Running an RPG really is an art. Primarily because rulebooks almost never even mention how you do that. The first edition of D&D was a collection of notes that an apprentice DM would get after he had been trained by an older master and somehow everyone kept doing that ever since. Two page Examples of Play are garbage. The only exception I know is the 1983 Basic Set for D&D. And that is after four decades and hundreds of games. Other than becoming an apprentice of the GM of the group you play with (simply observing a good GM at work won’t do it), the only people who can teach you to be a good GM are old GMs who explain it on the internet. The Angry GM and Matt Colville are both super helpful and there’s a lot of equally great stuff by The Alexandrian. (Links to all three are to the right.) I think I was a pretty poor GM myself until I started doing more research three years ago. Practice alone doesn’t make you any better if you just keep making the same mistakes every time. But all the methods to use as a GM are only options and not all of them work for every purpose. You need to have some reasonably clear image of what you want to do before you can pick the methods that work for you.
This post is basically my current playbook on how I am running games.
When in doubt, yes
Any time a player has an idea and you say no, the game stalls for a moment. Whatever vague plan was forming in the player’s mind will have to be discarded and a completely new approach considered. Saying yes keeps the game moving forward. Generally you want the game to maintain forward momentum and avoid situations in which players are overthinking things. For that you want to encourage and applaud their crazy ideas, not create the impression that they have to figure out what you think the correct solution should be. Even when the question is something like “Is the bridge made of wood?” and you have not thought about it before, answering yes will probably lead to the most interesting result. Whatever it is that the player is thinking about, it’s probably something that will only work if you say yes.
What are the players trying to do?
This is probably the most important thing that a GM has to understand. Before you decide the outcome of an action it is crucial to confirm that you and the player are on th same page. If players say they want to do something that seems nonsensical, it’s almost always because their picture of the situation is different from what you meant to describe to them. In an RPG the GM is the eyes and ears of the players. They can always only know what you tell them. If they get something wrong it’s your fault. And that means it’s your duty to make sure that you and the players are imagining the same situation and context.
Only when you know what the intended outcome of an action is can you decide what the best method is to handle it.
Do we need a roll?
Many rules heavy games have a mechanic for almost everything players might want to do. And that leads to a stong tendency to use mechanics for everything. But most of the time doing so is nonsensical. Most actions do not require a die roll. When it seems obvious that an action should succeed it should succeed automatically without a roll. If an action is effectively impossible, no roll is needed either. Dice are rolled when there’s a clear chance for both success and failure and failure also has a consequence. If the character could just keep trying until he rolls a 20 or a 01 then the question is not whether he can do it but how long it will take him. If there is no consequence of failure, no roll is needed. The most common forms of consequence are taking damage, alerting enemies, time pressure and anything that will make it impossible to try again. When player roll dice it should be a meaningful moment, not just tedious routine.
Sometimes only the first attempt wil have a consequence but successive attempts will not. For example a character kicking in a door could surprise the people on the other side. But only on the first attempt. If the first attempt fails anyone on the other side will know someone is trying to break the door and be alerted. A second or third failed attempt won’t have any further consequences and need no additional checks. It’s enough to tell the player that it didn’t work the first time and ask if he wants to keep doing it until it works or stop.
Only roll again if something has changed
Generally every action should require only a single roll. If a character disguises himself as a guard and fools the guards at the gate, no additional rolls are needed to fool the guards in the courtyard. That’s still the same situation of wandering around without being noticed. If the character would attempt to enter a restricted area or take a prisoner from the cells then he is doing something that would make people take another closer look at him and a new check might be required. If you would check for every guard the character walks by it would be only a matter of time until a roll fails and the whole plan of getting inside disguised as a guard becomes impossible.
What does the other side want?
Any time an encountet happens, whether it’s planned or random, the first thing the GM has to do is take a moment and decide what the NPCs or monsters want. It’s almost never to kill the PCs at any price. They might want to rob or eat them, get them to leave, prevent them from getting past, and possibly a wide range of other things. And usually, above all, they want to stay alive. To effectively play the people and creatures the party encounters, whether it leads to a fight or not, you have to know what they want to get out of the encounter. What tactics they use, when they might retreat or surrender, whether they attack first or need to be provoked to fight, and how they might negotiate all depends on what they want from the PCs.
