Failing Forward

[Lanefan]

If each check raises a serious prospect of failure, and each failure raises a serious prospect of changing goals, I don't see that dramatic arcs are going to arise. Dramatic arcs, especially in adventure fiction, tend to be generated by failures that are incurred while resolutely sticking to a goal. For my own RPGing, major dramatic arcs tend to unfold over many sessions - perhaps a year or two of play. Having the goal of attaining the pudding on Mt Pudding be abandoned because of a single failed Climb check doesn't strike me as very conducive to the sort of play I'm interested in.

But if that's what the players / characters decide to do, well...so be it. That said, most players / characters are a bit more persistent than that; and if they can't climb Mt Pudding they'll haul out the axes and chop it down!

The goal is to climb the mountain and get the pudding. Losing the divining rod is an impediment to this goal, just as falling down the ravine would be. It is a failure.

Getting the pudding is the goal, yes. Climbing the mountain is but one step towards said goal, but a significant enough step to call for its own check independent of any check required to actually find the pudding once at the top.

If you focus only on task and not intent, then you may not see it as a failure. But in "fail forward" games, intent is as important as task. I've already mentioned the BW rules and advice on this several times upthread.

The mere failure of the task (climb the mountain) leaves the overall success-failure status of the intent (get the pudding) still unresolved as there may still be other avenues allowing access to the pudding. Only once the character decides there's no way she's getting any pudding and thus abandons it for something else can the intent also be declared a fail.

The action does stop when the rod is lost, in the sense that that is the narrated consequence for the failed check, and now the player has to declare a new action for his/her PC. Maybe s/he dives into the ravine after the rod (or tries to climb down in search of it). Maybe s/he keeps going to the top without it.

OK, we seem to agree completely on this bit.

In the latter case, the GM may well not call for a roll, if there is no sense of any more interesting consequences resulting from the climb,

I'd say another roll is called for no matter what; as the character can still fall or find some other way to mess it up...or not.

One thing not yet really mentioned but worth considering: were this my game I'd have either myself or the player roll to see how far up she'd got before the failure occurred; and depending on why she failed this might affect the second climb roll - if she was 3/4 of the way up and lost some gear she'd get advantage on the second roll to reach the top, for example; but if the fail was caused by the mountain simply being too difficult to climb beyond that point she'd be at some sort of penalty were she to try and keep going anyway.

and the real action is in trying to recover the pudding without the rod. That would be an instance of "say yes or roll the dice", which is another technique fairly common in scene-framing, "fail forward"-type play.

The loss of the rod adds another challenge but doesn't negate the first one. It just adds more "real action", to use your term.

To me, this implies that a rod can be lost when a character falls due to a failed climb, but a rod can't be lost without the character also falling, because there is (in D&D) no separate mechanic for determining whether or not gear is lost on a climb.

In any event, reiterating that, for you, the only stakes to a Climb roll are "Do I climb or do I fall" is a clear reiteration that you don't like "fail forward"-type techniques. Key to "fail forward" is that the stakes are governed by intent as well as by task.

Intent or task notwithstanding, I think we agree there's more ways to fail than just falling. A loose foothold might give out leaving her stuck in place, for example, unable to keep going or to descend without falling but still safe as long as she can hang on; which she'll have to do until someone can come to her aid. Or she might find herself unable to progress further but safely able to return to ground. Or she might get her foot stuck in a crack in the rock. None of these have anything to do with losing any gear, they're just things that can go wrong while climbing.

Lan-"or she might just fall and die; that also happens sometimes"-efan

 

[Maxperson]

To me, this implies that a rod can be lost when a character falls due to a failed climb, but a rod can't be lost without the character also falling, because there is (in D&D) no separate mechanic for determining whether or not gear is lost on a climb.

In any event, reiterating that, for you, the only stakes to a Climb roll are "Do I climb or do I fall" is a clear reiteration that you don't like "fail forward"-type techniques. Key to "fail forward" is that the stakes are governed by intent as well as by task.

There are two fail forward techniques that are being discussed here. I like the one where if you fail, there are still other options available to you that can lead to success. With a few exceptions, I do not like the one where you succeed anyway, but with a cost.

 

[Imaro]

Of course the GM is narrating consequences that s/he thinks are fun and interesting! What else would s/he do - narrate consequences that s/he thinks are frustrating and boring?

