Failing Forward

[pemerton]

This question is related to a valuable precept of GMing and adventure design: Don't stipulate details before you have to.

<snip>

if I pre-write that, to get the pudding, they need the Rod of Pudding Detection (found in the ancient cave lair of the hermit-wizard Jell'O), and the Mace is in the River and they need that to beat the Abominable Sno-Cone Man guarding the dessert on the top of Mount Pudding... that's starting to look like a pre-written story - locations to visit in turn, with McGuffins required to reach the stated goal.

On the other hand Pathfinder's adventure paths seem to do the opposite of this, and they sell extremely well.

I personally don't feel the attraction of APs, but I agree they seem to be very popular.

There is certainly a widespread view about what being a good GM involves, and what scenario/session-prep looks like, which puts a high premium on pre-writing an adventure (complete with fetch quests, McGuffins, BBEGs, etc as Umbran describes). I happen not to share neither that view, nor the approach it advocates, but I think I'm in a minority of GMs.

 

[Manbearcat]

Quick question... so 4e sets the stakes for the consequences of a failed climb check, correct? If so, how does fail forward work here? If I as a player have read the rules and know that damage is the consequence of a failed check, do you as DM still reserve the right to create additional/other consequences... such as the alerting of the guards? If so as DM should you let me as a player know about these added consequences or should any and all "reasonable" consequences be expected... or, and this just occurred to me, is this type of thing best established before the game starts... perhaps in pre-discussion around DM/Player responsibility.

[MENTION=42582]pemerton[/MENTION] already answered this one but I'm just going to add a bit.

4e's noncombat resolution mechanics have three discrete functions in the game:

1) To interface coherently with the combat engine to facilitate precise and balanced adjudication of movement within the square-based paradigm.

2) To facilitate the resolution of player action declarations related to Stunting/Terrain Powers and Countermeasures (within the Trap/Hazard system).

3) To facilitate the abstract conflict resolution system of Skill Challenges.

Pulling this kind of triple duty means that the system and its attendant instruction needs to be robust such that the table can pivot/toggle intuitively and somewhat nimbly. There is disagreement on how well the designers accomplished this. I think there is room for some "fair grumbling" on this in the core books (I would have written them a bit differently or at least I wouldn't have had different people writing different chapters or I would have had a unified editorial influence to confirm the message coheres and is utterly transparent throughout - consider the difference between the 13th Age book with Heinsoo and Tweet. Further consider how easy writing "every moment of play should be about conflict and action" vs "skip the guards and get to the fun" is.). However, once we get into the extremely informative and "on-message" Dungeon articles, DMG2, Dark Sun, Neverwinter Campaign Setting, that "fair grumbling" vanishes fairly quickly.

On "fail forward", 3 is where you will find it in 4e. 4e Skills/deployable resources (as inputs to resolution) are extremely broad (by design intent). PC action declarations are meant to follow suit. In 3 (Skill Challenges), those inputs are meant to broaden further still (like 13th Age Backgrounds) with the outcomes (outputs to resolution) broadening in lockstep. If I'm facing several different varying levels of adversity (visual field issues, dealing with incoming artillery, riding horseback in a high-speed chase) concurrent with a primary PC action declaration that is about navigation (Nature), we're going to "say yes" to the PCs handling those secondary issues (unless the GM demands the players make a preemptive Group Check as part of the challenge...which s/he may very well do) and focus on the question of navigation. If successful, then the PCs get what they want and the fiction moves forward in a way as if they had earned the insurance of a Burning Wheel Instinct (I can't reframe the situation to bring about new adversity based on a navigation gaffe). If they fail, then my job is to change the situation dynamically (forward unless this is the final failure of the challenge). I may hone in on the 1st order input of navigation and create a like 1st order complication/cost/hard choice output for the PCs to deal with/endure; missing the trail "turn-off" and being cornered by Schrodinger's Gorge. Or I may nab a 2nd order complication due to the intensive focus on navigation (yielding an issue with one of the things we're "saying yes" on). Perhaps they find their way but a horse is lamed due to artillery finding home. How do they manage to evade the fast-charging pursuit now?