Enemies in videogames are stupid and all they can do is running straight at the characters and start hitting until they are dead. Henchmen in movies tend to do just the same. And most RPGs seem to assume the same thing of NPCs and monsters. But that’s a complete waste of interesting interactions and that’s what RPGs should be about, not combat. In the heat of the moment it’s often difficult to come up with interesting ways in which creatures react to spotting the party and so GMs usually default to instant attack. The reaction roll is a very simple and extremely useful mechanic to address that. If the reaction of a creature is not automatically apparent, roll a dice to see what it does. In B/X D&D immediate attack is a very rare result. Much more commonly they are hostile and waiting for a good opportunity to attack or just chase the party away, or they might wish to avoid a fight or even be friendly. The important part to remember when using reaction rolls is to question whether the reaction would really be automatically apparent. It’s a matter of tone for the campaign, but just because something is said to be evil or occasionally eats people doesn’t mean it has to attack. When in doubt, make a reaction roll.
Retreat and Morale Checks
Almost all people and creatures encountered by the party want to not be killed above everything else. When it becomes apparent that a fight is lost, have the enemies attempt to save their lives in whatever way they can. They might retreat, surrender, or try to negotiate.
But even when a fight can still be won enemies might get too scared by the killing around them and flee in panic anyway. This can be handled by a morale check. 2d6 are rolled the first time someone in the fight is killed and again when the enemy group has lost half of its fighters, either to death or being incapacitated. If the roll exceeds their morale score they will flee, regardless of the tactical situation.
In OSR games defeating an enemy in battle gives characters only very little XP compared to the considerable risk they pose and the resources it takes to defeat them. They are a pure nuisance and something to be avoided. Since random encounter rolls are made based on the time the party spends in a dungeon or enemy stronghold, the only way to avoid them is to be quick and don’t spend any unnecessary time in the place. Random encounter checks are also made any time the players do something very noisy that might attract attention, which is an incentive to be sneaky. Random encounters really only work if the players are not meaningfully rewarded by getting into fights. If a random encounter provides a good amount of XP and treasure then they defeat the purpose.
Most encumbrance system are way too complex to be fun and slow things down so much that they are just completely ignored. A much simpler system is to have every item have a weight of 1 and let characters carry as many items as their Strength score unencumbred and twice as many with being slowed down to half speed. Exceptionally big items like armor count as two or three items, very small items like keys or papers count as having no weight at all.
Going from full speed immediately to half speed seems very unrealistic but is by far the easiest solution to deal with turns on large maps. Having a 3/4 step in between would make it all much more unwieldy.
Turns and Zones
Most OSR games use turns of 10 minutes to track the durations of torches, spells, and potions and check for wandering monster encounters. Usually this is tracked by how much distance of empty tunnels the party has covered. This requires having the whole environment map out at great detail, which I am not fond of. Instead, for larger locations with a lot of empty space, I break the whole sketchy map down into zones of not clearly specified size and each time the party leaves a zone for another a turn has ended. If the party moves at encumbred speed, two moves have ended.
In addition, the end of a fight also means the end of a turn, regardless of how long or short it is. Tidying up after a fight tends to take longer than the fighting itself and for the sake of simplicity it’s assumed that another turn has ended once the party continues its exploration. A turn is only approximately 10 minutes, not a precise unit of time. Some turns are longer, some are shorter, and torches, potions, and spells don’t run on an exact timer either. It’s just a simple approximation.
No NPC or location is irreplaceable
Even though I was born in the mid-80s I am very oldschool in my approach that RPGs area about the players making their story, not about the GMs telling their stories. It has nothing to do with nostalgia or snobbery, I just found it to be a vastly superior use of the great potential of roleplaying games once I really understood the idea. Not understanding this is why my campaigns used to be bland and underwhelming.
Whatever you put into your campaign is there for the players to interact with. It’s not to show off your creative work. If the players decide to do something that leaves an NPC dead or a cool place destroyed then let it happen. If you are not willing to see it go up in flames, don’t put it in the campaign.
Don’t prep plots
The adventure is a story the players are making, not a story they are being told. A good adventure provides things the players can interact with in interesting ways. And GMs don’t know how the players will interact with the things they’ll find so you shouldn’t assume that they will do certain things and then prepare additional scenes based on that. A good adventure is an environment with multiple people who want and do various things. It’s wise to spend some thought on what would happen if the players don’t interfere but don’t write a sequence of scenes that are meant to happen. When you do that the players will make their decisions based less on what they would like to do but more on what they think they are supposed to do.