First let me say I'm having a hard time understanding what this is addressing in the post you quoted from me... nowhere in that quoted post do I say anything about a GM narrating consequences that are frustrating and boring... now that said I'll comment on it anyway.

Let's try to keep from being disingenuous here... there are a multitude of possibilities between... narrating consequences s/he thinks are fun and interesting (I'm assuming that also successfully engage the players) vs. narrating consequences s/he thinks are frustrating and boring (which I'm assuming do not engage the players.)...

From the GM narrating what he/she finds personally fun/interesting (that does not gel with or engage the players) to what the players think is fun and interesting (but is never considered or not deemed so by the GM). In other words, and I think you knew this, my point was that this gives an immense amount of power to the GM in pushing, shaping and creating the direction (not just challenges but the actual direction and shape) that the narrative takes... even when it has little to do with the mechanics the players themselves are engaging with... Loose an item because you failed to climb. Spring a trap because you failed to locate it. I am asking what checks are there to stop this from becoming a railroad and or stepping all over player agency? It's very easy for a GM using this method either consciously or not to push the "story" in the direction and shape they desire. Thus why I can see it being viewed as railroady... even though it doesn't have to be if done well...

As to whether or not the consequences of failure are spelled out, and perhaps negotiated, before each check - as I also mentioned upthread, with reference to both the rules and GMing advice for BW, this is an issue of table practice, GM/player rappor, the vibe of the moment, etc. But even in your example - if the player would have preferred his/her PC to fall down the ravine, then nothing is stopping an action declaration to that effect: "I jump down the ravine after my divining rod, hoping to catch it like Gandalf does Glamdring in the opening sequences of The Two Towers film."

But this isn't what I'm asking for. What if I believed I could grab an outcropping or survive the fall and thus would rather fall then loose my rod? Why do you the GM get to decide that is the consequence when the mechanics I was engaging with are the mechanics for climbing... not for dropping or loosing items...

 

[iserith]

OK, but I-as-player have no way of knowing that; and nor should I. To me there should always be a question hovering over something like this...did I not find it because it isn't there to find, or because it's there and I missed it?

And as a side note: this is why these rolls should ALWAYS be made by the DM and kept hidden. And even when a roll will auto-fail because the trap is 30 feet away the DM should still go through the motions of making one, to preserve the mystery.

I use the dice for resolving uncertainty - not for creating it.

Then again, trap-searching probably isn't the best example for a fail-anywhere discussion as it's pretty binary - you find it, or you don't; with the only other question being if you don't find it by search do you find it the hard way?

It's only "binary" in the way you suggest if you make it that way. I've already provided an example upthread of a PC searching around and stepping on a pressure plate after a failed check. The character has found the trap suspected to be in the area, but now his or her foot is on a landmine as the trap whirs into fell motion. I've narrated the result of the adventurer's actions as putting him or her in a spot and now more decisions need to be made to get out of the predicament safely.

In the doing, I didn't need to take the dice from the players or keep them guessing. I resolved the uncertainty of the player's stated goal and approach and framed a new challenge as a result of a failed check.

 

[Manbearcat]

I'm not sure I'm understanding the question... is it "mandatory" in what sense?

To satisfy your "agency-o-meter", let us call it. If you can, could you take a look at the example I composed above and maybe comment on the necessary PC build components and resolution mechanics to satisfy your "agency-o-meter" in a Basketball RPG?

Which all means you're probably best off just reacting to what the players give you. If they give you more detail than is really needed, resolve at that level of detail as it's probably what they want. If they give less than is needed, ask for more until the detail level is enough to give a resolution.

The issue for me is that a lot of this is contingent upon stuff that isn't malleable...stuff that is codified into system; the level of abstraction built into PC build and the resolution mechanics, the genre expectations, the the overall play agenda/priorities, and the GMing principles that all serve to drive play.

As has been mentioned by @pemerton and chaochou, Moldvay Basic, Tunnels and Trolls, and Torchbearer are very different games than Burning Wheel, Dungeon World, and D&D 4e.

A game that puts at its centerpiece the solving of long-term logistical puzzles (getting as much stuff out of this dungeon before you have to turn back because of depleted resources or before you're squished) is going to be very different from a game that puts at its centerpiece dealing with thematically compelling (meaning the PCs have buy-in) conflicts RIGHT NOW.