I've written it many times; Skill Challenges fundamentally do not work without deft GMing of the technique of "fail forward" and interesting/dynamic change to the situation on a micro-success. Every micro-failure must be forward while every micro-success must come with a new avenue of adversity that interposes itself between the PCs and their macro goal. This must continue until hard success or hard failure is ultimately cemented (which the system's framework does). When deft GMing is applied (assuming the players understand their role and the stakes) the abstract conflict resolution system of SCs is robust, coherent, and versatile. When it is not applied, or it is applied clumsily, things don't go well and the table is frustrated and/or bored. When it is not, you get a stagnant, unchanging fiction that yields PC action declarations in-kind ("he doesn't believe/like you"..."well then I Bluff/Diplomance MOAR/HARDER!").

 

[Umbran]

I personally don't feel the attraction of APs, but I agree they seem to be very popular.

I can see the point, for something like D&D.

In something like FATE, I can "wing it" for a combat encounter with ease - how many levels of stress and consequences? A couple thematically appropriate Aspects, and I'm probably good to go. I can throw in a Stunt or two to covver a weird effect the thing can produce, maybe, if I want. It fits on a 3x5 handwritten index card. D&D traditionally has so many tactical fiddly bits (the equivalent D&D critter is an 8" tall column of small typeface print) that this approach doesn't work well. The GM must prepare beforehand. And that preparation is costly. It pays to have someone else do it. And, soon enough, you've got an Adventure Path that provides pre-prepped stuff that will cover *months* of gaming. For the modern adult player, this can be a godsend.

 

[Bluenose]

You didn't answer my question... I didn't ask what pg. 42 or what SC's were (which by RAW use scaling DC's and were very much not presented as just improv tools but as a framework to run extended/complicated social and exploration encounters, irregardless of whether one was improvising or not). I asked if every edition (not just 3.x) of D&D uses scaling DC's (Which also encompasses SC's) which seemed to be what you were stating earlier in the thread. If not then 4e is different in it's approach.

The Rules Cyclopedia, which I have immediately to hand, recommends that DMs should modify the percentages on the thief special ability table for actions which are harder than usually (though not, interestingly, for easier ones) and also apply positive or negative modifiers for use of skills (the BECM version of Non-weapon proficiencies). There is a similar rule with regard to NWPs (the DM can apply modifiers) in the Dungeoneer's Survival Guide. The Holmes Basic set rules do not mention modifiers for thief abilities, and I don't have a 1e PHB readily available to check whether it does (and that might anyway need a DMG). But while it's not conclusive, there's clear evidence that a form of "scaling DC" with modifiers for tasks that would be easier or harder appears quite early in D&D.

 

[Ilbranteloth]

This question is related to a valuable precept of GMing and adventure design: Don't stipulate details before you have to.

For those who are so worried about the dreaded "DM's Pre-Written Story" one utility of this is obvious - the players cannot be following a predetermined story if we don't actually pre-write the story! On the other hand, if I pre-write that, to get the pudding, they need the Rod of Pudding Detection (found in the ancient cave lair of the hermit-wizard Jell'O), and the Mace is in the River and they need that to beat the Abominable Sno-Cone Man guarding the dessert on the top of Mount Pudding... that's starting to look like a pre-written story - locations to visit in turn, with McGuffins required to reach the stated goal.

You've already stipulated that to get to the pudding they need the rod and the mace. Whether you place them in a specific place or randomly is irrelevant, you've already pre-written the story, just with fewer details. They are still going to visit locations in turn, and, based on the concept that the mace may or may not be where they think it is, they may have to go to another location.

The only thing you have accomplished is that the DM didn't know where they were going to find them either. Is that a benefit?

