Wednesday, October 14, 2020

Pass/Fail Mechanics and Dungeons and Dragons


Picture by Grandanvil on Deviantart (

Had a great session yesterday and it made me realize something about pass/fail mechanics in D&D. I see a lot of criticism of pass/fail mechanics, and a call for “fail forward” mechanics as an alternative. The take I see most often is that pass/fail mechanics are boring, old fashioned, or they are claimed to be “limiting” and they force the game into “binary” choices. They demonstrate a “lack of imagination”, or are just “bad design”. 

I’ve always wondered about that, because I’ve been using pass/fail mechanics for years and they seem to work well. Part of the problem I think is one of formality. For example, I routinely assign a set of “better worse” options to an action if it seems appropriate. So if the party fighter wants to distract the guards so they can sneak by, and throws a rock, I might assign a d8 roll as follows: 1-2 the guards ignore it, 3-4 the guards look in the other direction for a short time, 5-6 the guards look in the other direction for a long time, 7-8 the guards run over to where the rock was thrown. Obviously the options represent degrees of success, with the 7-8 being the best.

Assigning “good to bad” options for an action is something that I (and I assume most) D&D refs do off the cuff all the time, in order to capture a range of options in cases where there are no explicit rules. But that sort of thing is not “baked into” the game itself. The closest AD&D comes to this is where Gygax suggests that in a case where the rules don’t cover something you should just assign odds as you see fit. Still, he doesn’t specify using grades of success or failure, and many of the mechanics in D&D are pass/fail, so even if this is done informally, formally the rules don’t give you examples of graded success or failure. 

Newer games are more explicit about this, suggesting that rather than failure you should use a graded set of results, so extreme failure, partial failure, partial success, complete success, that sort of thing.  Stars Without Number advises this explicity, suggesting that the ref should provide a range of results attached to the roll.

So with the caveat that many D&D refs do this sort of thing informally, what are the advantages to using pass/fail mechanics in the game at all? Are they “old-fashioned”, or “bad design”, is it the case that game designers are just better now so we should abandon the binaries of pass/fail mechanics entirely? 

Maybe not just yet. 

What am I missing? Yesterday's session coalesced my thinking, so I will recount it here.

The Job
A PC had arrived in my homebrew city, and was looking to join the thieves guild. So this was a solo adventure for him. To join the guild you have to pull off a job. In the case of my home brew city, there is a lot of second story work, property theft of valuable items from the homes of prominent locals to boost your reputation, that sort of thing. So when the PC approached the Guild, they proposed the following job.

There was a merchant, Voss Halma, who had enlisted the guild to lean on one of his rivals. However, he was late paying, and was making noise that he wouldn’t pay at all, as his rival went down for other reasons, and Voss figured he shouldn’t have to pay. 

Everyone has to pay.

So the Guild wants to send him a message. Voss had picked up a statue of a tiger, 1’ long, iron core plated in silver. When guild members met him in his home he bragged about it. So the guild had a duplicate made out of glass. The job: break into Voss’ home, replace the silver tiger with the glass one, douse it in oil and burn it. The glass duplicate would crack and burn, the smoke would fill his home, and he’d know it was the guild who did it, to show that he wasn’t safe in his own home. The player agreed to the job. He was given a crude map to show where the house was, and the glass tiger to take in to the house, which he placed in his backpack with nothing else but thieves' tools and rope.

Now, at this point I have no specific rules for “breaking and entering”, I have a thief and a target with an item to steal. Thieves have skill scores, and they are all pass fail, e.g. either you pick the lock or you don’t, you climb the wall or you don’t, you move silently or you don’t, etc. However, breaking into this house was one of the most exciting sessions we have had. How can a series of pass/fail rolls be exciting? Wouldn’t ‘fail forward’ options be better? I think they would be different, but not necessarily better. Why? In short, pass/fail rolls *look* boring and reductive as they are considered in isolation. This is a mistake.

Let’s see what happened. 

