Difficulty levels

[taalyn]

How do you folks determine difficulties? How do you decide what's a reasonable difficulty for any given task? Do you use probabilities at all, i.e. figure out what the average score would succed at 50% of the time, or something similar? Once you have a basic difficulty, how do you decide what's a good progression?

Aidan

[Mark Johnson]

Does the average major league baseball player hit 50% of the pitches from the average major league pitcher?

[taalyn]

I'm not a sports buff by any means, but if I understand it correctly, 33% is considered a good average in major leagues. On the other hand, a professional climber (almost) never fails at climbing a cliff. So how do you represent this in game terms?

Aidan

[bladamson]

imo the average difficulty for a task will depend on the task.  For example, I will succed 80% of the time trying toss something into a wastebasket (yup, I'm damn clumsy).  otoh, I'll succeed about 20% of the time trying to cook pancakes.  Both could be considered "very easy" tasks, yet the relative difficulty is vastly different.

I think one doesn't need to specify a general "average difficulty"; a good GM will scale all difficulties to the players anyway.

It's a more difficult problem than it seems.  Maybe better to design your mechanic so as to need not precisely know an "average difficulty".  It's also nice to be able to readily see "how well" the action has succeeded or "how badly" it has failed from the roll.  How tasty is the pancake?  How far away from the wastebasket is the latest draft of the end-all be-all mechanic?

But if I know what I was talking about,I'd be a millionaire, so you may not want to listen to me... :)

[Mark Johnson]

Is it necessary to represent difficulty and skill at all?  Is your game about a realistic assesment of difficulties vs skills?  If so there are two ways to do this: complexity and "winging it."  If it is not about this...  then maybe we need to be having a different discussion.

[Lance D. Allen]

It might be easiest to simply determine what constitutes a difficult task, and set that as an upper limit. If the maximum rating of a stat/skill is 8 (for instance) then that seems a decent level, at least for play-testing.

If I am that beat-cop with the legendary firearms skill, Cyan, 8 motes and I draw 8 motes to attempt a shooting task of nearly impossible difficulty (let's say a spriggan is hanging my compadre, and I'm gonna try the robin hood trick of shooting the rope... from half a block away) so my difficulty is 8. I draw my 8 motes and get: CMMBRCAR (randomly determined) That's 4 successes from my two cyans, now we get to see if I can make up 4 more from my remaining motes. I convince the GM that my two reds go into running pell-mell toward the target, shortening the range before I stop to shoot, and my blue is used as I focus my mind and clear all distractions so I can take a precise shot. Well, unfortunately that's only 7 of the 8 successes I need, so I blow a point of luck to draw another mote, which happens to be green.. not what I need. Well, unfortunately even legends don't always win... looks like my friend's gonna have to find his own way out of that noose, or hang in there until I can try another shot.

So a difficulty 8 task still managed to prove impossible for a skill with a hand of 8 as well. If I'd managed to pull one more cyan, I'd have had it, but that's the luck of the draw.

[taalyn]

Lance, the example is great. It's so cool to see my mechanic being used by other people!


Anyway, everyone, here's the issue I have. Is 8 hard enough to merit the (general) upper limit? I understand what people are saying about every case being unqiue, but I still need to provide some sort of baseline of what is a standard difficulty and where hard or really hard or really really hard lie. I need to have some idea of a range for standards - otherwise, how will I know what to consider a difficult task? "Ok, flying off the bridge, doing a triple flip with a 1/2 twist, and landing on one toe inside the tulip patch - that's difficult, um...42? 8? 637? 5?" How do I know what DL is worthy of the example? Telling me that it depends is so not helping.

Aidan

[RobMuadib]

Anyway, everyone, here's the issue I have. Is 8 hard enough to merit the (general) upper limit? I understand what people are saying about every case being unqiue, but I still need to provide some sort of baseline of what is a standard difficulty and where hard or really hard or really really hard lie. I need to have some idea of a range for standards - otherwise, how will I know what to consider a difficult task? "Ok, flying off the bridge, doing a triple flip with a 1/2 twist, and landing on one toe inside the tulip patch - that's difficult, um...42? 8? 637? 5?" How do I know what DL is worthy of the example? Telling me that it depends is so not helping.