I like the idea of something related to the skill or activity being tied to the check. I'm even OK with the rod falling out on a particularly poor check. You drop the rod instead of fall and die, for example. It could even be, 'you slip and slide down the cliff, with the strap of your pack catching on a protruding rock, suddenly halting your fall. You take 2d6 damage, and the sudden jolt dislodged the rod which bounces off the cliff with a loud clang before disappearing into the dark chasm.'

But whether the mace is there or not is not related to the skill in play at all. It might seem like it's related, but it's not. Either the mace is there, in which case the party finds it or not, or the mace is not there, in which case they don't find it.

On the distinction about "fail forward" meaning different things in different playing (or, I think more to the point, GMing) styles, I think you have a good point. "Fail forward" may very well be a cluster of techniques, some of which are useful for any given style. On the other hand, I am pretty sceptical about any "purist" approach to pre-authoring. I think most likely every GM has some things s/he authors on the fly - NPC intentions regarding the characters and dispositions being particularly common ones; details of the "small furnishings" in a room being another. I think it's more a matter of degrees than pure approaches.

I think a useful question as regards pre-authoring might be "what things are most usefully pre-authored, and what things are better authored in response to game situations as they develop?", rather than "what type of authoring is best*, pre-authoring or authoring-in-the-moment?"

(*: or even "do you prefer").

Agreed. In a good scenario I think there is a mix of both. In a completely random design, it would be extremely difficult to have a consistent believable world. The framework needs to be there to some degree or another.

Of course, in the fiction, the mace is either there or is not. The relevant question is when that authorship decision is taken.

In games that deploy "fail forward" in a systematic fashion there is no such thing as a "sideplot". The notion of a "sideplot" only has meaning in the context of there being a principal, non-"side" plot - one that (presumably) has been pre-authored by the GM.

In the sorts of games that deploy "fail forward" in a systematic fashion, "the plot" is whatever results from adjudicating the players' action declarations for their PCs. If a PC has finding the mace as a goal, and the player declares that the PC searches for it, then the location of the mace has become a central part of "the plot". If the search check succeeds, "the plot" is what the player (and PC) wanted - namely, the PC finds the mace. If the search check fails, then the GM is at liberty to introduce a complicating or vexing "plot" (ie "fail forward) - such as, in this case, the discovery of black arrows in the ruins of the (formerly) private workshop of the (now) balrog-possessed brother.

Something which I don't think anyone has raised yet is this: if, as a GM, I had no interesting idea about where else the mace might be other than in the ruined tower, it would have been just as reasonable for me to decide that, on the failed Scavenging check, the PCs find the mace and the black arrows. (This would be similar to the example that I mentioned some way ago upthread of the feather the peddler was offering for sale truly being an angel feather, as claimed, but having a curse on it.)

But I didn't do that because I did have an interesting idea about where else the mace might be, namely, in the hands of the dark elf.

This is a modest illustration of the general principle that - in "fail forward" play - consequences should be narrated in a way that maintains narrative and dramatic momentum.

Of course there are side plots. Or at least there should be. Unless you are implying that the result of every check always moves them on a shorter or longer path towards their goal. In which case it's a railroad.

Although this is a potential pitfall of this technique as it's being described as a tool to help write the plot. They should run across things that have nothing to do with the current goal. And if they choose to follow that path for a short period of time and come back, it's a side-plot. If they choose to follow that path and not come back, it's the new plot.

Ilbranteloth

 

[Imaro]

The Rules Cyclopedia, which I have immediately to hand, recommends that DMs should modify the percentages on the thief special ability table for actions which are harder than usually (though not, interestingly, for easier ones) and also apply positive or negative modifiers for use of skills (the BECM version of Non-weapon proficiencies). There is a similar rule with regard to NWPs (the DM can apply modifiers) in the Dungeoneer's Survival Guide. The Holmes Basic set rules do not mention modifiers for thief abilities, and I don't have a 1e PHB readily available to check whether it does (and that might anyway need a DMG). But while it's not conclusive, there's clear evidence that a form of "scaling DC" with modifiers for tasks that would be easier or harder appears quite early in D&D.

