat), this is often left for the GM to determine. Typically some general difficulty table is provided as a guideline (e.g. Easy: 3, Moderate: 5, Hard: 7, Very Hard: 9, or something along those lines).

The problem

This is where the ruleset's support often ends. Unfortunately, fictional events don't translate all that well into abstract mechanical terms. For instance, should disarming nuclear bomb be challenging (-20%), difficult (-30%) or almost impossible (-50%)? The GM has to arbitrarily determine that, representing the in-game reality in numbers based on its internal logic and his own common sense, knowledge or real-life experience. There tend to be only vague mechanical guidelines regarding the appropriate range of difficulty, usually too general for direct application.

Hands up everyone with real-life nuclear bomb disarming experience. Now, let's make that saddling robot unicorns. Or hell, when was the last time you bashed doors or bent bars?

Personally, whenever a game asks me to arbitrarily set the difficulty of a task, I want to cry. For instance, running Exalted years ago, instead of utilizing all five steps of difficulty, I eventually wound up setting all tasks at either 3 successes when it didn't seem like a big deal or 5 successes when it did. I still didn't feel good about it.

What I'm interested in

I'm interested in tools for deriving mechanical difficulty from in-game circumstances in "physics of the game-world" systems.

I wonder how to come up with a largely unambiguous formula for processing situations into numbers. A single procedure applicable to a wide variety of tasks outside finely defined subsystems. Rules, not guidelines.

What I'm not interested in

Now, the tricky part. I'm specifically not interested in the following solutions:

=> Forge-style tabletop games often circumvent this issue entirely. For example, there might be no tasks to resolve, no mechanical representation of in-game difficulty, or limited difficulty budget. I'm specifically not interested in tools for circumventing the problem; I already have plenty of those at my disposal when I need them.

=> Games like D&D 3.x or Mouse Guard neatly solve the problem by providing fixed difficulties for all commonly emerging circumstances. They still include general difficulty table, but I only recall rare instances when I had to resort to it in practice. Works well enough. Only, I'm not interested in this solution here; I'm looking for a single formula that would be applicable to a wide variety of tasks.

=> Some games, like NWoD or, I believe, the new edition of WFRP, suggest deriving the difficulty from the number of specific factors that hinder or help the action. This leaves too much wiggle room for my purposes, however. Say, the character is jumping between the roof at night, in bad weather. One GM could give the player -1 for challenging environment and leave it at that. Another GM could assign -2 penalty: at night (-1), bad weather (-1). Yet another could make it -3: darkness (-1), fog (-1), slippery surface (-1). That's already -1 to -3 difficulty range for identical circumstances, and I could dissect the situation further. It's just too easy to apply this tool inconsistently from session to session. I'm not interested in this solution; I need something more unambiguous.

=> Many games include opposed rolls, where the difficulty of an action depends on NPC statistics. Mechanically, NPC's profile is a set of fixed difficulties. Establishing NPC statistics is a potentially complex issue, but there are multiple working solutions available. Consequently, I don't want to delve into that in this thread; I want to focus on unopposed tasks.

=> I'm not interested in solutions that assign difficulty based on narrative needs (e.g. "This is the climax of the story, so it should be a real challenge" or "Time for the happy end, so let's make this easy") or social needs (e.g. "Bobby will be sad if he fails this roll, so I'll set a low target number to protect his fun"). Tools for both are numerous, I'd rather have more tools for preventing the GM and players from being reflexively overprotective when it comes to their characters or fiction.

=> I'm specifically not interested in "GM is the physics of the game-world" or "the group is the physics of the game-world" perspective. I want to approach the problem from "rules are the physics of the game-world" angle.

=> Solutions like "just wing it and have fun" or "good GM-ing skillz" don't work for my purposes, also.

[Callan S.]

Hi Filip,

Personally, whenever a game asks me to arbitrarily set the difficulty of a task, I want to cry.