The target’s house was located in a compound surrounded by a wall. First, the PC “cased” the joint. He walked around the periphery of the compound to see what was up. The wall was too high for him to see the first floor, but he could see the second floor and he saw that the west wall had no windows on the second floor. Every other wall had windows on the second floor. 

So he decided to climb the wall opposite from the side of the house without windows on the second floor, to minimize the odds of being seen. Of course, he would then have to move around to a side where there were windows in order to get in, unless the side of the house he arrived at had doors on the first floor, which he couldn’t know until he climbed the wall. In this situation there were unknowns that made it impossible to know if this was the best decision or not, but the player went with the most obvious advantage. 

Smart player.

He climbed the wall. That was his first roll, which he passed. He was at the top of the wall. Now, this particular house had wall to ceiling windows along the whole first floor west wall, and no windows on the second floor west wall. He saw that immediately on cresting the wall. 

There was also an alligator roaming the grounds. I rolled that it was on the West side of the compound right now, somewhere in the area where the PC was. I decided there was a 2 in 6 chance that the PC would notice the alligator in the darkness amongst the foliage. This was another pass fail roll. As it happens, he made the roll, so he saw the alligator moving along in the garden. I rolled to see if the alligator noticed him, and it did not. 

So I told him this, “the entire first floor west wall has windows along it, and there could be people inside looking out and seeing you at the top of the wall. There is also an alligator that just walked by. You can stay on the wall and survey the first floor to see if there is anyone inside looking out, or you can drop to the ground and hide in shadows right away to avoid being seen, but that might alert the alligator to your presence when you drop.” Again, this was all based around pass/fail rolls.

A pass fail roll would determine if there was anyone inside looking out if the PC chose to stay on top of the wall to look for people to see if he has been spotted. It would be a pass/fail roll if he dropped to the ground to see if he made noise and alerted the alligator. And if he did make it to the ground and there were people inside that could be looking out, it would be a pass fail roll to see if he could hide in shadows. 

So the player decided to drop down immediately and risk the alligator’s attention. He did so, and I rolled to see if the alligator heard. The thief made the roll and landed silently. Now I had to roll to see if the guards in the house (there were two) saw the player crest the wall. There was a small chance of this as he was only there for a moment, so I made it a 1 in 20 chance. Again, this was pass/fail, there were no fail forward conditions for the roll, but it was exciting as the player knew the jig was up if they happened to see him.

They did not. So now the PC had to hide in shadows to ensure that the guards didn’t spot him in the yard, and that the alligator didn’t either. Incidentally I rolled to see if the alligator turned around and came back this way. I rolled that it did not. Then the PC rolled to hide in shadows.

This roll was successful. So when no one noticed him he asked to look at the building in more detail and I described what was there, several different rooms, all visible through the large windows. And two guards inside. I decided they would be attentive or inattentive, just because there are guards there they won’t necessarily be doing their jobs well, so another pass/fail roll to determine that. 

I rolled “inattentive”, so I told the PC that they appeared to be playing cards, and that they did not appear to be looking out at him. So then he asked if he could spot the silver tiger in any of the rooms he could see. I told him that it was nowhere to be seen.

So then he had to decide what to do. He considered moving along the outer wall of the compound and heading towards the next wall (the south wall) of the house. He specifically said he wanted to stay against the outside wall and far from the house. As the alligator had moved on, the guards were playing cards and he was a distance from the house I told him he didn’t have to make a move silently roll in this case. The player was benefitting from his choice to stay far from the house and move around it rather than approaching, so he avoided having to make that pass/fail roll. 

And of course, if he stayed there in place the alligator might come back, I would roll for that. So even before having to make a pass/fail roll, the player has to make choices that he knows will possibly trigger a roll, and that in and of itself is part of the excitement. They know that if they are smart they can avoid certain rolls that might put them in a worse situation.

