Friday, 21 October 2016

On failure, its outcomes and its implications: consequences


There are basically three topics in this miniseries. Thread one is about (mechanical, die-rolling) failure per se and how it feels to me as a player. Thread two, today's topic, is about failures, interestingness and the Fail Forward/Roll Dice or Say Yes sort of concept. Thread three is about the interaction between failure, narrative and the coherence (in the non-Forge-jargon sense) of the play experience.

Failing Forward

One of the arguments that came up in our conversation was more or less this:

"Fail forward" systems (like Dungeon World), which insist on providing "interesting consequences" for failure, just create an unnecessary burden on players and GMs to improvise "consequences" for things that really don't need them, or which already have consequences that arise organically from the failure.

This touches broadly on both Fail Forward and Roll Dice or Say Yes. Both of these proposals work roughly on the basis that die rolls (or whatever resolution mechanic) should be used when you're indifferent to the outcome or when both outcomes are equally desirable. This idea is built into things like Dungeon World's dice mechanic, but also links to a lesser extent to Numenera and its GM Interventions. The principle seems sound, but in application is gets trickier.

I think there are several points to consider here, including:

  • Why are we rolling dice?
  • What is the resolution mechanic resolving?
  • What is a consequence?
  • What is "interesting"?

As a quick aside, let me say that I don't think Fail Forward is an inherently flawed idea, although I will be suggesting a lot of problems with it. I think it emerges from some genuine problems (like hitting a dead end or constant whiffing) and is a sensible means to address certain situations in games (more on this far below). I try to apply it in Call of Cthulhu, for example, to avoid discouraging players and keep investigations moving. However, I think generalising it to a standard rule risks creating a new set of problems. Like all tools it is best applied with care.

The reason I will begin by focusing very heavily on the "keep things interesting" usage of Fail Forward is that this seems to be the motivation behind the systems that build it into their mechanics. I say this simply because mechanically-mandated consequences don't really seem to do anything to address potential dead ends.

Dice are fun

As Dan pointed out, one of the assumptions that these philosophies seem to make is that rolling dice is a value-neutral activity, and that's not really the case for a lot of people. Picking up the dice and rolling them is fun. It is a moment of tension, an opportunity for glory or for dramatic failure or for opening up a new possibility space within the game reality, and you don't know which or what will happen.

This is simply not comparable to making a decision about what will happen. If you decide it's cooler that Gnurk the Barbarian snatches the goblet from the altar at the last second before the ceiling crashes down before rolling under a closing door, that's nice, and it may well be the best way to handle a particular situation (especially if it keeps a campaign going) - but it simply doesn't have the rush of satisfaction that comes from the dice and your character's mechanically-assigned skills pulling together to create that result. A successful die roll feels like an achievement; the possibility of failure lends it the tension that gives it punch, and makes whatever steps you took to contribute to that success feel like work well done.


The second point is one I've touched on before. There are many different types of situations that a mechanic may be resolving.

  • Do I spot the clue that allows us to keep the investigation progressing smoothly?
  • Do I spot the warning sign of an imminent ambush?
  • Do I correctly draw the magic circle to protect me from the demon I'm summoning?
  • Do I unlock the door?
  • Do I shoot the ork?
  • Do I convince the guard that I'm allowed in here?
  • How fast do I complete the race?
  • How long does it take me to do the jigsaw?
  • Do I find a secret door?

Some of these situations open up new possibility spaces: the situation was A, but it is now A+B. For example, opening the lock or finding the clue creates new opportunities to do things.

Some of these situations irreversibly bifurcate the possibility space: the situation was A, but transitions to either B or C. Either you convince the guard, or you alert the guard.

Some of these situations modify your circumstances: the situation transitions from A to B anyway, but you are more or less prepared for it (you might call these B1 and B2). The imminent ambush is a good example; you will be attacked, but are you caught off guard or ready to defend yourselves?

Some of these give you information about the game world. You might discover that there is a secret door, or that there is no secret door you can find. Often, these also open up new possibility spaces.

Some of these situations establish other things about the situation. Running very fast might impress NPCs, or create rivalries, or win you a prize. Doing the jigsaw slowly might consume valuable time, or prevent you from helping allies, or you might not finish it before B happens.