Aidan

In general, I see 3 important considerations in setting difficulty levels (which are closely tied to character ability levels as well.)

First, you need to determine when you intend players to check for actions. Is it going to be a base-line "sim" style emulation of character abilities. WHich typically breaks down into tasks being defined as automatic narration, trivial tasks such as walking or talking, while unhindered and such. Narrative Success, if you have a certain level of ability, you can peform the task as automatic narration, or your typical "chancey act', meaning the act is unsure and must be determined randomly. This is very sim application of the rules to define action within the game world.

At the other end, you can have players checking for "Dramatic Resolution" purposes, where they generally can narrate results, but use the check for guidance or detail when they feel like it. In the former case, you want to set your DL level low to allow proficient characters to succeed regularly enough for them to get things done. If the latter, it is ok to have slightly higher DL's, especially if you intend players to use luck. Which is also a consideration, how often to you expect players to use luck/boons to ensure their success? If players are expected to use them often, then it's ok to push the DL's up a level or so as well, so the luck will be needed.

Next, if you know what players are going to be checking for, you then need to consider the average character abilities. This is important also in relation to the character's 'power" relative to game environment. If the character's are all best of the best type heroes, then they should have scores that are going to be higher than the average difficulty.  Also, in defining abilities, characters should have higher than average abilities so they can succeed more often than average, as befits their conception.


Finally, with your token type system, it is similar to a dice pool in that it is success biased, in that the chance of getting a single success rises quickly as you add to your pool/hand. So, for average difficulty tasks, you shouldn't require too many successess.

I would reccommend you consider the "world" average ability character, with a moderate amount of training as your default ability, you will probably want to set your Difficulty Level at about half that Score.

At the extreme end, the highest difficulty should probably be equal to the maximum ability possible, max stat plus skill or whatever. In pool type systems, it is nicer to have extra space at the high end, since minimal success is usually trivial.

anyway, hope that helps somewhat, rather than being obscure and confusing.

best

[taalyn]

Rob,

 This helped immensely! Because of the way in which caerns work, at a hand half-caern size, the maximum # of successes are pretty much guaranteed - over that, and on average you're just adding non-relevant motes. So, that's approximately a hand of 15 or so - a working maximum DL.

 The average human has a 3 hand for any given stat. With the 2 DL I've been using, a normal guy will succeed approx. 50% of the time. PCs average 5, and with stacking will succeed on a DL2 a larger proportion of the time.

  So a scale might go like this:

  1 - fairly easy, why bother
  2 - average (1 correct mote or 2 narrative, stacking motes)
  4 - difficult
  6 - very difficult
  8 - whoa mama, that's hard!
  10 -  you'd better be a master
  12 -  masters can do it sometimes
  14 - Herculean, legendary
  16 - divine, epic, unbelievable

  chacks break down along both sim and dramatic lines - that is (since I tend to get the meanings wrong) that  chancey acts are checked only if it would really add to the drama of the game. Or, at least, that's how I play it - I think the game is actually more sim than that. Weird.

  Luck/Boons aren't extremely common for task resolution (Pcs usually have lots of other things to use their Luck for, since it powers magic too).

   Thanks again!!!

   Aidan

[Thomas Tamblyn]

For 'average difficulty' it should depdn on the game - and be hard-wired into it.

Unknown armies sensibly sets skill levels low so that an average check is average for a game where you are insane/running for your life/fighting for your life/attempting to ascend to godhood/cutting yourself for magical energy/whatever mad stuff you want.  That is - fairly low chances.

D&D on the other hand tells you right off the bat what an average check is - Basic attack bonus versus armour class.  The most important roll in the game has its difficulties right there and balanced for you.

