This is a bonus spin-off thought from the main mini-series. It doesn't quite fit into the main thread.
Caring and Minding
There's another case here which I think Fail Forward handles suboptimally, but which is relatively common. This is the situation where I care what the result is, but I don't mind.
Now, you can make a case that this is exactly the situation Fail Forward tries to create, by making both options equally interesting. But let's ignore that, and focus on situations where it's inherently true, rather than those where the GM makes it true.
This is actually a really common situation in the types of games I tend to play, which are generally relatively simulationist. The most obvious case is the very frequent one where I simply want to establish facts about the world around me. For example: Is there a secret door in this wall? Is this ruin historically associated with the Dark Lady? What kind of outfit is Hilgarth Enterprises? Can I hear anyone on the other side of this door?
I care what the result is, because otherwise I wouldn't have asked.* But I don't mind. I have no personal stake in the specific answer, but collapsing that particular waveform will open up new lines of enquiry or simply inform my future actions.
You can usefully distinguish it this way:
- I care whether there is a secret door, a letter for me, anyone inside the room, any reason to suspect Lord Surly of treason, or a pretty girl eyeing me across the room. I don't particularly mind whether those things are true or false.
- I mind whether my gold has been stolen, a balrog is chasing me, I can escape my manacles, I know a spell to banish this spirit, or I have any ammo left.
For example, if there is someone on the other side of the door, they might be a prisoner in need of rescue, or a dangerous enemy. Either of those might be good (I do a good deed and perhaps gain a reward or information; I test my mettle and gain XP) or bad (I have to keep the prisoner quiet, free them, wonder whether the guards will notice and raise the alarm, and complete an escort quest; the enemy kills me). I don't specifically want there to be someone there, but I do want to use my senses to establish that fact.
Similarly, if I'm playing in a setting with a Dark Lady in the backstory, and exploring a sinister tower where something has happened, it makes sense to rack my brains and check whether this tower is associated with her. If so, that suggests some things that might be happening: some of her followers might be here, or the tower might be rife with undead, or an evil influence might flow from a relic buried here. I can use that information to make guesses and prepare accordingly. If not, I want to think of other possibilities and carry out appropriate research. Either way, I want to use my historical and mythical knowledge to know the answer.
A Fail Forward system doesn't handle this particularly well, because it's not clear what reasonable consequence there could be for failing "I look for secret doors". In extreme cases it doesn't handle success well either. After all, you're not trying to create a secret door with sheer force of will; you just want to know what's going on. If I succeed at listening at the door, I don't want to have to tell the GM that there's someone inside.
In fact, Dungeon World handles this by not having any consequences for this kind of failure. Which I think is probably better than the alternative, but feels a little inconsistent. It does seem to highlight that adding consequences to an event is very much not always an improvement.
It gets even worse with "I check for traps". The very last thing you want is to try to ensure your safety by making sure there's no traps, and instead end up actively creating a hazard that then injures you. Or, coming back to our secret door, to suggest that there might be a secret door, fail the roll, and suddenly be ambushed by goblins emerging from a secret door that didn't exist until you thought of it. These are, to be clear, examples of reasonably bad GMing rather than game design per se, but I think they're a decent example of where the idea of dividing character skill rolls into two equally-interesting sets of consequences breaks down.
I also think in general it's worth considering GMing and design advice from the point of view of "what happens if the GM isn't very good?". You can't design away GMing problems, of course. On the other hand, there are always going to be situations where a GM is inexperienced, underconfident, takes things quite literally, assumes the advice will explicitly state caveats rather than assuming they will work it out, tired or drunk.
The thing is that if someone does fail a roll to look for a secret door, the chain of thought that leads to them being ambushed by goblins makes a degree of sense. You looked for a secret door and failed. Well, you can't have failed to find a door that isn't there, so maybe there was a door, in which case how does it go wrong? Ah, there were goblins hiding inside! Or there weren't, but some will come soon and sneak up on you!
Unfortunately this sort of thing affects what you might call the metagame. If, when you look for secret doors and fail, this action can cause you to be ambushed by goblins that otherwise did not exist, you are creating a penalty for yourself by doing something that wasn't necessary (but was fun and interesting) in the first place. The natural response is to avoid looking for secret doors.
This is true of any similar mechanic: if failed Social Interaction rolls can have you attacked by otherwise disinterested merchants, it really deters you from talking to anyone. If fighting Dreadbears gets you horribly injured, you'll tend to avoid fighting Dreadbears.
This kind of gameplay can discourage or sabotage certain playstyles. In particular it harms deduction-style investigation, where players want to painstakingly check leads and suspicions and eliminate possibilities. It also potentially harms cautious tactical play, where players want to use research and observation to gain as many advantages as possible.