What's the goal of fumbles?
If I were to implement it would be:
- Every character class/build should have a roughly equal opportunity to fumble
- Should be quick to resolve and not slow down the game
- Fun, but not game-breaking results
- Decrease in occurrence at higher levels
- Penalties should primarily affect the PC that fumbles.
- Be applied to monsters/NPC opponents in a way that works for the DM
I'm just not sure how to do it. Front line fighters always end up with the short end of the stick on both suffering penalties and getting more fumbles.
I like the question, and I'll get back to that in a moment.
Your answers are more about the goal of the mechanics of the fumble, rather than the goal of the fumble itself, but it's good stuff. There are two points I want to address.
First, I was thinking about the statement that front line fighters end up with the short end of the stick, particularly in getting more fumbles. Statistically, all characters who make attack rolls suffer the same amount of fumbles. However, since fighters attack more frequently, they will suffer more fumbles. That might sound like double-talk, but it's not really. It's like the statistic that most accidents happen within x miles of home. Part of the reason that's true is because of the sheer amount of opportunity within that radius.
So with this in mind, and the fact that fighters are exposed to potential critical fumbles more frequently, I'm OK with the idea that they are more likely to suffer from them as well. That doesn't mean that the critical fumble (if you use one) needs to be redesigned. Instead, it means we need to look for similar opportunities in the things other classes are more likely to do. A number have already mentioned spellcasting, and I think that's a great idea. I already have a wild magic surge system that can be triggered by various circumstances (like being struck for damage while casting or concentrating on a spell). Rolling a natural 1 seems like another reasonable trigger, considering I use that as a trigger for a fighter.
Second, the decrease in occurrence at higher levels. Sounds reasonable, but I think that's also related to the goal of the fumbles themselves.
So back to your first question - what's the goal of fumbles? I think there will be a lot of answers to that question. And I don't have a ready answer myself, so I'll go with what comes to mind right now.
The goal of fumbles for me is to simulate those points in a battle where the momentum changes, the luck changes, something shifts against one side or the other. It's not quite the opposite of a critical hit, but close.
Even the best trained fighter can find that a miscalculation here, an unexpected obstacle there, can change the course of the battle. Most combats are in unfamiliar environments, that aren't ideal circumstances, and aren't scouted for developing strategies, etc. (Which also leads to the possibility that a well-planned attack might suffer less fumbles).
A big example? The battle of Agincourt. The French were well trained knights, who knew how to fight. They made some major miscalculations in the terrain, they were forced into a bottleneck in a muddy field, in heavy armor. Yes, these were far beyond a critical fumble on a die roll. But historically there have been a lot of very good military men that made some very bad decisions. So it's not a perfect example, but I'm hoping it highlights that skill level doesn't always mean that you won't suffer mishaps.
I'm still on the fence about higher levels. One of the suggestions I made was to have a threshold, like a 1 plus miss by 5 or 10. That reduces the chances as you gain levels, but if a critical fumble (at least for me) is in part a moment in time where the momentum or advantage changes, often with unknown reasons, wouldn't it be more about the relative skill levels of the two combatants, rather than how much they missed by? Particularly since AC is not primarily a measure of skill?
But, part of that threshold is also to make them less frequent in general. A critical fumble happening 5% of the time is quite a lot. If it happens too infrequently, though, it's either meaningless, or really annoying the one time a year it happens.
Although our system has been working well to this point, I really want to go back and see if I can leverage this a bit more. I like the idea that it might be tied to the turning battle. Some expanded possibilities - the creature that makes the critical fumble can be "pushed" (more like driven, with an attack) 10 feet in a direction of the target's choice. Or the target can disengage as a reaction. I guess a better way to word it, is that the target can use their reaction to make an opportunity attack, push the creature that made the critical fumble 10 feet in any direction, disengage, and other possibilities.
Time for some thinking...