Some of them have long-term consequences which it makes no sense to try and determine now. Your magic circle might bind the demon for a hundred years, or as little as a single night. You might have persuaded the NPC to join your side, or they might be planning to betray you next week. You may have memorised the crucial spell to banish Yog-Sothoth, or not. You will find out when the time comes.

And some of them are more complex. If I shoot the ork, it may die. Or it may be injured and decide to take cover. Or it may be injured and attack ferociously to get revenge. The orc being injured may leave it vulnerable to an ally's attacks. It might scare the ork's boss, who decides the fight's going badly and retreats. It might hearten a scared ally. If I miss, the orc might decide I'm easy prey. Or it might decide I can be safely ignored. Or another character may take the orc down with a spectacular headshot. This kind of situation leaves the consequences of failure very much open.

What kind of consequence anyway?

In games in general, and perhaps in whiffy* games in particular, it can seem as though the possibility is: "something cool happens, or something cool doesn't happen". It seems superficially obvious that in such a case you'd want the cool thing to happen all the time, because it's cool. Except brains don't always work like that.

* i.e. games where it's common that you just fail at things, and often that your opponent also fails at things, so several rounds may go by without anything particular happening.

If my character is shooting a gun at an NPC, then in a diceless system I would typically choose for my character to hit rather than miss*. From a purely rational perspective it doesn't make sense for me to prefer a dice-based system in which I have a chance of hitting to a diceless system in which I can select the most appropriate outcome, any more than it would make sense for me to prefer a job where every month I have a 50% chance of not getting paid.

* assuming that it was genre- and character-appropriate, since a thing being cool is often a function of those, even when it's failure. In some cases, my character emptying a revolver at point-blank range without landing a shot is appropriate and cool.

Similarly, if I'm trying to open a safe, climb a tree or hypnotise a bear, I feel like I would generally like to succeed rather than fail.

This line of reasoning ties into the Fail Forward situation, where games compel players and GMs to add "interesting" outcomes to failure, so instead of deciding between an "interesting" outcome and a "boring" one, you're deciding between two interesting outcomes.

This relies, crucially, on an assumption that not-succeeding at something is boring. I'll return to this below.

In theory this solves "the problem" (whatever exactly that is).* In practice it solves some problems for some people (for some others the problem never existed in the first place). And I think this solvedness relies heavily on interpreting reality as consisting only of a subset of the situations I listed above.

*Establishing what The Problem is, and to what extent it exists, and for whom, is frankly beyond the scope of this blog. It's something to do with making things interesting, at least.

It seems to me (both from reading and from listening to Actual Play) that the Fail Forward model tends strongly to interpret situations as forks in the road. Reality is in state A, you attempt to do X, and reality shifts either to state B or state C depending on your success. Either it thinks most situations are like this, or, possibly, it thinks they should be like this in a game narrative.

I think this is a genuine weakness in the approach, which is partly tied into its tendency to assume a conflict resolution model rather than a task resolution model. This is particularly pronounced when we bear in mind that games involve multiple players, of which more later.

And I think this is a weakness because that's demonstrably false. If I search for a secret door, and there is no door, the most natural consequence is that nothing happens. If I try to pick a lock and cannot, the most natural consequence is that nothing happens. If I try to win a race and don't do better than everyone else, the most natural consequence which is simply that somebody else wins. If I try to shoot an orc and miss, the most natural consequence is that my shot goes astray with no further effect. If I try to build a bridge and fail, it might collapse any time in the next decade.

It's not that none of these things can have more complicated consequences; of course they can. However, I think Fail Forward systems tend in these situations to push GMs to improvise immediate consequences which don't always have much to do with the actions of the character, or which at least rely on specific interpretations of how things happen which don't necessarily match what the player intended.

The "natural consequences" offer several possible advantages. The possibility of trying an alternate path from the same state, so you aren't tied completely to whatever you first tried, and can indulge in exploratory gameplay to sound out options and approaches. The opportunity for another player or character to step up with a new idea or ability, which can be especially valuable for newer or less confident players. A potentially important element of predictability that allows planning.

The following example is combat-themed, but it's worth noting that our inspiration here, Dungeon World, doesn't use full-on conflict resolution for most combats, but models them at the level of a single exchange of blows. Typically, the consequence is simply a matter of whether you cause damage and whether you take damage.

