The finer details that enable meaningful conversation about it.
That is how to reach a sufficiently common understanding of "fail forward" that we avoid conversations of this kind:
A: Avoiding the game stalling by the player getting stuck on a thing is great, and is called fail forward.
B: I agree. I generally never allow retries, and make sure there are always other options to pursuie on a failure, making sure the game never stalls.
A: Wait a second, that is not fail forward because [definition I of fail forward]
B: Oh, then I do not think I like fail forward as it seem to encourage [undesired behavior X]
C: But [undesired behavoir X] is not fail forward because [definition II of fail forward]
D: But [definition II of fail forward] seem to allow [undesired behavior Y] which would not be normally allowed in my game.
E: I do not see the problem with [undesired behavior Y]
D: Explains why [undesired behavior Y] is not working out alongside their preferences.
F: Pointing out that it is perfectly possible to just not get into the problems of [undesired behavior Y] with fail forward because of [definition III of fail forward].
G: Agreeing with F pointing out that fail forward is perfectly ok as simply not allowing retry would be failing foward according to [definition III].
D: Expressing sceptisism is this is indeed fail forward.
G: Reiterating that according to G's understanding just not allowing retry is failing forward.
A: (Having missed some key parts of the context due to the thread moving at breakneck speed) G-s understanding is wrong because as pointed out earlier [reiterate definition I of fail forward].
We are in a situation where there are several ideas of what the word "fail forward" means, and everyone keep replying as if their own understanding of the term is the universally correct understanding of the term, so people keep correcting each other for misuse of the term.
"Things shouldn't hit a dead end" is a
motive for FF. It is not
what FF is. "Children should know how to read" is a motive for public schooling; it is not what public schooling is. A given attempt at/implementation of public schooling may or may not fulfill any given motive; any given motive may or may not be a good or sound one for why public schooling should exist. Likewise, "a literate population" is not public schooling, it is an effect of public schooling.
What Fail Forward
is is, IMO, quite simple. "When a player attempts an action, all results advance the fictional state in some new direction." Usually, this means that one
effect of FF is that, when an action fails, the state of play keeps moving toward...something. That "something" may be good, or bad, or merely different--but whatever it is among those things, it is not functionally unchanged.
Hence, with the locked
château servant entrance, it is
not Fail Forward to have a failure produce: "You cannot enter by this door. All remains quiet." That is the state of play remaining, functionally, completely unchanged. The characters' motivation remains identical to what it was before: find an entrance, remain undetected. The scenario in which they find themselves remains functionally identical: they are outside the house, not inside. And, since I was specifically told to
limit what assumptions I might make and to presume only the barest minimum of anything, this was
the only entrance the characters had, so not only have things remained unchanged, they're stuck with no further path forward.
Another common example (not previously used in this thread, to my knowledge) is one where the PCs are trying to stop an occult ritual, perhaps to save a captured victim's life, perhaps to identify the cult members. The characters
need to find the secret entrance in order to get to the cultists. There isn't any other way to get to the cultists
except to find the entrance. In traditional D&D play, this means you need to reach a certain minimum roll on a Perception check--so we get two situations if we presume failures. Either the characters just all attempt it and fail and then...the adventure just
stops. Or, they all keep trying over and over and over and over until finally someone rolls a nat 20 (or whatever) and finds the secret bookcase (or whatever the secret entrance is). Both of these are, generally, understood to be Bad Results, even by those who heavily favor ultra-traditional play approaches. Fail Forward addresses both by saying that a failure doesn't mean they simply cannot find the door at all; instead, it takes a slight step back and asks, "What does failure actually MEAN here?" Here, the party's goal is, as noted, to save a victim, or to unmask the cultists. Hence, a failure of Perception,
when they ABSOLUTELY MUST perceive in order to proceed, should not be understood as merely "you simply don't see anything, nothing happens".
Instead, since some kind of success (or perhaps some kind of failure!) is required for things to proceed, failure-vs-success is now a matter of "do you achieve the intent you sought", or to reference Lanefan's suggestion with the unavoidable iceberg, what is the nature of the collision? Is it a minor, glancing blow, something the ship can at least limp back to port to fix, or is it a tremendous, devastating blow that ensures total destruction? The act of treating the hypothetical "Pilot"/"Navigate" roll's success as "you mitigated the problem as best you could, and things might end up bad but survivable", and its failure as "you have turned a serious problem into an unmitigated, immediate disaster"--THAT is what Fail Forward is. Both success and failure advance the fictional state; they just do so in either beneficial or detrimental ways.
Looping back to the "find the secret entrance" example, this means the Perception check isn't--and shouldn't be understood as--"is it possible for you to perceive this?" Because that question is not contributing to play, as success is merely rubber-stamping continued play, and failure is merely bureaucratic hold-up that might evolve into total derailment. Instead, it should be understood in the same way as the hypothetical "Pilot"/"Navigate" roll mentioned above: success vs failure is a measure, here, of your
speed in perceiving. This idea, that a roll may sometimes be used purely to succeed on a discrete task, and other times to determine how much you can do or how fast you can do it etc., is perfectly cromulent with some existing D&D rolls, e.g. Acrobatics to determine how quickly you can move along a narrow ledge.
Fail Forward simply means that we make that apply a little bit more broadly. The point of picking a lock is not simply the determination that the lock is open (or not open, for failure). If that were so, most doors would just get
broken if they were locked, since that's almost always easier than picking a lock! Instead, just like with the "Piloting the Titanic" example, we care about things a bit further beyond: the overall safety and integrity of the ship, not merely whether the path it takes is the one you wanted it to take. With the lock, generally, one cares about remaining undetected in some way: residents(/guards/servants/etc.) don't know you're there, no visible damage is present, no obvious signs exist to link back to you, etc. Hence, just as "failing this Pilot roll means the Titanic is badly damaged" has been seen as a valid fail-state despite NOT being the instantaneous result of doing the piloting, "failing this Lockpicking check means someone is coming to investigate" seems a reasonable direction to take affairs. In both cases, they immediately lead to a new conflict of some kind, so the state-of-play advances, just in a direction generally undesirable to the characters.