Follwing through with all the wonderful advice and thoughts given on my previous thread about mysteries, I began to wonder; Is it possible to make step-by-step instructions for a GM to create a mystery to solve? See, I'm inspired by Dogs' instructions for creating a town. I mean, everything that a GM really needs is right there in black & white. It's just fantastic stuff.

I sat down & began scribbling notes. And threw them away and scribbled some more notes. I got nowhere.

So, I'd like to pose the question/challenge to everyone here; If you were to write a Mystery game, and were to include step by step instructions on how to write a mystery session, how would you do it?

-Eric

Mystery for whom?

Who is to "solve" it?

Why would anyone want to do that?

Mystery novels and movies are very enjoyable. It's important to undersrtand why. Enjoying a mystery novel or movie can include "solving" it, but that enjoyment is safe within the confines of understanding that upon finishing the book or movie, the solution will be provided.

How does that translate to role-playing? That is a rhetorical question, because there are many answers depending on the diverse ways and means of enjoying role-playing.

For instance, a particular set of players may be interested in solving the mystery very much in the sense that a viewer of a mystery novel might be, with the GM in the approximate role of the novel's author.. If this is the case, then it is absolutely necessary for the GM to make sure that the events of play will solve the mystery, just as the novelist does. The players' role is to add Color and perhaps to speed up the process, or elaborate upon how it gets solved.

Even that model has many possible avenues of application. Are the players competing to see who gets it first? Or not? That is only one possible issue to raise out of many.

Here's another, serious one: whose understanding matters, the players' or the characters'? Can the characters get it before the players do, perhaps with a couple of "good rolls"? Or what if a player figures it out when his or her character couldn't possibly know? Is the player then obliged not to blurt anything out until the character finally gets it? Or not?

And all of the above questions are predicated on the assumption that "novel-like" experiences are what's desired in the first place. There are at least four more ways of role-playing a mystery that I can think of which have nothing at all to do with sharing that assumption.

In other words, saying "play a mystery" or "run a mystery" tells us nothing. I have no idea what sort of role-playing you're talking about, or why the people are playing.

Best,
Ron

Certainly some clarification of what Eric has in mind would help -- what sort of mystery, for example? -- but I think the question is in general a good one. Here's a little essay in steps for how you might run the classic logic-puzzle mystery, a la Sherlock Holmes and so forth, as a Gamist challenge.

---

Consider the classic Sherlock Holmes-type mystery, in which the whole thing is in many respects a logical tour-de-force by the detective, culminating in a climactic confrontation of some kind (e.g. catching the bank-robber in "The Red-Headed League" or striking at the snake in "The Speckled Band").

The problem here, for the gamer, is that no GM can give all the clues available to Holmes, because it'd be fantastically boring. Take one of Holmes' "demonstrations," usually near the start of a story, in which he "reads" all kinds of detail about the client (or Watson) and then announces the solution; he then goes backward, upon being asked, to explain how he got there. The clues are things like chalk on an index-finger, a crushed line on a velvet sleeve, and so forth. These things could never be described in such detail by the GM. But when you come to think of it, they aren't described by Conan Doyle either: he allows Holmes to do the describing post facto. This makes it essentially impossible to figure out the mystery in advance, but also makes Holmes seem far more intelligent than we are (which is why Watson has to be there for reader identification).

Now compare this to the Nero Wolfe mysteries by Rex Stout. Not uncommonly, about 3/4 of the way through the book, Archie informs his readers that "at this point you have already figured it all out, and are wondering why I didn't see it, but you have to remember that I was there, with lots of other things going on, and you're just sitting back and thinking." This is a signal: all the clues are now on the table. It usually happens, Stout being a wonderful writer, just when the reader is completely in the dark. Once again, it shows that Wolfe is very clever, and it uses Archie to produce reader identification, but this time it assures us that every important clue is now available.

Okay, so problem #1 is getting the clues out where the players can see them without simultaneously putting neon signs around saying, "This is a clue."

Problem #2 is that even when we have all the clues, the players may well not be investigative geniuses, so they may get totally stymied just when the GM realizes that they have every necessary clue.

Problem #3 is what's sometimes called the "moving clue": if the players do not follow up what seems to the GM an obvious lead, they will miss an essential clue, so the GM moves the clue to wherever the players will in fact find it. Done inelegantly, the players may spot the movement and feel that the mystery is being cheapened or made dishonest.