No what you've shown is that D&D has recommended/provided modification of task difficulty based on the in game fiction... not that D&D has always used a certain range of DC's, modifiers, etc. for a particular level or range of levels and the fiction then created to fit said range. They aren't the same thing and it's why I don't think @Umbran's claim about details not being important is valid. It's like claiming D&D has always had 4e's powers... because in combat you always rolled a d20 and made choices... again it's not the same thing and details, process, etc. are all important.

 

[Balesir]

Where I diverge from fail forward (and this may be just because of my love of math and probability and how it interacts, and how I view it in the game, etc.) is that I dislike the tying of abstract thing to characters abilities.

In the example of the mace here is the way I see it playing out.
The DM does not know if the mace in the tower. With the fail forward example they will find it if it is there, and if it isn't there they will find the alternative path. So the difference between the 2 is not "did they search good enough" but is the thing they are looking for there.

My preference is (if you need to decide and the DM can't) just roll a die not tied to a skill (50/50).

One response to this was "we are tying it to "failure" not to the skill" but mathematically that is utter rubbish. You might mean that you don't care if it's tied to the skill, but it is easy to show that it is inversely proportional to your skill.
Skill will succeed on (p) so chance of failure is (1-p) so chance of mace not being there is (1-p)
As p goes up the chance of the mace being there goes down.

Amusingly (to me, at any rate), close acquaintance with probability and stochastic outcomes is a major reason why I see things quite differently in this respect. I'll see if I can illustrate why:

Consider a situation like the "finding the mace" case you cite. Imagine that we have a system akin to that you suggest, with one die roll - on a d10 modified by character skill - determining the thoroughness of the search carried out, and another roll - also on a d10 but unmodified - determining whether or not the mace is present to be found.

Now, consider further that we could devise quite easily a system that is exactly mathematically equivalent to the system above, using a single percentile roll.

In this percentile system, the character skill has an influence on the outcome, but - considering where the original system to which our percentile system is exactly equivalent - the chance of the mace being present to be found is clearly not connected to the character's skill level.

In most cases in D&D, skill level is not so overwhelmingly important that it is determinative of success or failure. It has an influence on the outcome, but does not (usually) make it a foregone conclusion. In this circumstance, I see nothing whatever wrong with viewing skill rolls as being analogous to the above "percentile roll". In other words, it judges ("resolves") success or failure at reacing a desired end-point based on a constellation of potential failure modes or reasons. In fact, given the general ways in which feats of skill work in real life, I see this view of skill rolls as far more plausible from a "verisimilitude" point of view than the "you either bungled or you didn't" perspective.

As soon as a DM says to me "If you pass the roll you will find the mace, but if you fail the mace is not here and I will give you a clue to it's location" the "Schrodingerness" of the situation is staring me straight in the face.

I think you are vastly overstating the sharpness of the divide, here. Not every failure will be taken to indicate that the mace is not there to be found; all it means is that the characters' best shot at finding it has failed to uncover it. Given how destructively thorough most players can be in their imaginations when "pixel bitching" a room, I will grant that the chances that the mace is there and still undiscovered by a balls-out search is slim, but strictly it's just "unknown".

As an aside, this is one of the things I'm liking a great deal about Dungeon World, so far (still reading and digesting - not run it, yet). Part of the GM's Agenda given in DW is "Play to find out what happens". This discussion has made me realise some of the dimensions of this; it is really talking about the "author only what you have to" that [MENTION=177]Umbran[/MENTION] mentioned upthread. If the PCs fail to find the mace (that might or might not have been there), what do you as GM take as "known"? Simply that they failed to find the mace - no more than that. It might now plausibly turn up somewhere else, or it might not (whereas, had the PCs found it, it obviously could not turn up elsewhere unless they either took it there or lost it again). Each resolution in play sets the parameters within which future resolutions happen, and as a GM you can be as surprised by what happens as the players. Looking back, there have been elements of this in my GMing previously, but I had not consciously singled them out as something "fun" I wanted to increase and develop. That may be changing.