I'm inclined to look at this critically and ask "Okay, what if the rules did set the difficulty and...it seems unrealistically hard to you, or unrealistically easy? Are you at the point where you want to cry if it leaves it up to you, but your not prepared to accept it when the rules produce something that is cockeyed to you?"

It seems like a lose/lose situation? Personally I don't like games that ask you to set the difficulty because they have no cap on the max amount (or floor on how easy you can make something) which to my mind leads to ouja board game play, where everyone denies that anyone would assist actions they want to happen and block ones they don't and they insist they are genuinely playing an unpredictable game. But that's another topic.

[Vulpinoid]

I don't want to cloud the issue further....seriously, I don't...but...

Isn't this where conflict resolution and task resolution really set each other apart.

A lot of game systems that you mention (eg. Various editions of D&D, WoD/Exalted) prepare the GM and players for resolving tasks, but they don't provide methods for integrating those tasks into a coherent storyline.

Many modern "story games" don't bother with the resolution of specific tasks, instead they seek to determine how the resolution of an objective affects the flow of the story. These games ignore the steps along the way, so you don't need to worry about the "Red wire/Blue wire issue", they just determine if the bomb was safely defused and the citizens got to safety.

Without knowing specifically what you're aiming for with this particular project (thematically, or otherwise), I'm afraid that most people will just be throwing you random suggestions. I would guess that you're trying to "Simulate" reality with the system you're devising (with a nod toward the relevant creative agenda), but even this becomes difficult because different people will have different notions of what is easy or hard.

Do you want crunchy simulationism, or soft? Crunchy takes a while, but on the other hand, there are some products really pushing the envelope with technology such as iPhones (see here). If you want crunchy, you can set up incredibly complex algorithms based on 2 parties, and environmental modifiers and anything else that becomes relevant to your game, then simply point and click. Character A (affected by circumcstances X,Y and Z) confronts character B (affected by circumstances W, X and Y), press enter...

Result [as per instantaneous calculation, table lookups, cross referencing, further calculation and pre-formatted output] occurs.

...or do you want something that a regular person can work their way through in a couple of simple steps?

I know it's said a lot around here, but you've got different tool to achieve different goals (AKA system matters).

I know it doesn't help much, but you're opening a big can of worms here.

[Jeff Russell]

With due caution for can opening and worm appearance, my off-the-cuff response is to perhaps look at it backwards. Rather than setting difficulty in a systematic way, perhaps you clearly define your resolution abilities (whatever they may be) and then have a set, constant difficulty for everything. In other words, rather than modeling 'this task is hard' or 'this task is easy' you have a threshold for 'stuff gets done', and then model character competence at accomplishing it. This doesn't really address super competent characters attempting super difficult tasks, and it might not be 'physics-y' enough, but just a thought.

[Eero Tuovinen]

Well, I'll throw out something concrete to get out of the speculative territory:

Here's a table of difficulties:

Difficulty:	Use this difficulty when...
1	When the task is pre-school in difficulty and people of the character's culture are looked sideways or pitied for not being able to execute it consistently. Example: using a key to open a door.
2	When the task is taught to all or nearly all people in the culture's schooling system. An adult not knowing how to do it would be exceptional. Example: pointing out China on a world map.
3	When the task requires specialized knowledge that is relatively common to the culture, but not omni-present, perhaps due to social stratification of some sort. There are schools for this type of knowledge or skill, but many people perform without schooling as well. Example: basic computer skills.
4

When the task is not commonly attempted by a generalist, but rather left for a specialist with specific schooling. Example: creating a website, fixing plumbing.

5	When the task requires not only special procedural skills, but those skills also cannot be applied without extensive theoretical background due to the non-mechanical nature of the task. Example: general computer programming.

The idea with that table is that you're not actually evaluating the difficulty of a task here, but rather checking your pre-existing knowledge of the type of schooling the character's culture expects of people who do this sort of thing consistently. We can say that disarming a general nuclear weapon would be at difficulty 5 here because it's not enough to know some specific complex procedure; you need to have a lot of knowledge about the theoretical construction of this type of weapon to be able to recognize and interact with somebody else's arbitrary design. Knowing how to disarm a singular, given type of bomb could be difficulty 4, I suppose.