A game that requires <--> level of zoom for each player action declaration is going to be different from a game that requires <-----------> level of zoom for each player action declaration. The first might require 5 steps of task resolution to get from one point of play to another. The latter might require 2. The first has granular PC build resources and resolution mechanics while the second has abstract. The first might tightly constrain individual outcomes of resolved action declarations to outcome a or b (and possibly c) with little to no advice to the GM on moving the fiction from action declaration to outcome/fallout (because all of that overhead is already performed/constrained by the task resolution system). Contrast with the second, where outcomes are not auto-performed/constrained by a bounded task resolution system, but rather by robust GMing advice/principles and the top-down agenda of the game.

There is some overlap in play priorities and in GMing advice, but it is not considerable (and that overlap is subordinate to other dominant play/GMing imperatives).

 

[pemerton]

A game that requires <--> level of zoom for each player action declaration is going to be different from a game that requires <-----------> level of zoom for each player action declaration. The first might require 5 steps of task resolution to get from one point of play to another. The latter might require 2. The first has granular PC build resources and resolution mechanics while the second has abstract. The first might tightly constrain individual outcomes of resolved action declarations to outcome a or b (and possibly c) with little to no advice to the GM on moving the fiction from action declaration to outcome/fallout (because all of that overhead is already performed/constrained by the task resolution system). Contrast with the second, where outcomes are not auto-performed/constrained by a bounded task resolution system, but rather by robust GMing advice/principles and the top-down agenda of the game.

Just to add to this with a concrete example. Let's compare opening a stuck door in classic D&D vs in Burning Wheel. ( [MENTION=29398]Lanefan[/MENTION], [MENTION=48965]Imaro[/MENTION], may be interested in this. Or not, as appropriate!)

In D&D (think OD&D, or 1st ed AD&D, or Moldvay Basic as the paradigms here) there is a chance to open doors that is determined by the STR stat. Each roll of that chance represents, in the fiction, one concrete attempt to bust down the door. A second attempt requires a second roll. (For some sorts of doors, like Wizard Locked ones or when attempting to bend bars/lift gates, there is no reroll permitted. This is an interesting early instance of Let it Ride, though I think more for game-balance purposes than narrative momentum purposes.)

Each attempt is likely to trigger a wandering monster check by the GM (due to the noise), and a failure will alert any inhabitants who are behind the door. While there is no in-principle limit on retries (subject to the exceptions noted above), in practice the passage of time (say, 1 round for each attempt) which depletes torches or Light spells, as well as the risk of wandering monsters or alerting inhabitants on the other side of the door, means that the players are likely to apply their own limits.

In BW, opening a stuck door would generally be a STR check. (In BW the STR stat is actually called Power, but that's a mere detail in this context.) Only one check is permitted (ie Let it Ride applies). That check represents the PC making his/her best efforts. There is an explicit rule (called Working Carefully) that permits the player to stipulate that the check takes 50% more time than usual, and which then grants a bonus die in the dice pool, but that gives the GM licence, if the check fails, to introduce a significant time-based complication. So a player could declare that his/her PC is opening the door "carefully" - which in this case probably means having more tries, with a longer run-up - but if the check fails despite the bonus then the GM can narrate a nasty time-based failure (eg "The inhabitants on the other side have heard your first attempt, and have readied themselves in an ambush and then pull the door open on you just as you hit it for your second try, so that you stumble into the room and find yourself at their mercy").

Those are different resolution systems which, in their application, generate different consequences. The D&D system, in its very application, tells you how many physical, concrete attempts were made to break down the door and what the consequence of each one was, whether success or failure. The BW system, as an equally inevitable consequence of its design and application, leaves this stuff unspecified until the dice are rolled and, in the case of a failure, the GM narrates the consequences of that failure.

Within classic D&D there is another resolution system which is noticeably less concrete and more abstracted than the one for opening doors. That is the one for thieves climbing. In the application of the climbing mechanics, there is no scope for taking account of considerations like finger strength vs balance vs facility with any equipment being used, etc. These are all just bundled up into an undifferentiated mass called the Climb Walls chance.