My suggestion is a little abstract, but I think has potential. It depends on understanding what C.S. Peirce called "abduction". Here's a fast rundown:

Imagine we have a bag of beans and a pile of white beans, and imagine three possible situations and ways of thinking about them:

Deduction

Case: These beans are from this bag
Rule: All beans in the bag are white
thus Result: These beans are white

In this instance, we know in advance that we got the beans from the bag and we know that the bag contains only white beans. Provided that Case and Rule are true, deduction is certain. If one is not true, deduction is faulty.

Induction

Case: These beans are from this bag
Result: These beans are white
thus Rule: All beans in the bag are white

Here we make a guess that prior experience suggests a Rule that will continue to be true. This is not certain, but it can be tested: take more beans out of the bag. Unless we empty the bag, we can never know for sure that we're right, but the more times we take only white beans from the bag the more confident we can be. Induction thus has a statistical basis.

Abduction

Result: These beans are white
Rule: All beans in the bag are white
thus Case: These beans are from this bag

This is the wild-card of logical thought, because it's not logically strong at all. It's a guess, a hypothesis. Furthermore, unlike Induction, it is also non-demonstrable. No matter how many beans you subsequently take from the bag, all you demonstrate is the Rule, not the Case. You must have other evidence to establish that the Case is true; the Result and Rule can never validate the Case.

Now the point is that investigation in a mystery sense always works by Abduction. You know the end-result (the guy is dead in a locked room, etc.). You know a lot of Rules about human behavior, physics, poisons, and so on. And from these you Abduce a Case.

Now Peirce indicates that there are several ways to strengthen such Abduction:

• You can move to Deduction: if the Case were true, and on the basis of another known Rule, what as-yet-unknown Result would also have to be true? You then go look for this Result. If you find it, it suggests that your Case is true. If you don't, it means that your Case is not true (assuming that the Rule you proposed is true).
• You can move to Induction: you can seek a new Rule that will either violate or support your Case. For example, you can try to find out how a particular poison works, or what sort of behavior is plausible for the deceased or a suspect, or the like. These produce new Rules, and you then test them against the Case by finding their required Results, and so on.
• You can compare Cases against the facts and look for the Abducted Case that covers more facts than any other. That is, you can assume that there are no red herrings, that everything is important, and try to build a Case that makes every known fact not only plausible but actually required.

And you can basically continue from there.

Now all this is very abstract, of course, but it seems to me that you can certainly construct a game in which the players are encouraged to do exactly this, strongly and often. In my own experience running mystery-type games, I find that players do not, as a rule, tend to want to make up Abduced Cases. They generally try to work Deductively instead, taking the end-result of the mystery and all the clues and such as so many Cases from which to derive Results. What makes mystery novels work, however, is having the detective Abduce a Case: given all these facts, what story might have been true that would have produced exactly those facts? So the first step, to my mind, is to get the players to Abduce.

Let's go back to Sherlock Holmes, and imagine now that he's a group of PCs played by players. Suppose we formally divide the mystery into blocks.

In the first block, the characters investigate the Result, trying to find out every possible detail, however trivial.

In the second block, the characters discuss (Holmes's "brown study"), and try to generate a small number of Abduced Cases that would cover all the known Results and depend on certain or at least likely Rules.

In the third block, the characters investigate possible violations of the Cases. They check to see if the Rules they have postulated are valid: does this poison actually produce this result? They check to see if further likely Results of the Case are valid: did the dog bark in the night?

In the fourth block, the characters rebuild their Case to be as watertight as possible, if necessary repeating bits of the third block until they're certain.

In the fifth block, the characters develop a plan of action that will confront the criminal with the complete Case in such a way that he or she cannot avoid it without demonstrable lies or the like, and they put this plan into action, producing a nice climactic scene.

We don't need the classical denouement here, because the detectives do not have to explain the Case or how they got to it -- the audience, i.e. the players, already know all this. So instead we might want to make sure that the conclusion of the fifth block will provide a satisfying wrap-up. Action-scenes are a good way to do this.

So let's go back to our original problems.

#1: Get the clues on the table
If the players are working abductively, they will create clues of their own accord, then seek them out. That is, if they hypothesize that the killer must have run across that flowerbed, they will realize that it will help the Case immensely if there are footprints in the flowerbed. If the Case is accurate on this point, the GM has them find the footprints.