The above difficulty classes are pretty deterministic in that if you don't possess the requisite level of knowledge, you'll fail automatically in the task. I'd probably use a second axis of task difficulty to represent the randomness involved in the task. For example, I could say that hitting a world-class baseball pitch successfully would be 3/5 or some such: it's a skill most people would be expected to be able to try, but success even then would not be guaranteed. In this manner I could differentiate between tasks that succeed automatically when you have the requisite skills and tasks that are still not guaranteed even if you execute the skill correctly.

If we'll look at theory, the above method of difficulty-setting relies on a clear pre-existing system of categorization to avoid complete arbitrariness. A GM using the above system could theoretically stop the game and log into the Internet to find out the sort of vocational level a given task is usually performed in if he didn't already know. For example, for any given medical procedure he could find out whether it's something a nurse, a general practitioner or a specialist doctor would do. The players could also argue the point if the GM miscategorized something.

[Paul T]

Filip,

You've set up quite a challenge with your set of conditions, and it's possible that it's an unsurmountable one. However, I'll give it a shot:

If you're generating a "target number" or similar "hard" quantity from a negotiated fiction, it's impossible to avoid some subjectivity altogether. Do you agree with me on this point? Because, after all, our communal imagined fiction is not a "hard" quantity--it's fluid and variable and requires interpretation by someone.

So, the way I see it, the key is to provide the person or people who are making the decision with a few very easy decisions. That way you approach "objectivity" by making the decision-making process as obvious as possible.

My suggestion is to do that by breaking down what you want the fiction to tell you into a set of decisions that are very easy to make individually, with the final "target number" emerging from a combination of these decisions.

For instance, you could break down a fictional situation into a series of simple "yes/no" questions a GM goes through, with each "yes" answer adding one point to the final target number, or something similar. It would probably be impossible to make an absolutely generic set of such questions that applies to any game, but you could make one that works well for a certain genre, or a limited set which the GM picks from in play.

For instance, imagine a list of, say, five "yes or no" questions, such as "Is the opponent clearly more skilled or experienced than the protagonist? (If unsure, default to 'no'.) If 'yes', increase die size by one." In play, some player (the GM; the player who's rolling; a player who's uninvolved) picks the three questions that are most relevant and answers those. That gives you a range of four possible target numbers.

You can probably also get a sense of objectivity by spreading the decision-making process out over several players. Have each player at the table select a question and answer it. If people seem to be affected strongly by other's answers, they could answer "blind", by picking a red card for "yes" and a black card for "no", and placing their decision face down on the table.

That's the best I've got.

[Callan S.]

A short idea comes to mind, where difficulty is based on the previous difficulty roll. Because you don't just wander across a nuclear bomb in the street - you've probably gone through some other tough stuff prior to that. So difficulty is around +4 to -4 of the previous difficulty (or some number range like that). Yes, it's still the GM calling it in the end. Because all there are are spoken words, vibrating air, around the table. Even if all that vibrating air conjures all sorts of images in your head, its just sound waves, it can't set a difficulty. Ever.

*side thought: Though wouldn't it be kinky if you had a device that measured the volume and consistancy of sound at the table and set difficulties based on that....that'd be kinky!*

[dindenver]

Fillip,
  this is how I approached it:
Base Difficulty Numbers
Easy: 6
Example: Changing a bandage
Simple: 9
Example: Disinfecting wounds
Average: 12
Example: Performing minor surgery (safely removing an arrow from the torso)
Challenging: 15
Example: Setting a broken bone
Difficult: 18
Example: Setting a bone of a struggling patient
Very difficult: 21
Example: Performing major surgery (Appendectomy)
Imposing: 24
Example: Performing major surgery on a struggling patient
Impossible: 27
Example: Performing major surgery with a improvised tools

