Tips for using fumbles/critical failures in your tabletop game:
1. Don’t. Critical failures are typically appropriate only for explicitly comedic games. This isn’t to say that funny things can’t happen in any game - certainly they can! - but by employing critical failures, you end up with a game where the rules themselves mandate a certain minimum level of slapstick bullshit, regardless of circumstance. It doesn’t hurt to consider whether that’s actually the kind of game you want to run.
(Note that comedy doesn’t just mean Looney Tunes. A dystopian bureaucracy milieu where people die horribly for making mistakes on paperwork is also a comedy game - for certain values of comedy! - so critical failures may be appropriate there. This guideline isn’t intended to restrict, but to encourage you to think about the rules in terms of what style of play you’re trying to foster.)
2. Respect player agency. A common error of novice GMs is to use critical failures as an excuse to hijack players’ characters and have them do things they’d never do voluntarily simply because it would be funny. An example I see frequently is “you fumbled your First Aid roll, so instead you stab your patient in the face with a knife”; that’s good for a cheap laugh, but unless the tone of your game is straight up Looney Tunes, it’s not a reasonable outcome of simply being very bad at first aid. Bad dice rolls represent errors in performance and judgement, not random demonic possession.
(Obviously narrative context is important here; “my character is possessed by a minor demon that forces her to do something pointlessly evil every time she rolls a fumble” and “my character is a literal space alien who often harms people unwittingly because she doesn’t understand how humans work” might both make the face-stabbing thing acceptable, because now it reasonably proceeds from established characterisation. Those are outliers, though.)
3. Keep the magnitude of fumbles commensurate with what was attempted. Inadvertently starting a war by critically failing a Subterfuge check to lie to the King may be reasonable; inadvertently starting a war by critically failing a Streetwise check to gather rumours in a tavern typically won’t be. Barring exceptional circumstances, players should have a reasonably good idea of what’s at stake whenever they pick up the dice, and disproportionately harsh critical failures make it impossible ever to know what’s at stake.
This also applies across classes of activities. If the typical result of fumbling an attack roll is “you pull a muscle” and the typical result of fumbling an Athletics roll is “you slip, break your neck and die instantly”, the message you’re sending to your players is that you’d prefer them to resolve their problems with fistfights rather than footraces in your game - which is a problem if that’s not the style of play you’d intended to encourage.
4. Don’t foreclose: redirect. A critical failure that simply blows the players’ plans out of the water and renders whatever they were trying to achieve impossible is bad not because it’s unfair, but because it’s boring. Blocking is poor improv, and it doesn’t become less bad just because the dice gave you an excuse. A good fumble simply creates complications that need to be addressed, or shuts down the players’ current approach while creating or highlighting a different route to the same goal.
(This is especially important to keep in mind when your players’ plans require more than one roll. Even if the odds of critically failing any one roll are very low, the likelihood that at least one of a long series of rolls will turn up a fumble is very high. If any one fumble renders the whole plan impossible, nothing will ever get done. Most tabletop RPGs are strongly informed by heist fiction, so take your cues from capers: disasters and opportunities are the same thing.)
@kittenclysm replied:
Fumbles don’t have to be nutty slapstick, that’s just what a lot of people do. Having your bowstring snap or your spell puff out isn’t necessarily comedic, it can also be serious. Very serious in some cases, that’s the point of fumbles.
They can be described that way, yeah. However, as fumbles are typically implemented (i.e., the “a natural 1 on a d20 is a critical failure” approach), the rules are working against doing do.
The compound probability issue cited in point 4 is your enemy there: even if critical failures are rare on any individual roll, risky activities - particularly combat, in most games - tend to involve a whole lot of rolls, so in aggregate they’re happening constantly.
Between the spectacle of a group of ostensibly highly skilled characters having improbably terrible things happen to them constantly for no reason and the accompanying creative fatigue of needing to describe that many critical failures, things are going to quickly slide toward comedy.
Let’s suppose that you want to avoid that issue but still want to have critical failures, though. It’s not my favourite way to play - I prefer to keep fumbles to their native idiom - but there are a few basic approaches you can use to successfully incorporate critical failures into a more serious game.
1. Make ‘em rare. The simplest way to avoid critical failure pileups turning your game into farce is to make them very unlikely on an individual basis - much more unlikely than the old one-in-twenty rule. You’ll still get a reasonable number of them overall because of our friend compound probability, but you’ll avoid big slapstick clusters of them.
If you’re using large enough dice - e.g., percentile dice - this can be as simple as shrinking the critical failure range. Dice pools, conversely, can make the odds of a fumble dependent on the acting character; many dice pool systems impose a critical failure only if all of the dice come up 1s, which means fumbles are fairly common when you don’t know what you’re doing, but become very rare when acting within your sphere of competency.
(If you’re not using percentiles or dice pools but still want to adopt this approach, an easy hack is to steal the crit confirmation rule from D&D3.5E/Pathfinder: when the dreaded natural 1 comes up, make a second attack roll or skill check with identical modifiers to “confirm” it. If the second roll also fails, it’s a fumble; otherwise, you’ve made a last-minute save and it’s just an ordinary failure. This belongs to the “critical failures are less likely within your sphere of competence” variant of this approach.)
2. Make ‘em voluntary - for the player, if not necessarily for the character. As above, there are two variants of this approach: making the circumstances under which fumbles can occur voluntary, and making accepting the fumble itself voluntary.
Chronicles of Darkness uses the first variant: if you end up in a situation where you’re eating enough penalties to push your dice pool all the way down to zero, and you choose to roll for it anyway, you can get a critical failure. On the flip side, any number of action games use a risk-for-reward implementation where you can claim a bonus on a roll by describing extra risks you’re taking, with the cost that if you take the bonus and fail anyway, it’s a fumble.
Making the fumble itself voluntary is more of a storygame thing, so how it works depends a lot on how the game in question is set up, but the gist of it is that when the opportunity for a critical failure arises, you can either claim a bonus for accepting it or spend some sort of game currency to reject it.
(A simple hack of of the second variant for D&D5E might be to change how inspiration works so that you can spend inspiration after seeing the result of a roll in order to reroll it, rather than spending it up front for advantage - though this alone may not be enough to avoid the fumble pileup problem unless you also increase how frequently inspiration can be used. This obviously makes inspiration a lot more powerful; if you’re adding fumbles to a D&D5E game, that may not be unwarranted!)
3. Make ‘em situational. This could be considered another variant of “make ‘em voluntary”, but it has enough potential wrinkles to be worth discussing on its own. The basic idea where is that critical failures apply to some but not all rolls - and the rolls they do apply to are rare enough that fumble pileups just aren’t a thing that happens, freeing you to keep fumbles frequent in that one specific context without the risk of descending into slapstick.
For example, in a game where the danger of magic is a major theme, you could have critical failures apply to rolls to cast spells, but nothing else. Beyond the Wall and Other Adventures - a game about community-building where magic users are cast as dangerous outsiders - uses this approach, paired with a variant of the voluntary-fumbles approach whereby a magic-user can avoid having a spell go out of control by choosing to becoming exhausted instead.
(Making “combat” the particular situation in which critical failures are used may be intuitive, but combat involves so many rolls in most games that you’re likely to just reintroduce the fumble pileup problem unless you’re talking about a game where fighting things is rare.)