This means that the GM's job is to have a clear Case in mind and to work very strongly on Deduction, prior to the game, to figure out what the clear Results will be. He should not provide too many red herrings; these can wait until the players are confident with the method and/or the third block of the story, once they're more or less on the right track.

#2: Players may get stuck
They're not that likely to get stuck on parts 1 or 2, if the GM provides the initial setup effectively. Any slob can make up a Case that will fit, and in fact many of them. So long as the GM gently reminds them every now and then not to decide on a Case too strongly in part 2, before they have enough evidence to make a valid choice, they should be okay here. But part 3 is where the difficulty may arise. An NPC investigator (such as a Lestrade or Cramer type) is very helpful here.

Have the NPC make a strong decision about the meaning of some clue and hare off after it, using all the resources of the police. If there is some antagonism between the PCs and the NPC, this creates time pressure (beat the police!). It also creates a classic Nero Wolfe situation: if the police have the story right, they will win, because they have enormous manpower resources. So the PCs can simply take it for granted that the police are wrong, because otherwise they cannot beat the police. This tactic by the GM should attract attention to the clue in question and raise the possibility that there are multiple interpretations. This should send the PCs into Abduction again: what other Case might produce this same Result? on the narrower scale of just this one clue.

Another thing that may stymie them is a red herring. If they get too hooked on one, have the NPC investigator make the same conclusion and get stuck, hard. He then admits his total confusion to the PCs. This suggests to the PCs that they have gotten off track.

#3: Moving Clue
This is a minimal problem if the players are pushed to Abduce. The GM actually doesn't have a lot of clues in mind beyond the initial setup, you see. He knows what the Case is, and the Result. The players must then invent all the other clues as hypotheticals. If their hypothetical clues fit the GM's Case, then those clues exist and are true; if not, not. This can also be used to head off the red herring problem: at some point the red herring must lead to a Result that is not part of the actual Case, and so long as the GM ensures that this arises squarely, the PCs must realize that their Case is wrong.

So how do we do this, as GM's?

Players Must Understand Abduction
It doesn't have to be in formal, Peircean terms, of course, but they must understand what they are doing and what is expected of them. The rules of the game have to be clear, in other words. Breaking the game into those five formal blocks will help, because it allows the GM to announce, "Okay, you're done with the first block, you have all the clearly-apparent pieces of the Result. Now go to the second block." The players will then make the shift and furthermore know that they are making progress.

Situation Must Provoke Abduction
This process is somewhat time-consuming, so there cannot be excessive time pressure. The police, or whoever has all the big manpower resources, must be a little dim and must be mildly antagonistic to the players. These things ensure that an abductive solution is by far the best way, and deters the players from simply knocking on doors and doing other basic police spadework.

Situation Must Be Bizarre
This is one of those classic points that Conan Doyle got dead right. Most crimes are either very simple or hideously complicated. Either you find someone standing over the body, weeping and saying, "He didn't ought to have said that about our Ken!" in which case you can pretty much wrap up but for the shouting; or you find that everything depends on elaborate and intricate interpersonal relations that are fantastically meaningful to the actual people but very dull for investigation. Some writers get around this one by having lots of very complicated and interesting NPCs, as it were, but in a game I think it's best to keep such NPCs to a minimum. You want one or two, to provide personal interest and so forth, but keep the focus on clues and logic.

The best way to do this is to follow Conan Doyle. Make something truly bizarre about the situation, along the lines of the Red-Headed League or The Speckled Band or The Copper Beeches or whatever. If the Result is exceedingly odd, there are very few Cases that will fit it. Make the NPC police not see this point -- they never do, and why should they? -- but ensure that the players realize that the very oddity of the Result makes their job simple and interesting. Whatever you do, do not make the oddity of the situation irrelevant, a red herring. If poisonous octopi are involved in the Result, they must be important to the Case; that cannot be a trivial detail.