Medicine
Description
  Characters with Medicine can aid other characters in recovery from Damage. With the right tools, characters can also help fight infection, disease and poisons.
Skill Pool: Wisdom (SA)
Difficulties
  Stabilize Bleeding: Simple, the die roll is modified by the patient's Stun Modifier
Note: Characters that are stabilized do not need to roll a Stun Save if they are hit for no Damage
  Disinfect wounds: Simple, Very Difficult for tainted weapons
Note: Allows characters to make another Survival roll to resist disease.
  Set Bones: Challenging
  Minor surgery: Average
Note: This would entail removing a foreign object, etc.
  Major surgery: Very Difficult
Note: In this era, this would be procedures like tracheotomy or removing appendix/tonsils. More complicated surgeries are not even possible.
  Minister wounds: Average, modified by the patients Damage Modifier
Note: Used to remove any serious Wounds from the character (broken bones, etc.)
  Relieve patient: Average, the die roll is modified by the patient's Stun Modifier
Note: May be attempted once per day, success indicates character may remove Damage equal to their REC stat,
  Scrounged tools: +6
  Minimal Medical equipment: +3
  High Quality, fully stocked medical equipment: -3
  Cooperative patient: -3
  Struggling patient: +3
Time and Tools
  At least some equipment is required to attempt to use a character's Medical skill. Stabilizing bleeding patients and disinfecting wounds takes about a minute, All other procedures except major surgery take 30 minutes and major surgery takes an hour.
Unskilled Use
  Characters cannot attempt Medical rolls unskilled.
Failure/Fumble
  Failure represents a setback, usually this is corrected by spending 10% extra time per point missed, but will result in the patient taking 1 Damage (DR cannot be applied to this) if the character cannot spend the extra time. Fumbles are not possible.

I dunno if this answers your problem or not.

Spycraft 2.0 has an interesting approach as well.

[Filip Luszczyk]

Ok, I have my solution. Found it on rpg.net, of all places.

It was a matter of a silly stumbling block, when I kept examining fixed difficulties in games like D&D and MG exclusively in context of their respective skills, oblivious to patterns behind them. After a few years of playing mainly games that more or less circumvent the problem, I pretty much lost any grasp of this "physics of the game-world" paradigm I once had, I guess.

I may still have some use for alternatives. I'll elaborate on the solution and respond to your individual questions and suggestions in a few days. You can safely give this thread a very low priority now, though.

[Paul T]

Please share the link as well, Filip!

[Filip Luszczyk]

The link. (high noice to signal ratio)

What I need this for

1. With a formula like this, I could hack some trad games on my shelf, and maybe find some use for the books again. I could go on and on about my dislike for trad, but a lot of that is frustration with mechanical minutiae that I could fix with various tools available to me currently, perhaps even without having to rewrite the entire ruleset from scratch. Those general difficulty guidelines were a rather fundamental problem for me, something very basic that I couldn't quite fix easily.

Like, I'm looking at Fading Suns now, and the rule is pretty much "make it more difficult when it's more difficult," and while some rather vague guidelines are given in the unified difficulty chart, the next paragraph is only asserting that I'm going to become better at this as my GM-ing experience increases (read: when I learn to abandon the ruleset or work out my own procedure for this through trial and error). But then, in the same book, the extended example of play starts with the GM basically overriding the result of a failed roll to "speed things up". So, uh, whatever.

Reading Burning Wheel recently reminded me of the problem and made me re-examine possible solutions. The procedure for setting difficulty in BW is crystal clear: "It is the GM's role to assign appropriate obstacles based on the inherent complexity of the task at hand," and the abstract general difficulty table follows. The procedure is also utterly useless to me. When the manual asks me about the inherent complexity of the task, all I can say is "it's adequate." I can almost imagine myself arguing with the manual:

The manual: What's the inherent complexity of the task?
Me: Adequate.
The manual: But what do you think is the exact difficulty?
Me: I think it's appropriate to the task, duh!
The manual: But what do you think is appropriate to the task?
Me: Whatever is adequate, obviosly!