 

[innerdude]

I personally don't feel the attraction of APs, but I agree they seem to be very popular.

There is certainly a widespread view about what being a good GM involves, and what scenario/session-prep looks like, which puts a high premium on pre-writing an adventure (complete with fetch quests, McGuffins, BBEGs, etc as Umbran describes). I happen not to share neither that view, nor the approach it advocates, but I think I'm in a minority of GMs.

It's not that there's any preconceived notion about what session prep should like.....it's that D&D gives you very little leeway to prep any other way because it's so stinkin' hard to prep encounters. Either you've got to pre-plan those encounters, or someone else has to, but for the most part "winging it" all the time probably isn't going to cut it.

When the bulk of your prep time is spent deciding what can my enemies do in combat (i.e., building their stat sheet), there's far less time to care about what the enemies are doing in the game world, and build natural/realistic/coherent/sensible intentions and interplays between the NPCs and the game world.

Typical prep for my Savage Worlds games now is 20 minutes, 30 minutes tops.

"What happened last time?" (Decide how the NPCs and world react and frame the scenes forward accordingly)
"Is there anyone or any place they're likely to interact with? (Build a short, 2-3 paragraph sketch of what those things are).

That's it. If I'm absolutely certain that they're going to have a combat with an NPC, I may take 5 minutes and sketch out their basic stats. But in Savage Worlds, that's literally all it takes---5 minutes.

If there's an easier system to prep encounters for on the fly than Savage Worlds, I've yet to see it.

The end result of all of this is that you can keep almost everything else fluid. There's no need to railroad, or plan out 15 sessions in advance. Set up the current scene, adjudicate what happens, do some quick calculating what the "spin off" effects are, and send the party forward to the next scene, and just see what happens. Your "grand plot" is little more than a vague set of basic ideas, tied to NPC motivations.

This is the greatest thing in the world, because you can apply "just in time" pre-authoring to nearly any situation, and to the players it feels like a natural outreach of what they've already done. And in my experience, this keeps everyone at the table highly engaged.

I can see the point, for something like D&D.

In something like FATE, I can "wing it" for a combat encounter with ease - how many levels of stress and consequences? A couple thematically appropriate Aspects, and I'm probably good to go. I can throw in a Stunt or two to covver a weird effect the thing can produce, maybe, if I want. It fits on a 3x5 handwritten index card. D&D traditionally has so many tactical fiddly bits (the equivalent D&D critter is an 8" tall column of small typeface print) that this approach doesn't work well. The GM must prepare beforehand. And that preparation is costly. It pays to have someone else do it. And, soon enough, you've got an Adventure Path that provides pre-prepped stuff that will cover *months* of gaming. For the modern adult player, this can be a godsend.

Pretty much this.

Of course there are side plots. Or at least there should be. Unless you are implying that the result of every check always moves them on a shorter or longer path towards their goal. In which case it's a railroad.

Although this is a potential pitfall of this technique as it's being described as a tool to help write the plot. They should run across things that have nothing to do with the current goal. And if they choose to follow that path for a short period of time and come back, it's a side-plot. If they choose to follow that path and not come back, it's the new plot.

To me this line of thinking is very much tied to the entire notion of "pre-authoring" to begin with. "Where's the plot? How are the PCs supposed to follow it? And if they get off of it, are they supposed to get back on it or not?"

Scene framing approaches don't really follow this line of thinking at all. Scene framing approaches say, "You're this character, framed in the fiction this way, with these goals/responsibilities/obligations. Here's what you understand is going on around you. Where do you want to go, and what do you want to accomplish when you get there?" Lather, rinse, repeat.

When I'm running a campaign, I literally have NO IDEA what's going to happen, or how it will end. NONE. For all I know the BBEG's might actually win . . . or maybe the PCs decide after killing the BBEG that they kind of liked the cut of his jib, and decide to finish what he started. Who knows?