GM Must Be Honest
The players must know that the GM will not alter his Case -- ever. Otherwise the whole thing becomes pointless. If the players are supposed to figure out the mystery, they have to be absolutely certain that when they get it right, they've really gotten it and can be proud of themselves. Do not undermine this by changing the Case. The players should recognize, however, that the GM has not necessarily gone around planting all the clues that could possibly be present; he has built a clear Result and has a Case in mind, but if he didn't happen to think of the point that there would have to be blood on the mantelpiece just behind the clock, then if the Case is accurate and the players have correctly Deduced this, then there will now be a dab of blood behind the clock. Note that the players must get into the habit of not saying, "Yeah, but we didn't see blood there, so we don't need to look again; we must be wrong"; they must say, "Well, we didn't look that carefully at the clock, we should look again." I don't know how you encourage this, exactly, apart from stating at the outset that this sort of thing is very important.

Game Should Be Progressive
Seems to me that a game like this ought to get harder over time. At the start, you can provide relatively small challenges to get the players into the swing of things. Keep the possibilities limited, keep the cases super-weird, keep the police turning up all the basic footwork stuff and providing it free of charge, and so on. Once the players get rather good at this, up the ante. Eventually they will know just how to go about things, and can deal with really very complicated crimes. I don't know for sure, but I think you could structure this stuff so that initially the crimes take a good evening to solve. Once everyone's on the same page, you go to the multi-session things. Now you're going to need the odd action sequence in the middle of various blocks; take a tip from Holmes and Wolfe mysteries and plant them so that the result of such action is an important clue.

---
All of this is predicated on the notion that the game -- in the sense of a campaign -- is going to be a mystery game. And furthermore, on the notion that this is going to be a basically Gamist thing: the players must Step On Up and try to meet the challenge. If you just want a little mystery to spice up another game, and the characters aren't necessarily detectives or even intelligent, you have a BIG problem that IMO is probably at least partly insoluble.

Ron Edwards wrote: Mystery for whom? Who is to "solve" it? Why would anyone want to do that?

For instance, a particular set of players may be interested in solving the mystery very much in the sense that a viewer of a mystery novel might be, with the GM in the approximate role of the novel's author.. If this is the case, then it is absolutely necessary for the GM to make sure that the events of play will solve the mystery, just as the novelist does. The players' role is to add Color and perhaps to speed up the process, or elaborate upon how it gets solved.

Here's another, serious one: whose understanding matters, the players' or the characters'? Can the characters get it before the players do, perhaps with a couple of "good rolls"? Or what if a player figures it out when his or her character couldn't possibly know? Is the player then obliged not to blurt anything out until the character finally gets it? Or not?

In other words, saying "play a mystery" or "run a mystery" tells us nothing. I have no idea what sort of role-playing you're talking about, or why the people are playing.

This is a great post, Chris. I don't have much to add, but I do want to offer public praise. :)

I agree - good stuff, Chris.

I would add that my successful mystery runs have always involved getting the PCs involved with a large group of NPCs and a situation they care about first. Then the crime/warning of impending crime comes, and the mystery becomes an alternate sort of challenge medium (alternate to combat and 'pure role-play') for PCs to work out their imaginary goals and establish character values.

This is a really fine post. I would add to the section on helping players get started that it is perfectly in-genre for the GM to say "hey, that's a red herring" by killing off whoever the prime suspect was.

The other thing that occurs to me is that the schtick in the Miss Marple books is that she is always saying "oh, this is just like that time when the postman was stealing herrings from the mail". In the books this isn't helpful, just a way to demonstrate the point that all crimes are like things she's seen before, but it seems like it might be an aid for the players trying to reason abductively if you can give them some cases that they remember that are similar.

This can be done as a hint without giving the game away -- the resemblance to the postman story might be because someone is untrustworthy, or because someone is abusing their position, or just because someone really likes fish. Furthermore, just because Miss Marple remembers a story about someone doesn't mean they're the criminal or even doing something wrong -- it may be a more positive statement about their character.

neelk wrote: This is a great post, Chris. I don't have much to add, but I do want to offer public praise. :)

I just wanted to say thanks for your thoughts in the step-by-step mystery thread. I'd read it earlier and was hoping something good would come along.

After reading what you had written, I closed that tab and went back to surfing the web... and then a second later leand back in my chair and said, "Shit. I know how to run a mystery now. That's so cool!"

Right off, I'd like to hug you Chris. You understood what I was looking for and wrote some gold in here for us.

What Eric's asking is how to construct such a mystery so that the players can solve it and feel pleased that they have done so.

Causality, folks! If the players have to solve the mystery, you have to take the random factor out completely. If they ask the right question, they get the right answer, period.