Yes, almost all adventures that are being released are scripts of scenes. And I also think most released adventuresare bad.
Always have alternative routes
Probably the biggest mistake that is constantly made in dungeon design is to put obstacles in the way of the party that have to be overcome or the adventure can not continue. In a well designed adventure absolutely everything needs to be optional. And the best way to do that is to always have more than just a single possible route to get to an important place. Failure to do that leads to a long list of really bad problems. If the adventure can only continue if a creature guarding a gate is defeated, then the players know that they can not lose a fight against it. It will be easy enough to defeat and if things go wrong the GM will help them out and make them win anyway. From this follows that everything they encounter will be defeatable without much trouble, which leads to the conclusion that everything is meant to be defeated. And just like that the game turns into a meat grinder where combat is always the default response to encountering a monster or enemy.
Add plenty of optional content
Players aren’t really into exploring. Players are into discovering. So add plenty of hidden stuff to adventures that can be discovered through the actions of the players. They will probably miss a good deal of them so prepare them in a way that takes that into account.
Don’t make important things hinge on a single die roll.
If you want something to be found for certain then make it impossible to miss. It’s not common but still happening all the time that very substantial parts of adventures will only be accessible if the players find a hidden thing, make a saving throw, or succeed on a skill check. When you include such thing, always ask yourself what happens if the players fail.
Make battlefields interesting
Some games have a lot of rules for battles and others have very little. Even if you have a complex battle system it might seem fun to just rely on the standard maneuvers but every fight will be several times better if it takes place in an interesting battle field that both sides can use to their advantage. Add rubble, pits, fences, pillars, tables, stairs, balconies, barrels with oil, giant spider webs, huge furnaces, and chandeliers to swing on. Nothing is more boring than a featureless square room.
In a campaign in which characters get XP primarily from retrieving treasure, my rule of thumb is to take all the possible XP from fighting opponents in a location and multiplying it by 10 to get the value for treasures that can be found in that place. A good amount of it will be in the lairs of creatures but most of it well hidden and easy to miss. In the long run this should lead to a ratio of about 1:6. Treasure is put where it makes sense. In a location with mostly animals, most treasure would be in hidden staches while in an area with many humanoids much of the treasure would be in their lairs.
To have an easier time with encumbrance each treasure item takes up one slot of inventory space and has a value of either 100, 1,000, or 10,000 XP.
Thief Skills (for OSR games)
Thief skills in OSR games are always a weird thing. Thieves have a skill to hide and to sneak but what about other characters who try to quietly move around unseen? And the chance to successfully pull it of is a measly 15% for first level thieves. With less then an 80% chance it’s not worth to even try. Sadly, D&D never explained itself on how this is supposed to work and all we have are various interpretations on what those skills and odds are actually meant to be for.
Stealth (for B/X)
The actual names for hiding and sneaking in D&D are Hide in Shadows and Move Silenty. This means hiding when there’s nothing to hide behind and moving without making any sound that could be detected. These are things that an average person could not do. In situations where other characters would automatically fail without getting a chance to roll dice, a thief might just be able to pull it off with a big amount of luck.
For regular hiding and sneaking I simply have characters make a Dexterity check, possibly modified based on how good their hiding places are and how much ambient noise is present.
Climbing (for B/X)
As with stealth, the climbing skill of the thief is actually called Climb Sheer Surfaces. Again, I agree with the interpretation that this applies to climbing things that would normally be impossible. And again, climbing things that are regularly climbable is done with a Dexterity check if necessary.
Locks (for B/X)
Unless there is time pressure, which there generally is not, a single roll is made to attempt to open a lock. This roll simply determines whether the thief can open the lock or not. If he fails then it’s beyond his skills.
Traps (for B/X)
I handle the detection of traps in two ways: A general quick lookover and a specific interaction. The thief skill is used only for the former. The player can declare that he quickly checks if something looks fishy and a roll is made. This roll is made in secret, which allows me to roll multiple dice in case there are multiple traps. If a roll is a success the character spots the trap. But any player can declare to do specific things to make sure there really isn’t anything hidden. A thief rolling to detect traps is only a chance for auto detection. Manual searching will find or possibly trigger a trap based on what the players are doing.