RPG Design - Elementary, my dear Watson!

Investigation lies at the core of A.D.A. gameplay. Or should be. Putting the mystery into RPGs isn't something groundbreaking and many ink has been spilled on writing on how it can be done. But it isn't easy to do it right. It has to feel satisfying. Especially for the players that want to solve a case like Sherlock Holmes (or that is what everybody is comparing it to).

I feel the current investigation mechanics lack that oomph it should have. So it is time to sit back and analyze what it means to investigate really and to do some reasearch. I've gathered some bits and pieces around the internet that sounded interesting and presented some potential for a future update of the rules. Maybe it also helps you to get some inspiration.

And apologies, this is going to be a bit of a longer post.

A study in Boardgames

I will not be the first (and probably not the last) that comes up with some ideas on how to handle investigations and mysteries. There is something to be learned from boardgames. There might be some inspiration for RPGs that came later. Lets have a look into the past and analyze some boardgames on how they handled mysteries.

Guess Who?

One came into contact with Guess Who? at some point or another in their youth probably, at least if like me you grew up in the 90s. It is a very simple two player game. Each of the players have 24 "suspects" in front of them (arranged in a 6x4 grid, but that doesn't matter). Each player then draws a random card depicting one of the suspects and the other player has then to guess which one it is. For that you ask Yes/No-questions back and forth until you guess the right person.

Each of the character portraits of the suspects had a name and some characteristics, that might be shared with others. So for example you could ask "Does the person have red hair?". And depending on the answer you could eliminate a bunch of wrong suspects.

One big disadvantage is that the game is actually solved. By choosing a question in such a way that you can always eliminate half of your remaining suspects, you have an high percentage of winning the game (this is not entirely true, as both players try to use the same strategy and somebody actually found a better strategy - but lets assume for the moment that it is the best strategy). This also means that the game is roughly solved in about 4-5 questions.

But lets analyse the game from a designers perspective. What makes this work is process of elimination. You have a fixed set of suspects that isn't going to change. So you can eliminate one-by-one each of the suspect until you come to a conclusion. What is also to note is, that each of the players is trying to find their own solution.

The second fact doesn't make it work in an RPG though. There should be only one solution that all characters agree upon (and that is hopefully the right one). The Gamemaster plays the part of the answering player but does not ask questions themselves. So with some simple modifications you could make it work in an RPG. The characters go around and asking people about certain characteristics of the suspects. But then they would need a fixed list of suspects beforehand. In a "real" world scenario this probably doesn't work. You don't know all the suspect, they might change and witnesses can be wrong.

But it could work in a murder dinner mystery kind of setting. All the suspects are in one place and it is the characters turn to find the murderer. Which brings us to ...

Clue / Cluedo

... the very definition of a murder mystery in boardgame form: Clue (or Cluedo in some countries). Hey it even made it into a movie.

The core of the game is to find the murderer, murder weapon and the location of the murder. In contrast to Guess Who? every player is trying to come to the same conclusion, but each of them is trying to get there first. What is also special is that every player has some hidden information. They get to know certain parameters (murder weapon, location, murderer) that are not part of the solution.

Gameplay is that everybody is then going in turn asking a question in the form "Was it with the in ?". It then goes around the table and if one player has the hidden information of one of those paramters he will share that information with the asker. Thus now both know that this isn't part of the solution. If no player shares anything, then everyone knows that this might be part of the answer.

The gameplay is a bit more intricate then Guess Who?. As the asker you gain some factual knowledge, but everyone else also gets some information. If you get an answer to one of the parameters, each other player knows, that one of them isn't part of the solution, but they don't know which one (except for the player that shared that information). So you have to ask cleverly.

To note also, is that every weapon, location and murderer is distinct. There isn't a possiblity to rule out one just because it wasn't another one (so there's no way to rule out all yellow weapons for example). And the form of the questions is also very restrictive. Still there exist some strategies to come to the conclusion quickly (and preferably before everybody else).

If this all sounded a bit confusing, maybe this video will help you.

What makes this game interesting from a design perspective is, although we still have a fixed list of suspects (and weapons and such) and we are still playing process of elimination, we start to have some sort of deduction in our game. Figuring out a good question to ask needs some clever thinking on the information you have to gain as much information, but giving others the least amount of information.