Technocrat13 wrote: Right off, I'd like to hug you Chris. You understood what I was looking for and wrote some gold in here for us.

I wrote: Causality, folks! If the players have to solve the mystery, you have to take the random factor out completely. If they ask the right question, they get the right answer, period.

Eric wrote: Take the random factor out completely? I'm not convinced that would be necessary....

I keep coming back to the classic investigator's scene where they spot the clue that no one else could spot, or understand. For example, in Chris' example of the Flower Bed & Footprint clues... Sure, the PCs -have- to find the footprints, but what else do they find?

I didn't think I was falling into "follow that clue", I thought I was following along with your Block 1 idea. Maybe I wasn't clear about what I was imagining.

Block 1: The PCs get the Basic Clues, as Provided by the GM

In the first block, the characters investigate the Result, trying to find out every possible detail, however trivial.

Technocrat13 wrote: I guess I just pulled the example from the wrong place in your write-up. You used the flowerbed as a Block 3 example, and I used it as a Block 1 example to set up for my proto-clue idea. Does that sound right?

I'm excited to hear what you think about the proto-clues. They itch in my brain like a good thing, but I can't quite work out all the details, so I can't tell for sure if they really can be a good thing.

More Points on Mysteries -- VERY LONG

Here is a more comprehensive sketch of how I think this would work.

I cannot stress enough that the players have to know how to be, as the kids say, “proactive.” They have to know what abduction is, and they have to understand what the rules of the game are. I can’t tell you how many mystery-type games I’ve run or been in where the players dick around saying, “Gee, I can’t think of anything, this sucks.” Because the constant assumption is that there are clues out there and you’re supposed to find them. In this version, there are no such clues “out there,” and in fact there is no “out there.” Everything is invented as a set of logical inferences coming from the players.

At the same time, clearly there are GM tasks here. This is very difficult actually to run, because GMs have bad habits too that run counter to how this mystery thing works. So let’s try to think through them. I mean, I may have seemed to answer Eric’s question, but when it gets right down to it what I’ve described is a product and some guidelines; I haven’t produced a step-by-step. Let’s give it a shot. Here’s my first draft; you guys pitch in and fix it where I go off the rails.

What I myself would really like to see as a continuation of this thread, apart from people revising and correcting my idiocies, is someone developing a complete 6-part Case and musing about what would and wouldn’t work about it. My sense is that this is something that takes tremendous prep from the GM, but very little run-time effort apart from consistent NPCs. I’d also love to see some Actual Play threads where someone tries this out!

Laying the Ground-Rules: Background
A great start would be to have them read this thread, frankly. In any case, write a little essay on the three kinds of logical inference, how Sherlock Holmes works, and all that. Then make the players actually read (or, we hope, re-read) some Holmes mysteries.

A very good example, and a famous one for semioticians (who all love Holmes), is “The Adventure of Silver Blaze.” A horse gets stolen one night. The client is sure that it was the guys from the next racing stable over, who are rivals. Holmes investigates. Inspector Gregory, one of the police who has a brain, is on hand.

Inspector Gregory: “Is there any point to which you would wish to draw my attention?”
Holmes: “To the curious incident of the dog in the night-time.”
Inspector: “The dog did nothing in the night-time.”
Holmes: “That was the curious incident.”

Holmes has made a preliminary Abduction, or rather has had one proposed to him by the stable owner: the horse was stolen by a rival from outside the stable. Holmes now moves from this Abduced Case to a Deductive inference:

Case: The horse was stolen by an outsider
Rule: Stable dogs bark at outsiders
thus Result: The dog barked in the night

But the dog did not bark; that is quite “curious,” as it suggests that the Case is wrong. If I recall correctly, Holmes has established along the way that the dog does indeed bark at strangers, checking his Rule, but it’s easy enough to test regardless. Now the only other possibility, because of the Rule chosen, is that the horse was not stolen by an outsider. It may have wandered off, but we’ve established that this is impossible. If may have simply disintegrated on the spot for no apparent reason, but this is outside possibility. Therefore the horse must have been taken by an insider. Interesting: we now have a completely different sort of Case, and some obvious avenues of investigation. Who left the house in the night? Who stands to gain by the horse vanishing? And so on.

I suggest that you walk your players through this a bit around the time of character design, having first given them a write-up and told them to go read a bunch of Holmes. Now they should understand their basic task: Abduce, then Deduce.

