I wanted to reply in greater detail but alas by 5 year old is fighting me on bed time right now. Let me just say a few quick things before running off.
LOL. Totally been there. Good luck.
For me, there's a difference between fumble and failure based on the way I use failure in my games. In games like D&D, there is no consequence of failure, unless specific to the mechanic being used (such as falling when making a climb check).
I agree. Again, more with definitions.
Critical Success: Not only did what the player propose to do happen, but he got some additional unusual benefit above and beyond what he expected.
Success: What the player proposed to do happened.
Success with Complications: What the player wanted to happen, did happen, but additionally something that the player did not want to happen also happened.
Partial Success: What the player wanted to happen did happen. However, the player only got part of what he wanted to achieve.
Near Success: What the player wanted to happen, didn’t happen. However, the player did at least achieved something beneficial that prevents this from being complete failure.
Failure: What the player wanted to happened, didn’t happen.
Fumble: Not only did the player not get what they wanted, but the outcome is now considerably worse than we’d expect of a failure.
D&D has no unified mechanic. Depending on what you are doing, you might have the outcomes "Critical Success/Success/Failure" or "Success/Failure/Fumble" or just "Success/Failure". In some cases you might actually have differing degrees of success or failure, which is a whole other approach (See for example Gamma World 3e or Marvel Super Heroes).
Some systems remove the "failure" option from the table. You either get something positive, or else you get something else but you never get "no change". An example would be a system where you announce your attention to attack a foe, and on failure you get damaged by the foe. That's "success/fumble" as outcomes.
All that is well and good if the stakes are known before hand, but some systems define the "failure" outcome (that is, your roll wasn't good enough) in an open ended manner. The Cypher system doesn't quite do that, but it does define the fumble outcome (you threw a 1) in an open ended manner because on a fumble you get a 'GM Intrusion' where the GM is empowered to complicate the situation by introducing new fiction. In the abstract, you can distinguish easily between "partial success" and "success with complications" and a "fumble".
But in general, if the system defines 'failure' as GM Intrusion, then the system becomes so open ended that in practice we won't be able to define failure. 'Failure' in this case will be a range of outcomes from 'Success with Complications' to 'Fumble', and it's a matter of rather subjective opinion which actually occurred.
Let me give an example.
Consider the case of rolling Athletics to determine whether a PC jumps a chasm. A non-open ended system might define the consequences concretely like:
Critical Success: The PC not only jumps the chasm, but moves with such speed and power than they can add up to an additional 2 meters as a bonus to their intended move.
Success: The PC jumps the chasm.
Success with Complications: The PC jumps the chasm, but stumbles on the other side, ending their move at that point. (Unless perhaps a second check is made?)
Partial Success: In order to clear the jump, the PC was forced to lay out, resulting in them landing prone in their intended square and ending their move.
Near Success: The PC failed to clear the chasm, but came close enough that they may grab the edge (perhaps requiring a second check?). They are as a result hanging off the edge of the chasm rather than at their intended point.
Failure: The PC failed to clear the chasm, and as a result fell in.
Fumble: The PC failed to clear the chasm or even to control their motion, resulting in them tumbling head over heels and landing on their head, taking more than the usual amount of damage.
We could probably hard define such a spectrum for almost any sort of check.
This superficially seems like a straight forward spectrum from ‘best’ to ‘worst’. But note we can probably always find edge cases where the above ‘Near Success’ was a better result than ‘Success with Complications’. For example, if the ultimate intention was to charge a bow wielding assailant, the PC achieving only ‘near success’ might at least benefit from near complete cover from missile fire.
If the system is open ended though, defining the clear cut case from 'best' to 'worst' becomes basically impossible, and if its all GM fiat, we can't begin to distinguish one case from another usefully. One GM might imagine that ‘success with complications’ in fact is hanging off the side of the chasm, while another might imagine that ‘near success’ is coming just close enough to the chasm edge that you miss the top but instead land on a lower ledge and take only partial damage. Or indeed, since the two GM’s can’t really compare notes, the reverse might be the case with a different group and a different GM ruling that a PC that achieved ‘success with complications’ is now clinging to ledge 10’ below the top of the chasm. Someone might decide that partial success is clearing the chasm but twisting your ankle on the landing. Someone might decide success with complications is clearing the chasm but landing in a hitherto unnoticed pile of shaving cream or perhaps even failing to clear the chasm but landing in a sufficiently deep pile of shaving cream that little or no damage is taken. And what happens with the GM starts feeling empowered to metagame, taking into account how much of a complication any particular alteration to fiction might be to individual PC’s or to the party? Now we have to deal with each GM’s own subjective sense of difficulty, fairness, and how antagonistic he should be.
As such, in any open ended system ‘near success’, ‘success with complications’ and ‘partial success’ are going to be synonyms. Indeed, I’m not sure any open ended fiat system has actually tried to mechanically require differences between the terms. Maybe there is a system out there were miss by 1 is "success with complications", miss by 2 is "partial success", and miss by 3-5 is "near success" but I haven't read the rules to such a system. Typically they just note that the GM is empowered to treat the failure result as any one of those things according to what they think is best.
I don't think I've seen a D&D tack on fumble chart that incorporated fumble effects that went beyond the immediate character (for example, when narrating the fumble of the archer, the GM describes the creature dodging his attack and tearing down the sails creating a complication for the entire party). The effect is primarily attached to a specific character.
This is because with the exception of 4e, D&D has generally avoided disassociated mechanics. If a player announces his plan to fire an arrow at a target, the implied stakes are that you will hit with the arrow or you won't. It's in the range of easily imagined possibilities that the arrow shot would be particularly poor or particularly good, but the cause "You tried to fire an arrow" and the effect "the monster tore down the sails" don't seem related. In D&D, complications like your sails getting tore down, don't happen as a result of your failure directly. They happen because the antagonist proposed to tear down the sails, and then you failed to thwart that plan and they succeeded.
This is a very different sort of game than open ended DM fiat. I'm running the antagonists, and they fumble as well. Sometimes NPC's go down hard to unforeseen failures and any intention I had for the NPC's to have an aura of suaveness is thwarted. I propose actions for the NPC's, without any certainty that the rules will allow those actions to succeed. My antagonists have limited resources and those resources can be depleted, leaving them without good options. My antagonists can only summon reinforcements if the established fiction says reinforcements are available. The PC's can only become entrapped in hitherto unseen bramble bushes if unseen bramble bushes exist in the fiction. I can't invent resources to complicate the scene. Well, I mean I can, but I believe that doing so would be inherently bad GMing on my part because a big part of my job is giving the PC's a fair shake and I don't believe I will be able to do that if I have an excuse to intrude into the fiction in an open-ended manner.
For me as a player, when my missed bow shot allows an Orc to blow a horn to summon reinforcements from distant rooms, I'm ok with that. But when my missed bow shot conjures a trumpet and a room of reinforcements into being, I just give up. I'm no longer interested in playing that game. I'm certainly not interested in running that game either. For that matter, I know from experience I can't run such a game according to its intention. I'm unable to run Paranoia (for example) as a wacky, gonzo, game where silly things happen. I try to run Paranoia and my basic sense of fairness, and my sense of realism and ability to paint a world very granularly, and my sense of drama and characterization, and my general lack of a sense of humor, ends up turning it into an angsty dark dystopian horror game.