If this inconsistency on my part annoys you, feel free to mentally substitute "I sever the chains of the snarling dire boar!" for "I chop the Orc's head off!".

Do I really want an Awesome Fork?

The assumption of Fail Forward in its starkest manifestation is that when I say "I chop the Orc's head off!" I want my assertion to irrevocably set the game down one of two paths: one in which I chop the Orc's head off and it is awesome, and one in which my failure to chop the Orc's head off creates an exciting and dynamic scene.

And let's say for a moment that this is true, that I explicitly want our next forty-eight seconds of gaming time to consist of either the DM saying "Grignr's axe bites into the Orc's neck, severing its head from its body and spattering the walls in its oozing, grey-black blood!" or "Grignr's axe swings wildly, severing one of the supporting beams of the overhead scaffolding, causing piles of mouldy orc-dung to cascade onto the battlefield."

The thing is, both of these outcomes are time consuming (especially because they don't flow as easily in speech as they do in writing), and both of them centralise me specifically.

Similarly, in a non-combat situation: if I attempt to lockpick a door, the DM might say "the tumblers clatter into place, and the door swings open, exposing the Duke's secret devil-worshipping chamber!" or "distracted by the clicking of the lock, you fail to notice the watch-panther padding along the corridor, and it springs towards you with a roar!"

Player creativity is not a finite resource,* but time and things-that-need-doing are. If we are fighting an orc, and I chop its head off, I have denied you the opportunity to deal with the orc by some other method. If my failure to chop the orc's head off causes us all to be buried in orc dung, I have denied you the opportunity to enact any plans you might have concocted that relied on our not being buried in orc dung. Or simply the satisfaction of killing the orc yourself. If my failure to pick the lock doesn't simply waste a few minutes but lands the party in a fight against a panther while creating noise that attracts other guards, the fallback plans laid for this stealth mission are worthless, and if you thought your plan was better you may feel justifiably annoyed with me.

I don't entirely agree with this! I'll touch on it later.

And those are consequences which do at least tie in logically to the events underway, whereas a hurried GM can easily end up suggesting what seems like a random event. "As Grignr rushes the orc, a rusting chandelier detaches from the ceiling and plummets towards them both!" "You are unable to convince the old man to tell you anything, and you suddenly realise your pocket has been picked!" "Your attempt to forge a letter of credit is interrupted by ninjas!"

In contrast, if we were resolving this in a system without Fail Forward, both failed rolls would typically result in nothing particular happening - and this leaves things open for someone else to try.

Organic outcomes are interesting

To put it another way, in a traditional RPG system, failure actually does have interesting consequences, but a lot of the time the interesting consequence is "another player has a go at resolving the situation, often trying a very different approach."

You can make a strong case that traditional systems actually produce the Fail Forward effect in a far more natural and organic way than systems that mandate consequences, but this is only apparent in retrospect looking at the way multiple players' actions and successes interact.

For my money, Fail Forward games place rather too much emphasis on consequences of failure rather than on the consequences of player action. If the PCs are going to wind up getting thrown in gaol after an altercation with an aristocrat, I'd far rather it happened because Brenda the Barbarian started a fight after Billy the Bard failed to resolve the encounter diplomatically than because the DM decided that "you get thrown in gaol" was an interesting consequence of Billy's failed Diplomacy roll.

Generally speaking, with something as simple and inherently task-resolutiony as an attack roll, I'd probably rank my preferences as "I hit", "I miss" and "I miss and something interestingly bad happens as a consequence". It gets trickier when you think about social rolls or rolls to interpret information, but even there I would prefer "I fail and a reasonable consequence ensues" to "something interestingly bad happens with no particular connection to your actions".

Dead ends

At this point I should perhaps come back to the dead end idea I mentioned at the start, because most of this post has been about keeping things interesting. I think this is the main point where Fail Forward genuinely helps out.

It's quite possible to hit a point in a game where players simply don't know how to make the game continue. This is a common complaint of investigative games, where players either run out of leads, haven't managed to put the pieces together, or blew some rolls and locked themselves out of information. However, it can also happen with certain physical challenges or "puzzle" situations, including things like geopolitical shenanigans.