The sad part is, that this doesn't really work in an RPG directly. All the characters (that have no information) play against the Gamemaster that has all the information and the only way to win is to eliminate one possibility at a time. Even if you would introduce some sort of character knowledge, they can just share the information freely. There might not be a direct pressure to be the first and only one to solve the mystery.

Sherlock Holmes: Consulting Detective

That guy had to come up at some time, right. The most famous detective (apart from the Batman of course) in the world. No wonder that at some point there were games made with him. One of which still receives "updates" to this day (even though it originated from the 80s).

I would argue it is a step further than Clue did go. Now you are playing cooperatively with other players to solve the murder. You move around London, interrogate people and investigate proper by trying to find clues and connecting them to hopefully solving the case. Pretty much what we want. Right? The suspects aren't known. The players work together. The grey cells are engaged.

There are two things to note here though. First, the mysteries are premade. Someone had to come up with all the characters, dialogues and maybe props. And once you solved the mystery, you can't really replay it. You know exactly (more or less) where to go and what to ask. In an RPG this would push the burden to the Gamemaster to author everything and to think of every eventuality and consequence. And if you ever GMed you know that one wrong word uttered, the players will chase it down no matter what and "derail" all of your hard work.

The other thing is, that when you miss a clue you can get stuck. There are solutions around that (more on that later) but it is something to keep in mind. The engaging part is to connect the dots and not comming up with all the dots. Its like doing a jigsaw with only half the pieces.

Mafia / Werewolf

A classic in my time during camps or gatherings: Werewolf. Quick to setup, but produced fun and incredible mayhem at the same time. Probably one of the first instances of Social Deduction games. To this day this genre has found its way into mainstream media (mostly video games).

The premise is simple: You all live in a small town, but amongst the innocent villagers there live bloodthirsty werewolves (or members of the mafia, depending on what version you were playing). There are also other villagers with secretive powers that can gain some information in order to help the villagers find all the werevolves before they get all eaten. Every night one player will be eliminated as the wolves eat one of the villagers. And in the day phase the villagers will nominate another player (hopefully a werewolf) to be hanged at dusk.

The deduction comes in the form on what the players say, how they vote and in general how they act. Can they lie undetected, or do they have an obvious tell?

There exist even variations of the genre that have all players work towards a common goal but one (or maybe none) is a traitor and working against them. This could work in a RPG well and isn't too tasking for the Gamemaster. He doesn't need to come up with a full list of suspects or clues. The players will generate it themselves. However there's still a fixed set of suspects, but possibly an unknown number of traitors. And the Gamemaster could even insert some fake "clues".

As mentioned: Designwise it is marvelous because the players are themselves generating clues. But it only works in a social environment and again with a fixed and known amount of suspect. It is kind of similar to Clue but instead of asking probing questions you observe the suspects in order to come to the conclusion. Still there is some fair amount of deduction needed.

However in an RPG setting you will probably not play multiple "rounds" of the game. It can be fun as a part of a campaign but players don't want to play PvP all the time.

The Final Problem(s)

When trying to adopt these concepts into an RPG, one has to be aware of potential pitfalls there might exist. Some of them I have already outlined above but lets reiterate and maybe even look at some more, and give a possible solution.

Finding a solution

Every mystery at some point will (hopefully) come to a conclusion. Clues will be gathered and theories spun. But who says that it is correct? In a classical scenario the solution has been pretermined somehow. In Clue the solution is determined randomly from a set of possibilities and in Sherlock Holmes it is predetermined by the authors of the scenario.

The solution dictates what clues you will get along the way. It doesn't make sense that you get a clue saying the victim was killed with a gun and you get clues all around that (caliber and type of weapon and so fort) but the solution is, that the murderer strangled the victim to death. The clues need to make sense and when enough of them are gathered should lead to one possible solution. The correct one.

This puts a lot of pressure to the Gamemaster or to the person designing the mystery. As I have mentioned, players are really good in interpreting stuff in very unique ways and putting focus on all the wrong things. And coming up with consistent clues isn't that easy.