He chose to move along the outer wall to the South wall of the house, it took longer (and thus risked the alligator coming back, I rolled for that!) but he didn’t have to roll to move silently and risk that failure alerting the guards. He made it around to the portion of the compound across from the South wall of the house. He then bolted to the wall and began climbing to the second floor. That was another pass fail roll, which he made. 

Then I had to determine if the windows were locked. Yet another pass/fail roll. I told the PC, “The window may or may not be locked, I’ll roll for that”. I rolled that it was locked. He then knew that I would roll for each window, pass fail each time, and that they might ALL be locked. Knowing that this is a possibility made the choice exciting, stay and try to unlock the window while hanging on the wall, or move on to the next one and hope to get lucky while risking falling while climbing across. Again, both pass/fail rolls, but the excitement comes from weighing the odds and choosing between risky options

He decided to move on, and made a successful pass/fail roll to climb walls to the next window. That window was also locked. He contemplated trying to unlock it but felt that would be too risky, so climbed across to the corner and around to the next wall. That’s when he failed his roll, and plummeted to the ground. Fortunately he was only about 15 feet up on the wall, and rolled low damage on the fall.  

I then had to roll to see if the guards or the alligator noticed the sound of him falling. The house is in the city, and there are always noises around. So even the fail result of a pass/fail roll on climb walls has more than just a binary result. I set it up like this, 1-2 the sound was not noticed, 3-4 it was noticed but ignored, 5-6 it was noticed and investigated. I rolled a 1 for the guards and a 6 for the alligator. I then had to roll to see how far the alligator was from the thief. It was close, 40’ away, at the alligator’s movement rate that would mean 4 segments to reach the thief. 

I told him he heard movement in the garden, what would he do? He decided to climb back up the wall as the alligator couldn’t follow him there. He made this roll and climbed the wall to the next window. I rolled to see if that window was locked, and it wasn’t! Success. Now he opened the window and crawled into a hallway with three doors.

The hallway was dark but the thief is a rakasta, so he has dark vision, this means he sees heat, and there appeared to be no living things in the hallway. He crept down the hall and found his first door. He opened it and saw into a room that appeared to be a kitchen, he looked around without going in and decided to leave. That took two rounds. He left and checked out the next door which revealed what appeared to be a pantry. I didn’t give details as it was dark, but his dark vision would allow him to make out gross differences. He looked around and left, another 2 rounds.

He went further down the hall and opened a door that led to another hallway, this one with two doors. Now, as it happens I had decided that there were guards on the first floor and a small group of flightless birds that roamed the second floor. Flightless birds are common in the city setting, and they are like cats or dogs, if they see anyone wandering around the house they will squawk and make noise, drawing attention to them. I would roll to see if one showed up every turn. Another round went by going quietly down the hall.

So he entered the hallway and tried the door on the right. He specifically said he would open it gently and slowly, and if it was making too much noise he would stop. He didn’t do this for the first two doors he opened, I guess he was getting nervous the longer he was creeping around. This is exactly the situation you want as a ref, building tension associated with player actions.

This door opened up to a bedroom, the merchant was sleeping on the bed and there was a large orange lizard curled up at the end of the bed, also sleeping. This was another one of the merchant’s security features, a trained giant lizard (the size of a large dog) in his room. I rolled to see if he woke the occupants opening the door, but I gave it a 1 in 10 chance of waking the lizard, and a 1 in 12 chance of waking the merchant since the player had just opened the door and specifically said he was doing so carefully and quietly .   

They both failed to wake. Then the player decided to stay out of the room but look through the crack opening to see if the silver tiger was in the room. Opening the door slowly and quietly, and then thoroughly surveying the room from a distance in the dark took 5 rounds. 

He then decided to check out the next door further down the hall. He took similar precautions in terms of opening quietly, and that took another round, so I had to roll to see if one of the flightless birds showed up. It did! Now, the player was far down the hallway in the dark, and the bird poked its head into the hallway through the door he had opened.