And then I'd probably do the same I used to do in trad, i.e. just set the Obstacle and proceed with the game breaking the rule, because the Obstacle I'd set would have nothing to do with my actual judgment of the difficulty. No such entry as "adequate" in the table, I'm afraid.

Damn, I've just remember how the first time I ran a game using a purchased manual, I tried to explain the resolution process to the player, and upon reaching the difficulty table he just commented it made no sense altogether. And I couldn't quite figure out what to say. I guess the only reason I bought into the idea myself was that since the product cost money, I expected it to just work. (Well, the very first game I ran, I designed myself with nothing but a single issue of an rpg magazine to introduce me to the hobby. It didn't have task difficulty at all. It didn't work, obviously, but it just didn't have that issue. It did have ninjas and the Power of Greyskull, though, so it sort of evens out; it took long years before I had those in my gaming again.)

2. I had this though about perhaps trying to design something GURPSy.

Throughout the last year, I designed a few generic games that do various things. All of them circumvent the issue of setting difficulties entirely.

In one of those, resolution is all about maneuvering within the situation and pushing one's agenda past the GM's blocking behaviors via simple and abstract resource management. Barely any representation of character effectiveness and no real mechanical differentiation (the little there is boils down to parcelling permissions for doing stuff). Task difficulty is not factored into the resolution at all, it's all players versus the GM.

Another one is heavily crunched up Otherkind Dice, and resolution is all about producing trouble. Here, I have plenty of character effectiveness differentiation, but success/failure is largely a matter of player choice at this point, with no representation of individual action's difficulty whatsoever.

The last one is a generic, streamlined Dogs hack, but I didn't playtest that one yet. I had some stumbling blocks working out NPC profiles, though. After a few hours of statting up various sets, I've just crossed it all out, settling on a single profile for all conflicts and a budget of d10s for the GM, fueled by concrete PC actions.

So, I thought it could be interesting to make a game where the ruleset explicitly provides "physics of the game-world" and that's all there is. I'm not really concerned with this being "realistic" past some very basic plausibility. I want some "three-dimensional", solidifying environment to emerge from the ruleset, however, with the players just going with it rather than challenging its integrity or, God forbid, contributing to it significantly. So, no voting for difficulty, figuring it out collaboratively, or objecting based on real-life. More like D&D, where the players are primarily concerned with exploring the dungeon as is rather than digging into the overall internal logic of the setting, or Super Mario Bross, where nobody negotiates the placement of platforms.

Still, I have something relatively compact in mind, hence the need for a unified difficulty formula rather than task-specific approach. Also, I'm just too lazy to write up a few dozens of skills in utmost detail the way Dave suggests

I think I'm aiming for something vaguely gamist here, but I'm not entirely sure about that.

My potential solution

About 5-10 Mouse Guard style factors, only a single general list rather than many skill-specific. Maybe the difficulty starts at medium, and each factor that applies bumps it by one step (or, with low granularity, perhaps after the first bump it's one bump per two factors, or something).

On rpg.net MetaDude suggested the presence of time limit as a commonly applicable factor, and I have a bunch of others in my notes, like whether there's combat or other physical danger nearby, expert training requirement, improvised tools, insufficient crew etc. Vitenka made an inspiring list on the third page, though some of those entries might involve more judgment than I'd want. I'm partial to conditions that depend on the player's preparation or their previous actions, or allow some way around with extra effort or risk.

So, essentially, what Paul suggested: a checklist of yes/no questions.

[Filip Luszczyk]

Callan,

I'm inclined to look at this critically and ask "Okay, what if the rules did set the difficulty and...it seems unrealistically hard to you, or unrealistically easy? Are you at the point where you want to cry if it leaves it up to you, but your not prepared to accept it when the rules produce something that is cockeyed to you?"