And if the roll to climb is a failure, nothing in the resolution determines whether the thief fell because his/her foot slipped, or his/her fingers gave way, or s/he reached up into the darkness hoping to find a handhold but the surface was smooth, or . . . If any sort of detail of that nature is to be introduced, it will have to be by the GM's narration of the failure ("Schroedinger's handholds"). If a player wants to have regard to particular details of handholds, of finger strength vs balance, etc, the system has no way as written to accommodate that. Either the GM would have to design a new, more detailed climbing system - and then, perhaps, "map out" the details of each wall that might be climbed - or else the player's description of how his/her thief PC uses the handholds and balances on this leg so as to be able to reach up to that small crack, etc, will all just be colour that is irrelevant to resolution.

It's in the nature of "fail forward" mechanics to work at a level of abstraction at least equal to that of classic D&D climb walls, and in most cases probably even more abstract than that.

 

[innerdude]

It's in the nature of "fail forward" mechanics to work at a level of abstraction at least equal to that of classic D&D climb walls, and in most cases probably even more abstract than that.

This is interesting, because this seems to go along with my general "off the cuff" times when I turn off "fail forward," based on stakes and opposed rolls.

In most cases, opposed rolls operate at a very low level of abstraction --- there's typically a very direct interplay between two opposing mechanics, or checks based on those mechanics. Thus, "fail forward" isn't generally appropriate, because the level of abstraction is low.

For very "high stakes" actions, this is also generally the case --- If something absolutely critical is at stake, the nature of a given check is generally highly specific. The stakes at hand and the consequences of success and failure are generally quite clear to the player. As such, "fail forward" in these instances often plays against what's happening "in frame" in the game world.

In thinking about it, though, I'm not sure that even "very high stakes" actions can't have elements of "fail forward" in them. This would be something interesting to discuss; I might posit a hypothetical around this in another post to see what people think.

I'm also re-thinking somewhat what I said earlier about "low stakes" checks. In hindsight, I don't know that it's a good idea to include "low stakes" mechanical checks very often as a GM. In context, it's not that a given check is ever totally "low stakes," it's that there's a strong clarity from the viewpoint of the players that there are other viable options should failure occur. It's "low stakes" in the sense that progress is not halted outright, it's only stalled temporarily. From a player and character viewpoint, though, a check is always at least "medium stakes" in the sense that any course of action represents an investment, and failure always represents to some degree a loss of resources (even if it's just time).

 

[Imaro]

To satisfy your "agency-o-meter", let us call it. If you can, could you take a look at the example I composed above and maybe comment on the necessary PC build components and resolution mechanics to satisfy your "agency-o-meter" in a Basketball RPG?

There is no specific level for me... it very much depends on the abstraction level of the game I have agreed to play. If your basketball game has a simple mechanic in which the entire success or failure of the game is decided by a coin toss and I have agreed to play... well that is all I need. However if your system has the granularity of attribute/skill scores & checks and through an unrelated skill check you have decided I was not smart enough (Int) to remember to tie my bag down properly, or too clumsy (Dex) to hold onto my divining rod, or so weak (Str) that it was pulled from my grasp without us finding that out through the mechanics then yes I feel like you are treading on my agency in that particular game. Same with a trap that says I must step in a specific space to activate it (and movement is tracked on a granular enough level to determine if I did) but because I failed a search check (regardless of whether I specifically moved into said area or not)... I must have stepped right on it.

 

[chaochou]

Those are different resolution systems which, in their application, generate different consequences. The D&D system, in its very application, tells you how many physical, concrete attempts were made to break down the door and what the consequence of each one was, whether success or failure. The BW system, as an equally inevitable consequence of its design and application, leaves this stuff unspecified until the dice are rolled and, in the case of a failure, the GM narrates the consequences of that failure.

I go all the way back to my scene-framing post and reiterate that the difference I see between the two systems is between Character vs Competence where 'character' means beliefs, flaws, dependencies, relationships, threats, hopes and fears.

Character forms the building blocks of BW. Stats and resources (competence) are the building blocks of D&D.

 

[Lanefan]

Just to add to this with a concrete example. Let's compare opening a stuck door in classic D&D vs in Burning Wheel.

In D&D (think OD&D, or 1st ed AD&D, or Moldvay Basic as the paradigms here) there is a chance to open doors that is determined by the STR stat. Each roll of that chance represents, in the fiction, one concrete attempt to bust down the door. A second attempt requires a second roll. (For some sorts of doors, like Wizard Locked ones or when attempting to bend bars/lift gates, there is no reroll permitted. This is an interesting early instance of Let it Ride, though I think more for game-balance purposes than narrative momentum purposes.)