So I had to decide how to handle this. First I made a roll to see if the bird surprised the thief, he might not notice the bird. In this case the bird did not surprise the thief. However, I rolled that the thief surprised the bird. So he could do what he wanted, enter the door to hide, run to the bird, that sort of thing. The hallway was dark and he was near the end of it, about 40’ from the bird, so the PC decided to try a hide in shadows roll to avoid being seen by the bird. It was a tortured choice! He knew that if the bird started to make noise he was sunk, stuck on the second floor with 2 guards in the building, an alligator outside and a giant lizard in the next room. 

He was successful at the hide in shadows roll, and the bird left. 

So now he tried the door, and it was locked! There was an immediate hue and cry in the room as they knew the clock was ticking, every moment that he was there made it more likely he would be found. But he decided that the locked door was a good sign he was in the right place, and he took out his lockpicking tools. Now, as it happens, the PC had picked up a pair of eyes of minute seeing recently (his only magic item) and he used the eyes to check the locking mechanism for traps, and to see how it worked to try and pick it. I assigned a bonus to both his find/remove traps roll (there was no trap as it happened) and his pick locks roll. He made the roll, opening the door. I told him that this process might wake up the merchant in the next room, or the lizard. It was a small chance, but just mentioning it created a stir, every time I had to roll it was exciting as it could change everything with one roll!

I rolled, no one woke. He looked into the room. It was a “trophy” room, filled with paintings, elegant rugs and sculptures of various kinds, one of them being a silver tiger!

So that’s where we stopped for the session. He has found his target, and so far evaded detection. Now he has to figure out how to get to the statue, if the room has traps or animals/monsters in it, and get out before being spotted. 

You don’t need fail forward mechanics to make this scenario exciting, what makes it interesting is a few things:

A) The way known choices interact, take the first example, if the PC stays on the wall to survey the first floor they will be more likely to know if there is someone there, but they also increase their odds of being spotted. If they drop to the ground they have less chance of being seen, but won’t know if there is anyone looking out of the first floor windows right away, and of course the alligator might hear them land. If the thief stays up then the alligator won’t notice as it didn’t notice him when he crested the wall in the first place and wouldn’t be likely to look up, but then they might get spotted on the wall. 

This situation is exciting as the nexus of choices doesn’t produce an immediately obvious best option, all are risky, but differently. The PC has a good sense of what the consequences are in this case, it isn’t a case of ignorance, but the choices are not clear as all options have risks. So it’s rarely just “pass/fail and you are done”, it’s “choose which of the pass/fail options you want then deal with the results when no one option is obviously better”. That’s exciting, and draws on player skill at making decisions, not pure luck. 

B) The potentially unknown consequences of the choices, depending on which choice is made, a different set of possibilities (and rolls) can be triggered. Since those sometimes can’t be predicted, they are exciting, and produce a degree of uncertainty that builds tension. Making the wrong choice can lead to having even worse choices to make. This is one of the things that is most often forgotten in these discussions, the ref knows what the consequences of the pass/fail roll are in many cases (e.g. in terms of what other rolls will be triggered), but often the player does not. So what looks like a simple binary option and seems quite uninteresting when discussing game mechanics is exciting for the player as they don’t know what the results will be if they fail.

Even though each individual roll is pass/fail, the result of each roll changes the situation and presents a new set of options, some of which are unknown at the time the player has to make the pass/fail rolls. Within the roll it’s pass/fail, but the uncertainty fuels the excitement.

C) The immediacy of the choices, because these rolls are pass/fail, they have immediate consequences that drive the action forward, EVERY ROLL is exciting as every roll could produce a fail result that could be catastrophic, or a pass result that could be life saving, each roll has immediate risk and reward. With fail forward mechanics one of the side effects is that each roll is less definitive as the results aren’t binary, they can lead to degrees of failure or success, putting off the very bad or very good results.

D) The possibility of “runs”, sometimes the player rolls badly, a lot, and a situation that is mildly dangerous becomes significantly dire due to how those consecutive bad results accumulate. That can be exciting, the player can see how they are spiraling to a bad result, and have to try and figure out how to stop it. 