In general this isn't a strong problem. One can certainly design an RPG around that and sell a bunch of premade mysteries (worked for Sherlock Holmes). Since I'm trying to design a Solo (or GM-less) experience this puts either the pressure onto me to come up with mysteries or isn't going to work, because you can't design a mystery and then go and play the mystery. You would have spoiled yourself. The other issue is, that you can play each mystery once and then it is solved, so you have to come up with new things constantly.

But you can turn the premise around. What if you have a bunch of clues and you come up with a solution that would make sense with the clues you got. What prevents you from saying that this is the correct solution? You could put a difficulty on it and then roll for it to see if it is correct. In a way it is a Oracle Roll with the Question "Am I right?". Then you would only need a way to generate clues and it is up to the player to combine the clues. In the example above with all the clues of the murder weapon being a gun, it becomes a fact for the characters in the mystery.

Depending on the way on how you generate the clues, you can replay the same mystery again and could come up with a completely different solution.

This idea is nothing really new. There are some other RPGs that go into that route, namely Brindlewood Bay or the Brainstorms of Atomic Robo. In Brindlewood Bay you still have a bit of pretermined mystery as the clues and scenarios are setup for you, but the actual conclusion and which clues are actually presenented is different from game to game.

The quote I found in a forum post pretty much sums it up

In a lot of games, the GM comes up with the mystery, comes up with what really happened, comes up with clues, and then hands them out to the players. [...] Running brainstorm based investigations is a bit different. The GM comes up with the premise of the mystery, and they come up with an initial selection of clues to hand out to players, but they don't need to actually know what really happened at all.

When incorporating mysteries into A.D.A. this is one of my design goals. The characters come up with their own solution and this is then true. I already have something akin to that in my current rules. You have the Draw Conclusions action that essentially does that, turns clues into facts.

Who is this Colonel Mustard you keep talking about?

The other part of a mystery is someone you can blame. A list of suspects that have something to do with the mystery and with the solution thereof. You can interrogate them or beat the living shit out of them to get some answers if you are so inclined to do so. But you need them. You can't punch air (well, you could and maybe it makes "sense" if you are fighting of some ghosts, but I digress). Clue has a selection of suspects and Guess Who wouldn't entirely work without them.

But during an investigation you uncover more and more and with it you get to know more people that you could add to your list of suspects. A hodded figure was fleeing the crime scene and later down the line you get to uncover who this might have been. The police doesn't know the list of all the suspects when they come to a crime scene. They might know some (husband/wife, friends or family are on the top of every suspect list). They should come organically. That is what Sherlock Holmes does: You go around gathering clues and find some more suspects to interrogate.

Again this poses some sort of a problem. A suspect needs to make sense in the context of the mystery. It needs to be linked to a clue you have. The hooded figure for example. If it fled from somewhere else than the crime scene I probably wouldn't investigate that thread further. But since it is linked to the crime scene it forms its own clue and its own suspect.

In Brindlewood Bay you still have a predetermined list of suspects that are linked to the scenario and you are bound to them. Theoretically as a GM you can introduce a completely new suspect out of the blue but in general it will be restricted to those people. From a solo perspective this makes it also not much replayable. Yes, you can have a different conclusion but you are always dealing with the same people.

You can also generate a random list of suspects. As I said above, some of the suspects are probably quite obvious. So you can just randomly determine a wife or close friend to be on the list. You could link it to a location. You will encounter different people in a small town than in the big city. And when you create a new suspect (because you encountered a new NPC) you have to link it to a clue to tie them in into the mystery.

Searching the Room

The biggest offender in mystery games is the ability to find the clue. Since we're often dealing with randomness in our actions in RPGs, the act of searching a room can lead to differing results. You failed your Investigation or Perception check, well you missed the critical clue and are thus unable to solve the mystery. Better luck next time. Sherlock Holmes suffers from that. You have to find all the crucial clues otherwise you can't proceed in the story and this can become frustrating.

The GUMSHOE system is often the first thing you hear when you ask about designing a investigation scenario for your favourite RPG. Searching for clues shouldn't be random (or not entirely). The characters are often capable detectives so let them get the clues without rolling any dice. When a room is searched with some of the expertise the detective has, the clue is immediately detected. You could give out additional clues with a roll to make the investigation easier, but you don't need to. The real game is to come up with a conclusion and not finding clues.