Each attempt is likely to trigger a wandering monster check by the GM (due to the noise), and a failure will alert any inhabitants who are behind the door. While there is no in-principle limit on retries (subject to the exceptions noted above), in practice the passage of time (say, 1 round for each attempt) which depletes torches or Light spells, as well as the risk of wandering monsters or alerting inhabitants on the other side of the door, means that the players are likely to apply their own limits.

In BW, opening a stuck door would generally be a STR check. (In BW the STR stat is actually called Power, but that's a mere detail in this context.) Only one check is permitted (ie Let it Ride applies). That check represents the PC making his/her best efforts. There is an explicit rule (called Working Carefully) that permits the player to stipulate that the check takes 50% more time than usual, and which then grants a bonus die in the dice pool, but that gives the GM licence, if the check fails, to introduce a significant time-based complication. So a player could declare that his/her PC is opening the door "carefully" - which in this case probably means having more tries, with a longer run-up - but if the check fails despite the bonus then the GM can narrate a nasty time-based failure (eg "The inhabitants on the other side have heard your first attempt, and have readied themselves in an ambush and then pull the door open on you just as you hit it for your second try, so that you stumble into the room and find yourself at their mercy").

Those are different resolution systems which, in their application, generate different consequences. The D&D system, in its very application, tells you how many physical, concrete attempts were made to break down the door and what the consequence of each one was, whether success or failure. The BW system, as an equally inevitable consequence of its design and application, leaves this stuff unspecified until the dice are rolled and, in the case of a failure, the GM narrates the consequences of that failure.

There's no great difference, in the end.

The D&D system doesn't know what will happen until the dice are rolled any more than the BW system does. And if a failure (or success) generates consequences in either system then the DM will (I hope!) narrate such.

I tend to prefer what you call "Let It Ride" in D&D as well; if a second attempt is to be made at a stuck door I insist there be something materially different about it e.g. getting an ally to help, as you've already given plan A your best shot. And there's no reason at all why a failed attempt might not attract unwanted attention in BW the same as it does in D&D. And even though it seems BW kinda forces consequences to occur there's no reason a GM can't override this and simply say "The door remains resolutely closed but nothing else seems to have changed."; just like a D&D DM is always able to narrate a set pull-the-door-open-at-the-right-moment ambush if she wants to.

Within classic D&D there is another resolution system which is noticeably less concrete and more abstracted than the one for opening doors. That is the one for thieves climbing. In the application of the climbing mechanics, there is no scope for taking account of considerations like finger strength vs balance vs facility with any equipment being used, etc. These are all just bundled up into an undifferentiated mass called the Climb Walls chance.

True, just some modifiers for the type of surface being climbed. Balance is kind of in there as part of Dex, which does modify your climb chance. I can't offhand remember whether Str affects climbing in RAW 1e, it does to a small extent in my game but that might be a local tweak. Equipment is variable enough that it's almost up to the DM to modify the chance each time based on what's being used and by whom.

And if the roll to climb is a failure, nothing in the resolution determines whether the thief fell because his/her foot slipped, or his/her fingers gave way, or s/he reached up into the darkness hoping to find a handhold but the surface was smooth, or . . . If any sort of detail of that nature is to be introduced, it will have to be by the GM's narration of the failure ("Schroedinger's handholds"). If a player wants to have regard to particular details of handholds, of finger strength vs balance, etc, the system has no way as written to accommodate that. Either the GM would have to design a new, more detailed climbing system - and then, perhaps, "map out" the details of each wall that might be climbed - or else the player's description of how his/her thief PC uses the handholds and balances on this leg so as to be able to reach up to that small crack, etc, will all just be colour that is irrelevant to resolution.

It's in the nature of "fail forward" mechanics to work at a level of abstraction at least equal to that of classic D&D climb walls, and in most cases probably even more abstract than that.

A DM is always free to narrate whatever makes sense at the time; a player also could in this case, in character.

But here we get back to the same argument as with Mt Pudding: can you in any way reach the top (i.e. succeed) on a failed roll in either system? If yes, then "fail" is probably the wrong word to be using.

Lanefan