The Colors of Magic - Available Now

December 30, 2015

Fail Forward

You may have heard of the term "fail forward" used in RPGs.  In the business self-help world, the concept means "failing because you took a risk and it didn't work" as opposed to "failing because you did not take a risk."  It's meant to urge people to take risks, and remind people that successful entrepreneurs are always failing because they take more risks than typical businessmen.

The term was adapted to RPGs because it sounds good.  This is a very bad reason to pick a term.  Worse, if you google "fail forward" you find a lot of websites full of business jargon.  What's a confused GM to do?

Let's start with an RPG definition of Fail Forward.

When people talk about Fail Forward in RPGs, they mean that failure should not stop the action, and failure should always have interesting consequences.

I suggest that we stop saying "fail forward" now, because it's confusing, it's business jargon, and googling it finds all the wrong links.  I don't need to make up yet another term to replace it.  Instead, I suggest we just start using the term for it from Fate Core, "succeed at a cost."

(If you're really wedded to the term "fail forward" just use find-and-replace.)

Why should I use the "succeed at a cost" technique?  

Every time the dice come out, there are two possibilities.  Things might go the way the PC wanted, or they might not.  Degrees of success, critical hits, botches, and other rules are just degrees of those two possibilities.  Duh.

So when that roll comes up a failure, you want it to have interesting consequences, but you can't have those consequences stop the action.

Example:  Imagine you're running a Vampire: the Masquerade game, and the Nosferatu, Sai, is searching for information on a bizarre Greek translation of the Book of Nod called the Gennimata Annotations by calling his academic contacts.  
GM ("Storyteller" in Vampire):  OK, give me a Charisma + Etiquette check to get them to open up about such a dangerous book.  You can add your Contacts to the roll, but I'm raising the difficulty to 9 because the Book of Nod, especially the Gennimata Annotations, terrifies mortals.  
Sarah (Sai):  8, 8, 4, 3, 2, 2, 2.  Fail.

So what do you do?  It seems like the stakes for that roll were "track down the Gennimata Annotations or fail to do so."  Just because the dice failed to roll high doesn't mean the character failed to achieve his goal.  A bad die roll just means the character performs poorly; not that the character just stops.

Think about it:  Let's say you're calling around looking for a copy of the hot new indie RPG.  You call several game stores and check Amazon, but everyone is out of stock.  Do you just give up?

Well, maybe.  It's just a game.  You can wait and see if they restock later.  And in a boring story, Sai would just give up, too, because finding the Gennimata Annotations wasn't really that important.

But it is that important or else you wouldn't have a plot about it!

So what's the GM to do?

Traditionally, here's what happens.

GM:  Nobody Sai knows can tell him where to find a Gennimata.
Greg (Galdos the Tremere):  OK.  Well, let me check my occult connections.  I know a bunch of thelema temples.  Maybe one of them will have a line on a Gennimata.
GM:  Sure, make a Charisma+Etiquette check.
Greg:  Cool, one success.
GM:  OK, they know there's a guy who has a copy, Professor Helmut Knecht.  He acquired it in the 80s and has rejected all offers to buy it.

There are more disadvantages than advantages to doing it this way.  The main advantage is that more than one player got involved in the scene.   The disadvantage is that the pace slowed and the table wasted time.  This is the "inevitable success shuffle."  If everyone gets to roll something until someone succeeds, success is inevitable because failure means the game is over.  Too many RPG investigations work like that.

You may be more familiar with the D&D version of the shuffle.

Rogue:  The old monk said there was a secret door in the narthex of the old cathedral.  I search for secret doors.  16.
GM:  You find no secret doors.
Fighter:  I see her searching and join in.  18.
GM:  You find no secret doors.
Wizard:  I attempt to Aid Another.  8.
GM:  No good.
Cleric:  I guess I'd better help search too.  21.
GM:  At the base of a column, you notice a geometric pattern.  When you press one of the triangles, the column sinks slowly into the floor, old masonry, dirt, and dust falling away after it.  The mechanism must be hydraulic, as you notice the cracked fountain on the East side of the room gurgling and spurting black, fetid water all over the floor.

What a waste of time!

At least in our vampire example, the Nosferatu and Tremere were engaged in slightly different activities, highlighting their characters' roles and resources.  In a way, that's not so bad.  But what if the Tremere failed, too?  How long would the table spend just trying to get the next clue?