So the takeaway is, that when trying to find clues, there should never be a fail state. There should always be something found. Otherwise the story comes to a halt. But I would go a step further (and other blogs seem to agree to that one), that it can introduce obstacles. A failed roll gives you a clue but it comes with a cost. It complicates the investigation and introduces conflict. It fails forward.

I think this matches perfectly in the Create Advantage action already. A clue can be an aspect and the action never gives you a hard fail but you can always choose to get the aspect at a cost.

Sometimes a chair is just a chair

Player can be very imaginative. That is good, that is part of a game. But as a Gamemaster you have to be very careful what and how you say it. Or otherwise a chair can become suspect number one for your group:

But sadly the other way around is much more common. You mention something and think - This is so obvious - but the players just ignore it and move 180 degrees in the other direction and follow an imaginary squirrel into the woods.

So you have just given them a clue for free using the advise above, but you have to scream at them to tell them that this is a clue and it is very crucial and you should further investigate on that. Basically solving the mystery for them. That's certainly not fun.

The other thing that is quoted so many times when you ask about investigation scenarios is the Three Clues Rule:

Because the PCs will probably miss the first; ignore the second; and misinterpret the third before making some incredible leap of logic that gets them where you wanted them to go all along. [...] For any conclusion you want the PCs to make, include at least three clues.

To note is, that this doesn't mean that your mystery should have three clues and then it solved, but rather that you need to give the same clue multiple times to the characters, so that the players register it as being part of their investigation and they need to followup on that. So if the clue for the murder weapon is Old-Timey Gun, then you need to repeat that same clue at least three times to the characters, in differing ways. For example the medical examiner could postulate that normal guns don't leave such wounds, a suspect that found the body could say that it smelled like black-powder in the room or the victims room is filled with old guns they used for war reenactments. This should then lead them to track down suspects that handle in antique weaponry.

The other thing is of course, that you need to be a right place and time to maybe find such clues. In the example above you need to go to the medical examiner or talk to suspects. So having multiple clues in more than one place gives the characters a chance to at least find one. There is also a great followup article to the Three Clues Rule that tackles that exact problem.

In a solo game, this might not be entirely an issue. You (hopefully) know when you create a clue and when not. But it might not be immediately clear if it has any significance in the overall solution. So when a clue comes up again or other clues point into the same direction it enforces the initial clue and should thus take more significance in the overall solution.

The Sign of Four

With A.D.A. I'm in a bit of a special case of an investigation scenario. There is no arbiter that has all the knowledge and trickles information bit by bit so you can come to a conclusion. And I don't want to invest too much time into generating scenarios and premade mysteries that get solved once and then it is done. To summarize the points that I feel are most important:

  • Clues and Suspects are generated by the system and are connected
  • Clues are guaranteed, but can have consequences
  • The solution is determined by the player
  • Replayability

It is to note that the current implementation probably covers most of these points in some way or another, but it doesn't fell complete. There are also other topics and thoughts I want to "briefly" cover.

It is all connected

When thinking about investigations and detectives, I always have this corkboard with various newspaper clippings and photos in my mind.

That is a feeling I want to invoke to my players as well. They are connecting the dots and uncovering this big mystery under it and then they feel awesome when they solved it, no matter how waky their theory was. What was very clear for me in the beginning was, that the investigation itself needs some sort of character sheet as well (Case File). It also made sense in the theme. As a bureaucratic entity A.D.A. wanted to have concise notes about the investigation. And I'm not alone.

A system that does this quite well is outlined in Mythic Magazine Volume 6. It pretty much includes a Mystery Board were you fill in your clues and connect them to a bunch of suspects. There are mechanics around on how and when a mystery is solved. I think this is a great starting point and some of the mechanics can be probably translated into A.D.A.

What is also important for me, that there is some sort of Showdown. The mystery is solved but you have a final confrontation, where you need to grab the artefact. There is also some uncertainty on how to produce clues and suspects organically.

You meet in a tavern

Generating suspects should be the easier part. Suspects are mostly tied to a location. What that means is, that you encounter the town mayor or local sheriff more likely in a small town, whereas in a forest you encounter wild animals and a hermit (which is probably hard in a big city). In a way I'm already prepared for that as I have Sites and Singularities as a way to vary the mystery. The site is the major driver for the suspect generation.