Conversely, when there is a series of successful rolls it can be beautiful, the PC pulling off some crazy stuff because the dice. They feel like they are “on a roll” and every new roll is exciting as they want to continue with their success. Conversely, consecutive rolls like this can create excitement when there is a string of successes because the player intuitively knows that this can’t last forever, sooner or later you will fail your roll, and the consequences will be dire. That is also something that produces anticipation and excitement. I believe that part of this excitement is that the pass/fail option of each step means that a serious consequence or reward is waiting with every single roll. 

E) Mimicking degrees of failure, when people focus on individual rolls they forget how the rolls are rarely one and done. In most cases, the result of a roll triggers a situation that leads to additional rolls. The set of rolls together often mimic an individual roll with “fail forward” conditions, but instead of one roll where you have a set of options from better to worse, you get a set of rolls where the player gets to make individual decisions along the way and thus has more control over what is happening. Essentially a sequence of pass/fail rolls mimics the individual roll with degrees of success, but it does so organically as each step in the process changes the circumstances.

F) The ability to avoid rolls / weigh choices, player decisions can lead to the mitigation of rolls, so even if you are using a pass/fail mechanic, player ingenuity can lead to avoidance of rolls entirely, which is challenging and fun. Also, the player is often presented with choices between actions that have different costs, for example when the PC had to decide between trying to unlock a window while hanging on the wall and moving to another window to see if it was unlocked. Both odds require simple pass fail rolls, but have different risks and rewards. Weighing those risks and rewards is part of the fun. If you wrap all of the degrees of success/failure into a graded fail forward kind of roll then you are giving the player fewer opportunities to shape the results with their decisions.

G) The consequences of pass/fail are up to the referee, there are cases where the result is pass/fail, e.g. you are noticed by the guards or you are not, but the consequences are graded by the referee. So what do the guards DO if they notice the thief? The fact that the move silently roll is pass/fail doesn’t mean that the consequences need be binary. In fact, they may require a further dice roll to resolve, or the ref may assign consequences that are on the scale of success. Take the case where the PC failed their pass/fail move silently roll but they are not visible to the target, in that case the response to hearing a noise isn’t obvious. I’ve had many cases where players have suggested that a result of a pass/fail roll might lead to another roll or a new condition I hadn’t foreseen. 

Fail forward mechanics are not bad game design. They are good game design. They take a situation with a binary result and open it up to other possibilities. Those possibilities can be negotiated by the ref and the players, the players can even decide on the options to make it more interesting and collaborative. 

But the use of pass/fail mechanics isn’t old fashioned or boring or binary. The individual roll is binary, but when you put it in the context of the choices made by the player and the evolving situation created by the results of those choices, when you see the individual roll as part of risk/reward management, the mix of known and unknown consequences, and a larger series of actions and rolls that make up the actions of the session, you can see that they aren’t boring or binary, they are nodes in a complex, multifaceted and evolving process. 

The players LOVE this process (or at least my players do) as it has risk at every step, but the choices made by the player can mitigate or even avoid those risks entirely, which makes the process engaging and exciting. The burden on the ref with pass/fail mechanics is to use them fairly, estimate odds fairly, do a good job determining how each choice changes the situation and determining what the next rolls will be. The job of the player is to try and avoid having to make these rolls, and when they do make rolls hopefully end up having to make rolls where the odds are in their favor, or the consequences are less dire. 

I find this process to be significantly interactive, challenging, exciting and engrossing. An hour of our session yesterday was dedicated to this solo adventure, I checked in with the other players repeatedly to see if they were enjoying it, as they were spectators. I offered to cut away and run the session for the other players, cutting back and forth so they would have something to do. 

They adamantly refused, watching the thief navigate his choices, watching him roll each time to see if things were going to go south, they were riveted by the process. 