Laying the Ground-Rules: The GM’s Promise
The GM will never change the Case. The GM will not invent a Case that depends entirely on coincidence or luck. The GM will not invent a Case dependent solely on specialized knowledge not readily available to the players. The GM will not “bait” clues. The GM will not lie. The NPCs will not always be smarter than the PCs nor try to mislead them. The GM will not play, “Ask the exactly right question.”

Changing the Case
Once you have invented the Case, i.e. have determined the sequence of events that led up to the Result, it is set in stone. You should be able to write up the Case in a series of a few clear steps, all logical and coherent, and put it in a sealed envelope. You needn’t actually do this, but you must be as locked to the Case as the players are.

Cases of Pure Coincidence and Luck
Coincidence and luck are fine, but the Case must not be entirely dependent upon them. For example, suppose Sir Nigel was shot in the heart at 9:00 pm by his wife’s lover, firing through an open window and using a powerful air-rifle. The butler cannot accidentally drop a pistol and also shoot Sir Nigel through the exact same spot at 9:10. On the other hand, Sir Nigel might at 8:00 have drunk an enormous amount of opiates to commit suicide and drifted off to sleep – and death – at 8:45, only to be shot by the wife’s lover who doesn’t realize that the very still form of Sir Nigel in his favorite chair is already dead. This sort of weirdness should be reserved until the players already know the game, but the coincidence must be plausible in the sense that it’s not totally unbelievable if you were to read it. (I find that Agatha Christie often violates this one.)

Another example. Lady Margaret dies when a big rock falls on her from the battlements of Castle Harlech, where she was being a tourist. That rock cannot just at this moment have become loose; it must already have been known that it was loose.

I strongly recommend that all murders be cold-blooded and pre-planned. If someone in the heat of the moment grabs a poker and beats someone else’s head in, he should be covered with blood and all that, and he’s very unlikely to get away clean. Lots of easy evidence for the cops to follow up. But if someone devises a fiendish plot, that means on the one hand that the cops will screw it up, but on the other hand all the clues have in a sense been pre-laid. Holmes always talks about the “bizarre” and “outré”, and you should follow the Master’s advice.

Specialized Knowledge
Make a distinction between specialized knowledge that is clearly needed (but not available) and specialized knowledge that is not clearly needed.

For example, suppose a man dies of falling from a window. Upon investigation, it is found that he has a broken neck and some strange marks on his face that look like little pock-marks or something. The Medical Examiner says he doesn’t know what the pock-marks are. Faced with this, the players do not know that the pock-marks are something requiring very particular specialized knowledge: if they are actually sucker-marks produced by a bizarre hybrid strain of giant blue-ringed octopuses (which are very poisonous), the players are now totally screwed... unless, of course, you also immediately provide them with a strong clue that says, “Look into poisonous cephalopods of Indonesia!” And it’s hard to see how this could be done smoothly.

But suppose on the other hand that the Medical Examiner says, “You know, if I didn’t know better, I’d say that these pock-marks are sucker-marks from octopus tentacles. My son keeps the things in an aquarium, and I once had to pull them off myself. But, no, that’s impossible.” Ah. Now the players know that specialized knowledge is needed, and they have some rough sense of what to do about it – although of course finding the right sort of specialist may be no mean trick.

You can extend this principle into any forensic details, railway timetables, tobacco ash, tattoos, signs of secret societies, and the like. These are the sorts of things that make Fu Manchu stories terrible mysteries, for example, though they’re good fun to read.

The GM will not “bait” clues
Do not place clues like bread-crumbs to mark a trail. You find X, so that leads you to Y, and that leads you to Z, and so on. The GM must not plan a trail like this. It sounds like a good idea, maybe, but if you do it you are predetermining the direction of investigation, and if the PCs depart from that track, they are simply lost.

Conversely, do not drop red herrings from the sky to deflect the PCs from solving the mystery too soon. The GM may, if he likes, develop some preliminary red herrings and run with them. But if the players are outsmarting him, he may not put bright, shiny clues in their way de novo to keep them from beating him like a red-headed stepchild.

The GM will not lie
This seems simple, but it’s actually very tricky. If you read a lot of really good mysteries, you will note that the narrator must be quite cautious at times. My favorite negative examples here are Rex Stout and John Dickson Carr, two authors I like. Every now and again, the narrator apparently tells you something that turns out to be a bare-faced lie; later still, usually right at the very end, you find out that it was literally true but absolutely untrue in its apparent sense. This sucks. I love both these writers, but this device just plain blows. You cannot do this to your players. They must be able to take you at your word.