I experienced this in some games, but not in others. Largely depends on how the game handles failure, but also whether there's any point to improve effectiveness past a certain point. It's also a matter of gameplay building GM's emotional investment into the player's success or not, I guess.

I had no problem with this in Mouse Guard, for example, where the player needs a certain amount of both successful and failed tests, and on a failed test, the GM has the option to grant the player success at a cost. Also, the last time I ran WFRP 1st ed., where the characters where literally made of fail, I had no problems with the massive whiff factor or implausible outcomes. "Thomas always had it worst in life; even when he actually achieved something, it was never anything he was good at," one of the players commented about his character, heh. However, with the percentile roll-under resolution I did not use the general "make it more difficult when it's more difficult" rule at all in that campaign, applying only fixed modifiers.

In Exalted, on the other hand, it was silly how the manual offered a wide choice of skill-boosting magic, when it was common for starting characters to routinely succeed at maximum difficulty tasks within their realm of specialization with raw mortal skill alone.

A short idea comes to mind, where difficulty is based on the previous difficulty roll. Because you don't just wander across a nuclear bomb in the street - you've probably gone through some other tough stuff prior to that. So difficulty is around +4 to -4 of the previous difficulty (or some number range like that).

It sort of reminds me of HeroQuest, which we tried out recently, and which sucked. The specific problem with something like this, however, is that I still have to think up the number. I hate having to make up numbers out of nothing, even when it's saying how many goblins are in the room just like that or something. Numbers hard!

Michael,

Well, I've found that a few years of playing games like those made it very hard for me to think in "physics of the game-world" terms. Story thinking, especially, seems to be pretty incompatible with this approach. Story = poison.

On the other hand, a friend had some interesting war stories about his old GM who applied this "physics of the game-world" paradigm to quite an extreme degree. In one of those stories, the GM opens the game: "You are standing in a city street," to which the players immediately respond: "We move to the sidewalk."

So, with this, no story to speak about. When some stray player comes to the game to listen to the story, or to participate in the story, or expecting the story to happen, I want them to suffer. If for some reason they decide to stay in the game, I want it to be a pure act of masochism on their part.

Jeff,

I'm not sure if I understand your suggestion. Is what you advise assuming the same difficulty for all tasks, with character skill being the only differentiator?

If so, while I can see the functionality of the solution, it still circumvents the issue. I already designed a few games like that.

Eero,

Still too much case by case assessment for my purposes with this approach, I think. I had something similar in my Exalted hack I ran last summer, where I assigned difficulty based on what type of Exalt the scale of the task seemed appropriate for (i.e. mortal 1, heroic mortal 2, terrestrial 3, celestial 4, solar 5). This worked better than abstract difficulty categories like easy, medium, hard etc., but figuring this out every time was still exhausting in the long run, and despite my rather strong sense of power-level benchmarks, I've been uncertain way too often. Also, temptation to assess lazily or dishonestly.

[Jeff Russell]

Filip,

   Sorry for not being clear, but yes, I was suggesting a constant difficulty with character ability being the only changing factor. Now that I better understand what you're shooting for and why, I see that that was not the right suggestion for what you're going for. I guess I was going for an 'outside the box' solution when you were really looking for the right inside the box solution. If you can meet your design goals here, that'll be pretty darn impressive, I look forward to seeing it.

[SageThe13th]

I know you said you found a solution that will probably work already, but I've working on an idea that may also do what you need.

The general idea is that you approach task resolution the same you would when making an antagonist or a combat encounter.  Which basically means you "stat out" the environment the PCs act in.  This usually means additional prep-work when planning the game session, though there are some ways around this. 

I'm going to go with the idea that the context of a situation is not as important as just generating target numbers to roll for.  The big question for the task at hand is therefore not how hard is it to disarm this bomb?  But, what is a challenge appropriate for the party given the reward?  And, is disarming a bomb an appropriate rationalization for that challenge?  Of course this line of thinking is less effective when it's a PC choosing to do something because he or she thinks it will be easier or because his or her character is good at it.  But, I'll cover dealing with that later.