So by all means use fail forward mechanics, they are fun and interactive and are part of exciting games. But don’t reject pass/fail mechanics as they are “old fashioned” or “bad design”. They are part of a system that rewards player creativity and problem solving, and can lead to surprises for both the ref and the players.

Game on!


  1. Hey friend!

    I think you have misunderstood "Fail Forward" mechanics, mostly because in your critique of them, you describe gameplay that is ABSOLUTELY using fail forward mechanics all of the time.

    Fail Forward Mechanics is this: Failing a roll does not stop gameplay.

    Failing the roll does not stop gameplay, instead it opens up new scenarios to play within.

    It was created to clarify that people should do WHAT YOU ARE DOING HERE, and to help teach people how to do those things. The example it is supposed to stop is this.

    DM: There is a locked door
    Player: I try to unlock it
    DM: Roll lock pick
    Player: 12
    DM: You fail to unlock the door.
    Player: can I try again?
    DM: uh, sure? I guess? I mean, I haven't come up with stakes, or any consequence of failure, so sure?
    Player: I rolled an 8
    DM: Uh, I guess you fail again. DO you wanna try again?
    Player: Sure. 24?
    DM: Ok you make it through.

    The door in this example is an example of the play stopping. There is no consequence for failure, there is no potential for the story to GO anywhere. Fail forward mechanics instead suggests that, when you fail to pick the lock, the guards route progresses and they have a chance of discovering you, or whatever.

    So literally EVERY example you have in your demonstration is.... an example of Fail Forward.

    If, on the other hand, you are speaking about Graduated Tests (which you may be?), then you may have a point. NOT ALL TESTS NEED TO BE GRADUATED TESTS. But I don't think anyone is really arguing that they have to be?

    Check out writings on Fail Forward mechanics, you'll see that this is what you are doing. The principle is just that a failed roll does not stop the action, but does change the situation.

  2. I was trying to say that in entry E) Mimicking degrees of failure. I think that what I am doing is essentially fail forward, but instead of it being with one roll, its a series of rolls. But its the same idea, the game doesn't stop. The reason I hesitate to use the terminology is that I think of this as a 1e thing, not a new thing. I have always played with DMs who play like this. They make the players roll for things and either with graded results or the aggregation of many different rolls is the map of those graded results. I have never met someone who said, "OK, you failed picking the lock, you just have to keep trying". Fail forward mechanics seem to just foreground that stuff into the design so you have accounted for it. Doing what I do and what many old school DMs do is just to assign odds to each step the PC takes (where odds aren't given by the game, e.g. a to hit roll) and let PCs choices in respose to each step define the graded result. In short, I agree, I think you can do "graded tests" in different ways, either with an individual roll with a really bad, bad, neutral, good and really good result for example, or with indiviual aggregate pass fail rolls. Both work. I only mention any of this as I regularly see people saying that D&D is a "binary" game with "either/or" mechanics, when it is more than that. How the binary mechanics interact also determine how the game is played. Thanks for commenting.

    1. Hey my friend!

      The problem is that the article you have written sounds like a condemnation of something that you then describe. The fact that someone said "Hey, we should NAME this thing that we've been doing, call it 'fail forward' or something" is not a reason to condemn the practice.

      Here is your opening sections primary point "I see a lot of criticism of pass/fail mechanics, and a call for “fail forward” mechanics as an alternative. The take I see most often is that pass/fail mechanics are boring, old fashioned, or they are claimed to be “limiting” and they force the game into “binary” choices. They demonstrate a “lack of imagination”, or are just “bad design”. "

      Ok, sounds good. NOW SHOW ME who is saying that, in what context, and then critique THAT. You are arguing against a mythical bogie-man and it makes the whole discussion beneath it a little jarring/weird. "Fail Forward Mechanics are bad! Here's me doing fail forward mechanics and having a great time as a critique of fail forward mechanics" is not a sensical argument.


Randomization - It’s Not What you Think! After seeing a number of tweets on monster HP and such, I think it’s time for a discussion of rando...