Sorcerer defines its difficulties relative to the character's abilities which is a lot easier to judgement call than "4 is difficult, 5 is really difficult, 6 is a bit more difficult, 6 is really difficuolt, but not as difficult as 7 which is less difficult than 8 but still pretty damn tough..." which so many games resort to.

Bascially I think that games should a) have a default dificulty and b) make it an appropriate one.

Things to consider - Low probabilities of success make the game more random and can sometimes make players feel innefectual.  High probabilities take away uncertainty and make skill level a little less
relevant.  Finally speed, low proabbility means many rolls in action scenes to get stuff done, high probability means fewer (unless, like d&d, success doesn't necessarily translate into a large effect).

Hope thats helpful.

[taalyn]

Thomas,

 Yeah, that's pretty helpful too. I recognize that players are going to have different priorities: some people want to play godlike beings, where I'm content to be a bookstore clerk dealing with the fact that my landlady is one of those godlike beings. When I make up my little chart of DLs, I'll make sure to note what is baseling for different styles of play.

 I am really curious, how does Sorcerer define difficulties relative to Character abilities? This sounds like something I need to ste^H^H^H think about, as it lines up with my ideas and understandings of how characters should work. I once has a VERY early release of Sorcerer (6 years ago?), but I know it's changed since then...

  Aidan

[Ron Edwards]

Hello Aidan,

Well, your question about Sorcerer opens up a whole issue that I've been thinking about how to address from the first post of this thread.

They're based on this idea: "difficulty" does not apply to characters or to in-game actions at all. It never does. It never can.

Instead, it refers to the real people and their actions: dice, dialogue, and all the agreements that go into someone saying "what happens." Whether you call it Difficulty, Target Number, Saving Roll, or whatever, that's what it is.

If you want this "real difficulty" to reflect or represent an imaginary difficulty in the game-world, that's OK. That's one way to role-play. As a first priority, it's called Simulationist play.

However, it's also perfectly possible to throw out the whole idea of representing "how hard" doing something is. Think instead about the whole point of play at all, staying with the real people at the real table - is it to struggle with unpredictable outcomes, having to re-adjust one's position a lot? Is it to slowly build up more and more motivation, ultimately to deliver it in a climactic punch? Is it to experience a transformation?

One concept we throw around here a lot is Task Resolution (which I think is pretty much where your mind's at, at this point) and Conflict Resolution. A lot of games focus on the latter alone. Sorcerer play is about resolving conflicts, and so is Hero Wars play, Castle Falkenstein play, Orkworld play, Prince Valiant play, and a few others. Tasks, per se, are only interesting as a detail of conflict.

A lot of people nod and go, "Yeah, yeah" at this point without really understanding at all. Let me give you an example from Hero Wars.

Say my character has a Sneak ability at 18 (out of 20). Say the place he's sneaking into has a little lap-dog sleeping there, who has a Yap of 5w (which in Hero Wars means 25 out of 20).

If you go by task resolution as found in most games (and as found in most role-players' heads), I would roll all kinds of things: a roll to climb in, a roll to orient myself in the dark, a roll to sneak, and so on. The little dog would roll some kind of perception in order to wake up, and then maybe roll its Yip-yap or whatever ...

See? The conflict exists as a built-up, multi-step construction of tasks.

But now let's do it Hero Wars-style. I roll my Sneak. The dog rolls its Yap. Someone wins.

Here's the point: if the dog loses, that does not mean that it tries to Yap and fails. It means that my Sneak utterly outweighed all aspects of the dog's potential to Yap ... an appropriate narration would be, "The dog doesn't wake up."

Whoa, says the role-player. How can the dog roll its Yap if it's not yapping? I don't get it! This sucks!

Same goes for Sorcerer. How can you get bonus dice for announcing an intended action, if the system permits that action to be aborted? If in-game cause is the golden metric, this makes no sense at all and drives role-players absolutely crazy. These are the guys who are always whining about "initiative" in Sorcerer, which is determined retroactively rather than prior to announcing actions. However, since in-game cause is practically just Color in Sorcerer play, rather than its motor, we focus instead on conflict resolution.