A suspect doesn't neccessarily has to be a person. It can be a plant, location or something else. Even a chair or a recently installed traffic light (I knew watching that show would come in handy sometimes).

What needs to be figured out how the suspects are generated. Is there a big list of possible suspect and you roll on one of them. Or is it more of a construction system, where you roll on different tables and in the end you get a suspect.

All about that clue

More important will probably be clues. What is nice about it, that you can interpret clues as aspects. This ties in nicely with the rest of the system as you can exploit them as well, which makes thematic sense. Exploiting would mean that you interrogate a suspect about a particular clue and see how they react. Then you could use that reaction to drill them some more.

Similar to suspect, the hard part will be to come up with a way to generate clues. And like the suspects are tied to a Site, the clues will be tied to the Singularity. The effects of the artefact will be producing signs and evidence that is left behind and picked up by the characters.

I see two ways to do it currently, both of them outlined in Brindlewood Bay and Mythic Magazine Volume 6. In Brindlewood Bay you have a list of about 20 clues you roll on. So a predetermined set of aspects that you then have to interpret. An example would be a changed testament. It is up to you what was changed about the testament (names or what each person gets and so forth). It depends on the suspect this clue is linked upon (so the change has something to do with them), but it is determined that it is a change on the testament. The next time you play the same Singularity (or Scenario) it can come up again. Or it might not make any sense in the current situation. For Brindlewood Bay this makes sense, as the clues are linked to the scenarios and the same group will play the same scenario only once.

Mythic Magazine has a bit of a different approach - and this is mostly because that is how the Mythic system works in general. Instead of having a big list of premade clues, you will roll on a big table of keywords and you have to interpret them yourself what it means. So for example two rolls on the table could be Change and Document. It can mean something similar than before - a changed testament. Or it can be something more mysterious like the document is changing right before your eyes. It can be adapted more easily to the current situation and make more sense. But it also puts a bit more work towards the player. The challenge could also be to make these keywords matter in the context of the Singularity.

Four Big Questions

My current system is focusing on answering four big questions: Type, Effect, Owner and Downside. I don't think they make much sense in such a system anymore. And in a way they get condensed into other aspects. For example Type and Owner are compressed into the suspect somewhat. And the effect and downside is visible in the clues to some extent.

That isn't directly bad, but the concept of the big questions was to give the investigation a goal and a way to contextualize a showdown. This will be somehow missing and needs to be replaced by another mechanism.

Let's get the party started

Another thing that I noticed is the start. The characters are put into a case without a real starting point. Yes, they know were to go and roughly what they are looking for (in the form of the singularity). But apart from that, there is no real call to action. So I think an Inciting Incident would be helpful to give a headstart into the investigation and set the first scene.

This also made me thinking about the type of investigation. There is the classic Whodunit. You see the consequence of some action and you need to figure out who did it. But there could be other investigations. For example you know who did it, but now you need to find them in their hiding place. Or the agents are affected by an artefact and they need to find out what is happening to them and how to stop it. The system should work with different type of investigations and ideally the Inciting Incident should change the type accordingly.

Last Bow

So now I have a plan to restructure my investigation rules according to my findings:

  • Inciting Incident: During the Ping I can let the player roll on a list of possible starting point, probably related to the Singularity choosen. This makes thematic sense. There was an actual Incident that lead the agency to investigate.
  • Finding Suspects: Suspects are generated from the Site. How exactly (fixed set or construction) needs to be figured out still.
  • Finding Clues: Similarly to suspects they are generated from the Singularity either by a list or using keywords. Clues are gathered by using a Create Advantage action and are in general handled as aspects.
  • Mystery Board: Something like the Mystery Board in Mythic Magazine Volume 6

Some of the mechanics already fit nicely in the current system such as clues being aspects and how they are generated. This also removes some of the "intermediate" actions like Draw Conclusion and boils it down to the initial three actions.

What it is missing though is the how to push everything into a showdown and make that work. It probably will still be some sort of scene challenge but the parameters are probably not so easy to determine as I will not have the four big questions anymore. That is another point I will loose. On one hand it gave some sort of structure to artefacts (and would later make the cataloging easier). On the other hand it maybe was to restrictive into gameplay. You had a very strict order on how to proceed in an investigation.

There is an east wind coming and I feel change.