You can very easily hit a dead end. Your only way out is apparently through the locked cell door, and you can't pick it. Your only lead is this neighbour you just insulted. You've tried every configuration of this weird science device but you botched your roll to make it teleport you home. The group sits scratching their heads and looking at the GM, who is wondering why they paid no attention to half of the clues that were laid out, abandoned their door-removing equipment before entering the complex, and both shot the teleport operator and set fire to the manual. Or maybe things just didn't quite go as expected. Or the scenario is not as robust as you hoped. Or it's just been a really long day.

In these cases, Fail Forward is one perfectly good way for the GM to recover momentum. Here, it's not about adding interest to a failed roll per se, but using that circumstance as an GMing opportunity to throw in a clue or hook, or just to shake up a situation where the players were stuck and frustrated.

Just being stuck is not necessarily enough. Sometimes you may want time to sit, talk and mull things over and aren't really worried about whether you're "progressing the scenario". But often you are.

So you fail to pick the lock, which the GM rules means you're still fumbling when the door opens and slams you in the face; the guards sigh, threaten the rest of the party, and one of them drags lockpicker off to the infirmary, where they'll have a new chance to escape, overhear some news, or palm vital equipment.

You get an earful from the neighbour, and after he slams the door, a couple of local kids pop round the corner to sympathise, ask for a fag and hint that they know something - because that guy's always loud and they've overheard some of his other complaining. If he doesn't like you, you must be alright.

Your bumbling attempts at operating the device instead trigger a completely different function, and now a coolant maintenance bot is being despatched to the teleporter. Maybe it's intelligent enough to talk to, or has a telecomms function so you can call for help, or a company datalink that'll give you access to that manual you destroyed.

Content is not the same as interest

So, that "not-succeeding is boring" business. I think when Fail Forward starts to fall apart is where it assumes that "explicit consequences" are always more interesting than "no consequences" and should be mechanically mandated, and I think that's often not the case. Quite frequently, "that thing you were trying to do just doesn't happen" actually ends up being a more interesting outcome than "your attempt to do that thing has dire consequences", and for many people it is certainly more satisfying and even interesting than "that thing you were trying to do just doesn't happen and, for reasons that are game-mechanically connected but have no in-character logic, this other bad thing happens." The "consequences for all!" mindset basically seems to tie into an assumption that more stuff is better.

This is simply not true. Too much stuff clutters up a game. To take a simple example, I have played board games where every single turn involved doing several things: you always rolled a die for movement, you always moved, you always drew a card based on the square you landed on and a consequence ensued.

And generally these games were so boring they should have been wood-eating beetles.

In the case of an RPG, the problem is that you are generally trying to Do Something, and that it is surprisingly easy for complications and consequences to clog up the works. One of the advantages of failure-means-whiffing systems is that they tend to also be simply and relatively fast. If a bad roll means nothing happens, you can simply move on to the next character.

If you need to wait for the GM to think up and then narrate a consequence, and make sure you understand its implications, and then the next player to act has to reconsider their intentions in the light of a potential major change to the situation, this can really slow down play. It can also lead to layers of Things To Deal With piling up on the game, and making it increasingly difficult to follow, or to decide what to do. Constant consequences can act like a Cat's Crade, making each move seem to only tangle the characters, story and even players more deeply in a web and paralysing the game, or making it seem like a stream of consciousness.

And that's what I'll be discussing next.

1 comment:

  1. A house rule my main group uses with GURPS is a "luck roll" - roll 3d6, 6s are good luck, 1s are bad luck, a mixture is complicated. "Complicated" is always the hardest work for the GM, who has to improvise something on the spot.

    I think that at least some of fail-forward comes from attempts to replicate a cinematic feel: when a filmic hero can't get through the lock, the film "keeps things interesting" (the guards show up, he kicks the door down and hurts his foot, whatever). What he doesn't do, that some games support by implication, is try it again until he gets it right; what that gives you is a random delay. One of the Things I Always Say is that RPGs can evolve their own style of narrative rather than copying approaches taken from films or books. (And one thing GURPS can do is convert a success/failure roll into a time-taken roll - it's a 30-second lock and you failed Lockpicking by 5? It'll take you 15 minutes before you can deal with it. Now what's going to happen in that time?)

    The ambush is an interesting example of Rule of Cool breaking down: it's not immediately obvious whether it's more "cool" for the character to spot the attackers and marmalise them, or to be taken at a disadvantage but still win the fight. Depends on the character and the genre and quite possibly how the player feels that day.