One way to keep yourself honest is to lean on the monosyllabic answer, then amplify: “No. I saw nobody that night.” Don’t say, “I saw nobody out of their rooms that night,” then later on say, “Well, see, that witness thought that by ‘their rooms’ you meant that group of people, so when Sir Nigel was in Lady Margaret’s room in the night, that didn’t count because the two of them were ‘they’ because they’re lovers.” Your players should kick your butt if you do this.

On the other hand, you don’t have to be Mr. Honest. “I saw nobody that night” can validly ignore, “Except for Rover, of course.” It is perfectly reasonable that somebody with nothing to hide would neglect to mention the dog when asked if anyone was out of his or her room at night, because the structure of the question presumes that we’re talking about people. The correct answer can also be, “I saw nobody that night, except Cook of course, but she’s always in the kitchen at 4:00.” This may send the players haring off after the poor cook, while missing the important fact about Rover, but that’s not the point.

The NPCs will not always be smarter than the PCs nor try to mislead them
Most NPCs do not know what has happened; if they do, they’re accessories after the face. They are scared and unhappy and deeply worried. They may well resent the police. If the PCs are polite and charming, they have no reason not to want to help out. Some people will dislike them, of course, on general principles; some will think privacy is everything. These people are a pain in the ass. But lots of people will help just because they are basically honest people.

Most NPCs are basically not so bright about things like murder (or whatever). They don’t know anything about a murder investigation, they don’t know anything about crime, and they don’t care. They have their own lives to lead. If they try to deceive the PCs, it will be about stupid things: “No, I wasn’t there with my girlfriend, we’re not that kind of friends, I was, um, ah, playing pool. Yeah, playing p-pool. And nobody saw me. Okay? Hah. So there.” The PCs basically know from this (1) he was screwing his girlfriend, probably, and (2) he’s not so bright, and (3) he probably had nothing to do with this anyway. You just either catch him in the lie if it matters or push on if it doesn’t. But a perfect seamless lie from every NPC makes a case impossible and basically dishonest.

The GM will not play, “Ask the exactly right question.”
Again, people are basically honest and decent and want to help, to some degree. They can at least be convinced to help in some small way, with the right coercion. If you ask, “Did you see anyone coming out of the house that night?” and the answer is, “No, I didn’t,” the full answer shouldn’t usually be, “No, I didn’t, but I did see someone go into the house, but you didn’t ask that so I won’t tell you, ha ha.” Not everyone is an asshole. People volunteer information constantly. In fact, one of the main problems for real detectives is that people volunteer far too much information, and most of it is irrelevant.

If you’ve ever played text-based adventure games like Zork, you know the frustration of the situation in which there is one exact command-word or -phrase and every other one that is basically the same is no good. Don’t do this. Make a difference between an RPG and a bad old adventure game. For goodness’ sake, you’re a human being, not a computer!

Inventing The Case

Step One: The Straight Story
Who? What? When? Where? Why? How?

No fillips, no complications, no weirdness, no coincidence, no luck. Keep it straight. At this stage, you can have the fiendish plan be as fiendish as you like, but there must be no accident; it all goes perfectly according to plan. Be sure that the plan is indeed fiendish; a simple murder is hardest to solve.

Some people like the “heat of the moment murder with an elaborate cover-up.” This is tricky, but possible. I recommend against it for a first shot.

A few essential questions should take you further on.

Why now? If Dave has hated Phil for many years, why did he decide to strike now?
Why this way? If Dave used rat-poison, why did he choose it?
Why there? If Dave did it in Phil’s home, why there? Why not at his office?

Step Two: Messiness
Something went wrong. There are no perfect crimes, and clues were left.

What were they? Why were they left? Who saw or heard? What did the killer forget? How did he deal with it?

If you want to get complicated:

When something went wrong, what coincidence happened that allowed the killer to get away with it? Until the players know the game, avoid this trick.

Step Three: Bystanders
A man dead of a gunshot wound in a secluded apartment where nobody knew he lived, with no friends or acquaintances, and only footprints leading away, leaves nothing to work on.