Note, I'll be giving examples in more or less D&D terms since I don't know what system you actually want to run.  The system works by deciding on the what level or challenge rating various factors are, usually with regard to the party's level.  Each level a factor has provides it with resources that are spent to increase the difficultly of the tasks associated with it.  I can't be to much more specific than that without getting into exact mechanics, so I'll be using an example from here on out.

First pick a universal difficultly to use as a base.  This is just a place to start and it will probably be the same through out the whole campaign.  I'll go with 10.  Next, I start planning my session.  The PCs are all level 7 and have been getting through the regular level 7 content without to much hassle.  However, this session I'm going to have them go for a generic item of great power, that either gives awesome buffs, is plot important, or both.  It really doesn't matter.  So, I decide that this time they get to tackle a level 8 dungeon.  But, before I get to that I start with the town they're starting in.  The town is a level 1 environment, so it isn't a threat, and most of the town's points are put into making the buildings sturdy.  The general idea for the town ends up being that the weather is clear and bright, the town itself is small and easy to navigate, and there aren't a lot of obstacles around, like high walls and such, but you can't easily knock the buildings down.  You could count NPC competences and attitudes as part of the environment or handle them separately, your choice.  Well, the town was easy enough, now for the dungeon.

A level 8 dungeon is obviously a serious threat to the level 7 PCs.  It will probably have a large number of points to spend, that would be generated through some formula which takes into account not only dungeon level, but also how many skills can be applied within it, so that difficulty remains somewhat consistent no matter how many different situations pop up.  My first option for spending points is increasing the universal difficultly within the dungeon.  This represents a constant all pervasive force that hinders the party at all times.  I decide to up it to a 15 giving all actions a decent chance of failure.  To  rationalize this I say that the dungeon is a huge labyrinth inside of a volcano and that the hazardous environment is do to noxious gasses.  So the idea is not in that dungeons inside volcanoes are difficultly to survive in, but that the dungeon has a lot of resources to throw at the PCs.

Next I get into more specifics and buy all the different difficulties for various trials that must be overcome to get out alive with the loot.  I'm going to start with the concept this time and buy based on that.  I go through most the basics making it hard to navigate, giving it locked doors and what not, but decide that this dungeon, like any good dungeon is all about the traps.  And so spend the points to make sure the traps are suitably lethal.  Again it's not how deadly arrow traps are, but how many points can I spend to make my arrow traps deadly.

Well, after that's all said and done it's time to run the game.  Of course, things won't always go the way you expect them too.  Having a rigid preplanned dungeon may not end up being the best thing if it's the only plan you've got.  Which is why challenges also have a number of flex points that can be spent in game to buy new difficulties as they come up.  So, let's say I forgot to buy something obvious like making it so that you can't just beat down the walls.  So, while there are traps and locked doors everywhere they are easily avoided because the party discovers they can break through the walls to get around the traps and tear the doors off the hinges to get through them.  I can then block these actions by buying the difficulty of the strength checks need to do this up mid game.  So why use points to do this?  Because they run out.  This will allow players to "out smart" the dungeon if they continue to try off the wall strategies, though they may run out of usable skill checks first.  Of course if you like the approach they come up with the first time they try something different there's no need to spend the flex points just cause you have them, and instead you could  just default to the universal difficulty rating.  On top of that, maybe you don't know how you feel about a given skill check suddenly coming up.  At this point you could roll a die to determine how many points you spend, giving the approach at least a chance of working, while not having to spend time debating what to do.