Most conflicts can be brought to a matter of character vs. character. In the Hero Wars example, never mind how hard the wall was to climb, or how creaky the floor is, as separate tasks. Those things exist only to modify the Sneak vs. Yap conflict, so they're just penalties. It's all about the character vs. the friggin' little dog.

Same goes for Sorcerer. Trying to nail down the cellar door with special nails, so the Thing doesn't claw its way out? The conflict is You vs. The Thing. The door's rottenness or the character's familiarity with carpentry are modifiers, not tasks in and of themselves. Roll your Stamina (with or without modifiers, depending) vs. the Thing's.

Same goes for all kinds of situations. Almost every roll, even if the character is all alone, constitutes a character-on-character conflict. If it's not, just consider the object or situation in question to be a character in its own right. Yeah, that's right. The door in the character's way is a character, trying to stop him. Think of all conflicts like this and you're all set.

What's the conflict? How drastic is it? What degree of in-game plausibility do we want to preserve (which is an issue, I'm not saying it's totally gone)?

That's all there is to it.

Best,
Ron

[taalyn]

So, my DL list would be better described as relative "ranks" for characters, then. THis seems more Narrativist than Sim in some ways, which is great, since I'm aiming for a nar tendency behind/in addition to the simulation. On the other hand, the Narrativist approaches (as in Theatrix, Story Engine, UnderWorld, and others) as what consistently attract me to a game. Also, since skills mostly boild down to 2 main categories (inspired by Alternacy and Mnemonic), Nature, describing a cultural background and all the skills that go with it, and Profession, the same for the PCs job, it seems silly to get anal in Simiulationist details.

If I'm understanding you right, the difference is that Task resolution focuses on the parts of the action chain, whereas Conflict resolution focuses on the entire event at once.  

Here's the question then - if resolution always resorts to characterifying an object, or relating to an NPC somehow, with modifiers to the whole event based on situational issues traditionally approached as unique tasks in themselves, there can never be unopposed rolls, correct? Difficulty in this sense is always relative to a character, whether NPC or NPC-ified door.

If that's the case, how are tasks like the following handled? These would be traditionally assigned DL-type tasks/conflicts. Here is my babbling to myself on them:

- see if the character can run for an hour (but I can see this within a greater context where the PC needs to outrun the bad guys, thereby introducing the needed opponent. In any case I can think of, it won't matter if he's just running because, as who cares? How will it impact game play?)

 Actually, I think that was the only counterexample I can think of, and I've already addressed it!

  What about combat? This is one of the things I really like about UnderWorld - the two sides throw their HeadCount, and then use contextual differences (greater numbers, better weapons, poor light, etc) to flip heads to tails. It can also be a bit more gritty/Sim as well, using the same sort of action to detail individual "rounds" of the fight.

  In terms of Aisling, I can easily assign a value to the Door that Jonas needs to get through, call it "Doorness", and range that on a similar scale as characters. An average Doorness is 3, excellent is 5, and so on, just as for Characters. What about magic?

  Basically, a PC will draw motes, explaining how results are melded together ("this red and green become amber, which I need for XYS reasons") or woven ("and this red for damage, and that red for fire, and the blue for distance..."). The spell is a weaving of various motes/threads together and then 'powered up' with a blast of Luck. After all the weaving, and the Luck expenditure, the player draws on the PCs technique, and the Difficulty is equivalent to ? The number of motes in the spell? What about melded motes - diff is the number of motes drawn? And since motes are kept until the spell is cast, the PCs caern is low, totally messing with probabilities. All of these bits conspire to make me think this doesn't work for magic, but how else to model a conflict with reality (i.e. a spell)?

 As for in-game plausibility, I think my mechanic preserves that even with  conflict vs. task resolution. Since the motes are explained as colors of energy that 'power' certain kinds of actions (an athletic person draws red a lot because red motes/energy allow athletic actions to be taken), conflict res is discussing, ultimately, the source of opposing energy. The Doorness of the door, or the team cohesion of the bad guys.