 

[innerdude]

In response to, "I'm still a bit baffled why Fantasy Craft didn't become more popular among the 'I don't really like 3.x, but don't want to move to 4e' crowd."

Too much crunch was a big part of it I think - and too little marketing.

See, this is what's weird to me. A majority of 3.x adherents migrated to Pathfinder. I've played Pathfinder. I've GM'd Pathfinder. And Pathfinder isn't any less crunchy than Fantasy Craft.

In fact, from a player/GM perspective, Fantasy Craft is better, because it changes the ratios of the crunch. The bulk of the "additional" crunch is player-side, while the crunch from the GM's side is radically reduced due to FC's encounter template design.

The marketing part, on the other hand, makes sense. That, and I think Crafty Games missed the peak window to capture the broadest audience. If FC had been on the market in late 2008 / early 2009, before the "full drop" of Pathfinder Core Rulebook, it could have made a bigger impression. As it was, by the time FC made it on to shelves in early 2010, the Pathfinder core rules had already been around 4-6 months and stolen the thunder.

But ---- even then, it seems in retrospect people weren't ready to jump away from 3.x. And if the goal was to maintain 3.x compatibility, FC certainly didn't go that direction. In my opinion Fantasy Craft is an objectively better game than anything in the 3.x line, Pathfinder included, but it seems like most people actually WANTED 3.x.

 

[Manbearcat]

As an aside, this is one of the things I'm liking a great deal about Dungeon World, so far (still reading and digesting - not run it, yet). Part of the GM's Agenda given in DW is "Play to find out what happens". This discussion has made me realise some of the dimensions of this; it is really talking about the "author only what you have to" that @Umbran mentioned upthread. If the PCs fail to find the mace (that might or might not have been there), what do you as GM take as "known"? Simply that they failed to find the mace - no more than that. It might now plausibly turn up somewhere else, or it might not (whereas, had the PCs found it, it obviously could not turn up elsewhere unless they either took it there or lost it again). Each resolution in play sets the parameters within which future resolutions happen, and as a GM you can be as surprised by what happens as the players. Looking back, there have been elements of this in my GMing previously, but I had not consciously singled them out as something "fun" I wanted to increase and develop. That may be changing.

I'll aside to your aside!

I've been GMing like this for literally as long as I can remember. I've never run APs or modules. The very first thing I knew about my preferences as a GM was "I want to be in on the wild-eyed wonder (as much as possible) of what emerges from play, same as my players!" Then I thought "how do I accomplish this?" Prep what is only utterly necessary for a session (shockingly little to many GMs who have played in my games in real life). Develop a few guiding principles, techniques, and mental-overhead-management strategies which center around how to best evolve post-resolution fiction (to maintain continuity, genre constrains, pacing, and dynamism). Develop the in-situ clerical side (high utility short-hand and make flash-cards my friend). Mature to mastering adlib/improv skills.

When I first read Dogs in the Vineyard, my mind said "holy crap!...this is it!" Then I read Apocalypse World and finally Dungeon World. Then my mind said "holy crap!...this is it-ER!"

EDIT - for the below

It's not that there's any preconceived notion about what session prep should like.....it's that D&D gives you very little leeway to prep any other way because it's so stinkin' hard to prep encounters. Either you've got to pre-plan those encounters, or someone else has to, but for the most part "winging it" all the time probably isn't going to cut it.

Xp for a good post except wanted to comment on this bit.

Winging it is entirely possible. You just have to (a) master the ruleset, (b) be really good at improv/math/tactics, (c) and the ruleset needs to be replete with (1) precise challenge math (monsters/traps/hazards and overall encounter budgeting), (2) easy and robust monster/trap/hazard/terrain creation, (3) all things interfacing nicely to create dynamism within the combat engine and a foreseeable (GM-side) narrative.

I have probably pre-built (only when I was assured they would take place) perhaps 20 (ish?) encounters in 4e since the end of 2008 (in probably 60 levels of play).