System note:  Gumshoe system games make it impossible to fail to move the game forward on an investigation action.  If the players are seeking information, as long as they have the skill, they automatically succeed.  But the GM can still make "succeed at cost" happen.  Let's say you're playing a Bookhounds of London Gumshoe game and the players are searching for the Gennimata Annotations.  One player says that they have Research, so they can find out who last acquired a copy.  The GM has to give that player a clue to move the game forward, but they can add a complication:  The Gennimata Annotations can shatter minds.  You can track down who last acquired a copy, but if you don't give me a Reassurance or High Society spend, you'll leave the collectors who know about it gossiping about you behind your back...

In the D&D example, there is absolutely no reason to keep rolling checks.  Eventually the party would find the secret door.  If they all failed, what would the DM do?  What would the players do?

I'll tell you what they'd do.  They'd go back to town and bring that old monk.  And if that didn't work?  They'd hire henchmen.  More wasted time!  Statistically, they're eventually going to succeed at the check.  That's why it's the "inevitable success shuffle" - so it seems pointless to call for a roll at all.

Or is it?

With "succeed at a cost," we can still have stakes for a die roll, but "halt the action" doesn't have to be the failure condition.  There are other ways to screw up, after all.

How do I use "succeed at a cost"?

There are two ways to use "succeed at a cost" depending on when you decide to implement it.  If you decide to set the stakes for the die roll ahead of the action, you can use "succeed at cost" instead of "failure" as your stakes.  Otherwise, you just have to describe failures in ways that change the situation and don't hold the game back.

Just think up how things could go wrong for the PC that don't necessarily involve failing to move the game forward.  Here's how we'd do it with our two examples.

GM ("Storyteller" in Vampire):  OK, give me a Charisma + Etiquette check to get them to open up about such a dangerous book.  You can add your Contacts to the roll, but I'm raising the difficulty to 9 because the Book of Nod, especially the Gennimata Annotations, terrifies mortals.  If you fail, you'll lose one of your contacts for a while.
Sarah (Sai):  8, 8, 4, 3, 2, 2, 2.  Fail.
GM:  You learn that it passed through one of your contacts' hands in the 80s.  At first she acts like she doesn't know what you're talking about.  But with some prodding, you unlock her repressed memories of the horrible thing.  It's basically a book of living nightmares.  She only saw a few pages, but that was enough to traumatize her mortal mind.  The words come out along with the tears.  So many tears...  Your Contacts goes down by 1 for a month, but you learn that she acquired the book for a Professor named Helmut Knecht in the 80s.  

Not only is the consequence for failure harsher (loss of a Background point for a month), but this way the GM has an opportunity to accelerate the pace.  This description of the Gennimata Annotations drives home how awful the book is.  What kind of professor would buy such a thing?  What's he been doing it with it for a decade?

Here's the D&D example:

Rogue:  The old monk said there was a secret door in the narthex of the old cathedral.  I search for secret doors.  
GM:  OK hold on.  You're searching a crumbling cathedral for the entrance to the dungeon for tonight's game.  You're going to find it.  But if your roll doesn't come up 20 or better, it takes you all day, and you'll be going down into the dungeon in the dead of night.  You'll be rolling for the whole group.  Take a +2 to represent their help.
Rogue:  Ah crud.  18.
GM:  Hours after twilight, you've burned through six torches and still nothing.  In your despair, you slump against a column and hear a loud THUNK!  You must have hit a hidden switch by accident! The column sinks slowly into the floor, old masonry, dirt, and dust falling away after it in the dark.  The mechanism must be hydraulic, as you notice the cracked fountain on the East side of the room that Fighter was examining starts gurgling and spurting black, fetid water all over the floor.

Instead of the consequence for failure being wasted table time, the GM has decided to make the consequence for failure be wasted game world time.  Obviously both are "bad," but wasted table time is bad for the whole game while wasted game world time is bad only for the characters.  For the players and GM, it adds to the sense of urgency and danger of exploring the ancient dungeon.  So it's good for the game.  (Remember the fun formula.)

But failure still happens, right?

Sure.  Sometimes failure itself is interesting and drives the game forward.  When narrating failure, don't narrate a "nothing happens" failure.  That always leads to the "inevitable success shuffle."  And that's dumb.  Instead, make the consequence of the failure itself move the game forward.

(This is why people latched on to the term "fail forward" - it's a failure that still moves the game forward. If that term was not already taken by business jargon, it would be appropriate.  But it is, so we really shouldn't re-use it.)

Consider failing to disarm a trap, setting it off, and breaking your thieves' tools.  That's cool!  Consider pleading to the proud Baron, only to make him angry and exile you.  That's an interesting twist!  Consider trying to intimidate a crooked cop, only to have him draw his gun on you - that ratchets up the tension!  Consider trying to talk a spy into revealing information, only to have him demand an exorbitant price for it - ouch, that smarts!