Now, what if things really don't go as planned?  Let's say the party thinks that the item I want them to go after doesn't sound that awesome and powerful, and they decide not to leave the town this session.  Well, I could just let the dungeon I planned go to waste.  However, because the dungeon is in essence just set of target numbers and flex points I could easily use it in a different session with different flavor to dress it up in, there's nothing saying it has to be a maze inside a volcano.  Though, in this case I opt for the more extreme option.  If the party won't go to the dungeon I can bring the dungeon to them.  So I make the volcano erupt and light the town on fire.  Now the level 1 town has been transformed into a level 8 death trap.  The 15 universal difficulty is now due to everything being on fire rather than noxious gas, and my traps are replaced with falling debris and lava flows.  I replace the other challenges as they come up applying new rationalizations to fit with the new situations.  I could even go as far as coming up with a new reward if they make it out alive.  Though, if I'm vindictive I could have them go through it all for nothing.

That's the basic idea.  The system would have to be written out in more detail and reworked to fit with the balance of any game you wanted to use it for.  It also may be a tad more complicated than you want.  I don't know.  Tell me what you think.

[Paul T]

Filip,

That's exactly what I was suggesting, yes! It's the only way I can imagine to solve your conundrum, although I would be surprised if there weren't others we haven't thought of.

"SageThe13th"'s model is an interesting one, for a given style of challenge-oriented/Gamist play, but I'm guessing you're looking for something that's challenge-agnostic--something that's more based on the concept of "simulating" what's happening in a particular fictional situation: "What's the most likely outcome here? How difficult would success be?"

Without having thought it out in great deal, I'd imagine something like this:

* Success at a task requires a roll of 3, by default.

The GM looks over the following list of questions, and notes how many "yes" answers there are. For each "yes", increase the required level of success by one:

* Is the character lacking important tools/resources/equipment normally used in this task?
* Must the task be completed as fast as possible/urgently?
* Must the task be accomplished with an unusual high level of skill, finesse, or attention to detail (e.g. "with style")?
* Is the character faced by opposition with a significantly higher level of skill?
* Is the character outnumbered by his/her opposition?
* Is the character attempting something he/she has never done before, and knows little about?
* Must the task be accomplished while paying attention to some other danger, like remaining hidden from guards?

Then the player looks over the following list of questions. For each "yes", decrease the required level of success by one:

* Is my character extensively skilled and/or experienced at this task?
* Has my character made efforts to be prepared for this task, in such a way that puts him/her at an advantage?
* Is my character receiving help or assistance normally completely unnecessary for this task?

Rules:

1. Ignore any question that doesn't seem relevant.
2. If you are uncertain of an answer--i.e. the answer isn't unanimously obvous to the group--the answer is "no".

So, for example:

* Fighting off an enemy: 3.
* Fighting off an enemy, for someone with extensive combat experience: 2 (3 - 1).
* Fighting off an enemy who is wearing armor when you are not: 4 (3 + 1)
* Fighting off several enemies who are better armed and armored than you: 5 (3 + 1 + 1)
* An extremely skilled warrior fighting off two attackers: 3 (3 - 1 + 1)
* An extremely skilled warrior inexperienced with stealthy operations attempts to knock out a guard with a naked sword, without raising the alarm: 5 (3 - 1 + 1 + 1 + 1).
etc.

I think it would be impossible to make a concise list suitable to any game and any genre (for instance, the issue of "character skill" may be irrelevant to some games, while existing in an enormous range in others, such as a superhero genre), but a concise for a specific game could be quite doable, I think. Proper wording may be achieved that makes questions fairly unambiguous (my wording for questions above is off the top of my head, and not very good).

Another approach to consider is what Vincent Baker has done with Apocalypse World:

The "difficulty" of a task is always the same.

* However, the GM is given guidelines for how often a roll must be made: the more challenging the circumstances, the more often the character might roll.
* The outcome of each roll is determined by choosing from a list of options which produce different results in the fiction depending on the circumstances. (For example, one outcome might be something like, "Your opponent gives ground." That has little impact in a duel in an open field, but severe consequences if you're fighting on a narrow ledge.)

Yet another option is to rig something like the Otherkind dice idea with variables for fictional circumstances. I have a brief ruleset outlined for this, inspired by something you wrote on Story Games a while back. Let me know if you're interested; I can describe it, too.