  Thoughts?

  Aidan

[Ron Edwards]

Hi Aidan,

You wrote,

If I'm understanding you right, the difference is that Task resolution focuses on the parts of the action chain, whereas Conflict resolution focuses on the entire event at once.

Not quite. Don't confuse Task vs. Conflict with Action vs. Scene. Those aren't really the same thing.

Action vs. Scene would be running a fight hit-by-hit (most games), vs. running it as one whole event (Tunnels & Trolls, Story Engine). Please note that this whole spectrum can be handled as Task Resolution; the only issue is scale.

Task vs. Conflict is far more about what is being opposed, and answering that question based on real-people priorities as opposed to in-game physical causes. From this point of view, a "fence" can represent an owner's desire to keep you out, in the sense of a long-standing rivalry between the two characters. Think of the way the two characters in the movie Nobody's Fool keep stealing a lawn-mower back and forth - their efforts to (e.g.) climb fences or drug dogs are only manifestations of that rivalry.

Your paraphrase:

resolution always resorts to characterifying an object, or relating to an NPC somehow, with modifiers to the whole event based on situational issues traditionally approached as unique tasks in themselves, there can never be unopposed rolls, correct? Difficulty in this sense is always relative to a character, whether NPC or NPC-ified door.

... is dead-on target. As you say, I think you already addressed your proposed counter-example.

[Caveat: you can also achieve the very same effect by having no rolls for NPCs/objects and work only from target numbers, always, as in Legendary Lives or The Whispering Vault. Either way.]

Finally, you have also happened upon the wonderful outcome: that in-game plausibility is never challenged by Conflict resolution. Everyone who's used to, and emotionally committed to, Task resolution thinks that in-game plausibilty will somehow crumble and fall apart unless it's given first-priority ... but it doesn't. In order to validate the conflict, the in-game reality must be taken into account and validated as well, secondarily, and people are very good at doing this with little effort or disagreement.

Best,
Ron

[Mike Holmes]

A slight nuance correction:  Task Resolution deals with the results of Tasks, and Conflict Resolution deals with the outcome of Conflicts. The length of the chain is irrellevant though one could generalize that most conflcits would cover more than one task. However, if a Conflict were, say, trying to convince someone to come with you on a trip, then that might only represent one Task in length.

The real point is that, with Conflict Resolution, you don't much care to define the steps involved before hand. You only care about the possible outcomes of the conflict. Thus, a fight is a Conflict that has many, many possible outcomes, all of which may take seconds or days to achieve. This relates to the Fortune in the Middle (FitM) concept. Basically that once you've determined the winner of the Conflict, you are relatively free to describe whatever tasks were accomplished to come out on top.

Thus a combat win could be described as several exchanges of blows that result in you wounding the opponent several times, wearing him down until a final swordstroke puts him away for good. Or you can describe it as a single sudden well-timed strike with the flat of the blade to the side of the head that renders the foe unconscious.

Does that illustrate better? The Chasm example is often used. In Task resolution, the task is to leap the chasm. The character is described running up, and leaping, and then the roll is made, and success means that he gets to the other side. Failure means that he jumped, and came up short in sme measure. In Conflict Resolution, the Conflict is getting to the other side of the chasm. If you choose to use jumping skill, you roll, first. Success can be described as a successful leap. But failure can be described as running up and then suddenly stopping before making the jump. Or not seeing someone hiding in the shadows who steps out and blocks your path. Or any of many things.

See the difference now?

BTW, this is not the same as Scene Resolution wherin you determine the results of a whole set of potential Conflicts and/or Tasks with a single roll, thus creting a theme for the Scene.

Your statments about opposed rolls gets you Mike's Standard Rant #4: the Myth of Opposed Rolls.

Basically, use whatever makes the most sense for your game. None is better than any other, but one should consider the ramifications closely.

Mike

Edit to note cross-post with Ron (second today :-) ).