Here are a few ways to make failure interesting:

  1. Add a game complication (broken thieves' tools): Game complications can be as sweeping as changes to the game itself, or as simple as losing a piece of equipment (or the lost Contacts point in the Vampire example, above).
  2. Add a story complication (exiled by the Baron):  Introduce a new obstacle that either needs to be dealt with right now, or could be a serious problem in the future.  The lost time in the D&D example, above, is a story complication.(By the way, the best story complications connect to the players' character hooks.)  
  3. Raise the stakes (crooked cop draws his gun):  Make the consequences of future failures even worse.
  4. Charge for success:  Give the PCs the choice to fail unless they pay something that the game makes it hard to get back.  "Your contact won't talk unless you give her one of your healing potions."
Notice how none of these consequences are boring, and none of them allow your players to engage in the "inevitable success shuffle."  

In each example, there's a bad way to handle failure that is quick, simple, obvious...   and wrong:  You fail to disarm the trap; you fail to persuade the Baron; you fail to intimidate the crooked cop; you fail to get the contact to reveal his information.  

You can even put this on your GM screen to remind you of your options when you run a failed check or set the stakes ahead of a roll:

  Succeed at a cost
  Game complication
  Story complication
  Raise the stakes
  Charge for success

Remember that "rolling to succeed" implies "...and to avoid a consequence."  If it's not clear if there is a consequence, you're thinking about it wrong.  Failing to climb the wall doesn't mean you simply walk up to the wall, grab a rock, strain, slip, and shrug your shoulders.  That's not how humans work.  They don't give up that easily, and nothing is ever that simple.  Failing to climb a wall means...

  • You climbed the wall, but twisted your knee, had some hard slips and falls, and cut your hand for a total of 1d6 damage.  (Succeed at a cost)
  • You tried to climb the wall, but you put too much weight on a lower handhold and broke it off when you slipped.  Now anyone trying to climb the wall has a -1 penalty.  (Game complication)
  • You tried to climb the wall, but fell noisily.  Now the guards probably know you're here.  (Story complication)
  • You tried to climb the wall for five minutes, with no success.  Now you're running out of time and getting nowhere.  (Raise the stakes)
  • You can't figure out a way to get up this wall without leaving the rope and pitons behind.  (Charge for success)

Never just say "you fail to climb the wall".  That's not failure.  That's a waste of everyone's time.

"Nothing" is not a consequence of failure.  

It's literally what happens when the GM isn't doing their job.  

If you want to make "nothing" happen, just sit there and play on your phone.  

Your job is to make the world react to the players' actions.  

"Nothing" is not a reaction.  

Do your job!


  1. Do you have any experience with the Star Wars and Genesys system from Fantasy Flight Games? I love the blog, and want to apply the content to what my friends are playing these days.

    1. I do! I ran the Edge of the Empire starter set and it was a lot of fun. You can certainly apply a "success at a cost" approach, because the GM still has total control over check framing. What I mean by that is you can use this sort of check frame:

      Example: The party, a bunch of galaxy's most wanted rebel scum, have put on disguises to sneak into an Imperial Ball.

      "Make a Deception check to get the storm troopers to let you through without making them suspicious."

      That check frame has "success at a cost" built into it: If they fail the check, they let you through, but they become suspicious - they might tell other storm troopers to keep an eye an you or tail you or double check your ship registration at the stardock. FFG's system has neat "and also!" mechanics with every roll that make success at a cost check frames even more exciting!

      Fail: "You make the stormtroopers at the gate suspicious. One steps aside to radio in the 'anomaly' of your presence."

      Advantage: "...but you get a look at the guest list when he does, so you get a bonus on your next attempt to avoid suspicion."

      Triumph: "...but when he suggests someone check your ship registration, he gives the wrong name, which will buy you some time."

      Threat: "...And the other one starts following you, giving you a penalty to your next check to keep your cover."

      Despair: "Roger. Let me see... The port of origin has no record of a light freighter with the designation, 'Heart of Gold.' I'm sure it's just a mistake, but we're gonna lock it down until we clear it up."

  2. The original context I learned it in (I believe Dungeon World?) gave the example of a thief successfully stealing the king's diary then running into trouble on the way out.

    This is different from "success at a cost". There's no additional obstacle or spent resource to get the diary. They just straight-up succeed at their intended action, with the rolled failure pushed forward to a separate event.

    This forward failure can be immediate, but it doesn't have to be. Especially if you're thinking offscreen. I've had a couple times when it made narrative sense for a miss to hang over the table for half a session before the consequence dropped.