Who are the friends? Who are the servants? Who are the lovers and family? What’s the story about the victim? How do these people feel about the death? Who gains? Who loses? What’s in the will? Who knows that? You don’t need to know all of this, but you need the gist.

Step Four: The Result
This whole story is now in motion. What was the timetable of events? Who was where? What are the basic clues left? Don’t get cute; just keep it straightforward. What, at base, will the police find out within a few hours?

For every NPC who isn’t involved, have them have a reasonably normal schedule that they generally stick to. One or two NPCs who couldn’t follow their routine because of some personal complication is fine, but a lot is annoying. Consider one NPC who couldn’t follow her routine because she got a phone call telling her her mother was dying, but when she got there her mother was just fine; the phone call was placed by the killer to get this NPC out of the way. That’s a good clue, but you have to make damn sure that it gets revealed or isn’t really all that necessary. The easy way is to encourage the players to run through the NPC routines, and have that NPC tell this story immediately, no mucking about.

Step Five: The Difficulty
Why can’t the police solve it? What makes the whole thing seem much more complicated than it really is? A classic here is that half the innocent people won’t talk to the police honestly because they think their innocent but peculiar personal stories will incriminate them. So what are those stories? You don’t need a lot of detail, just the basics. Don’t do too much of this until, again, the players know the game.

Step Six: The PCs
Who calls them in, and why?

The point of this whole exercise is to generate an actual Case that seems very complicated but is actually extremely simple. Once you have very clearly in your head who did what when, and what is immediately apparent on initial investigation of the crime scene, you’re set to go.

Starting the Ball Rolling
You need an investigative group who work together and have complementary abilities. In a modern story, maybe there’s the forensic medicine specialist, the charming talker, the ballistics guy, the photography and sound guy, the computer guy, and so on. I recommend against too many people: three sounds perfect. Be sure that at least one person can talk charmingly and get along well with people. Ngaio Marsh’s Alleyn and Fox are good: Alleyn can talk to wealthy people very easily, and Fox can get pally with servants in seconds. Rex Stout’s Archie Goodwin is a genius who can get along with everyone immediately – especially women.

The group should have a decent working relationship with the police, but should also have some rivalry going. Maybe it’s just a question of earning a fee, but at any rate the police should be helpful in their way (they want the murder solved, however it happens) but the PCs should want to get there first. Inspector Cramer hates Nero Wolfe in many ways, but he also respects Wolfe’s brilliance and he really, really hates having a big-news murder unsolved; he’d rather cooperate with Wolfe than let the murder go unsolved and get bad press for it. Unfortunately, Wolfe is such a swine that Cramer is usually steaming by the time he leaves the office. Ah well.

The group should have an established way of meeting clients, a type of detective practice, a general type of client, and so on. If they have a very upscale practice, like Nero Wolfe’s, they will get rich clients who want them to do things their way. If they have a grimy, back-alley practice, they will get creepy thugs who are afraid of being jailed for crimes they did not – in this case anyway – commit. Stick to one shtick.

This stuff should all be worked out as a group.

Now produce the client – who may be a policeman, depending on the situation – and provide the outline of the case. The client should have a clear direction for how the case will run and who is probably guilty. The client should also be wrong. If the client is a policeman, he should have a clear sense of what has occurred – which is wrong – and be totally lost as to how to proceed.

Stage One: The Scene of the Crime
Now we move to the Scene of the Crime. The investigators proceed to the scene and get the basic rundown from the police. They look at anything and everything. You need a list of what the basic clues are, the facts they must know; this is essentially the Result. Do not let them leave until they get all this information, one way or another. The police should provide stuff the PCs do not find, as necessary.

Stage Two: Abduction
The players/characters now discuss. They must come up with a number of hypothetical possibilities that fit the known facts. They should be encouraged to find as many solutions as they can, and to grade them by how many of the known facts become predictable by this story. They should also seek the simplest possible hypothesis.

Once they have a prioritized list, they are ready to go track things down.

Stage Three: Investigation
The investigators now make a series of logical inferences from their various hypotheses. If X were true, then Y must also be true, so we look for Y. If the killer ran across the flower-bed, then he must have left footprints, so we look for prints. If the killer used rat-poison, then the stomach must contain this poison, so we ask the medical examiner.

Unless they have immediately chosen the correct Case, they will often run against things that are not true. Furthermore, they will have to ask NPCs about these things, and will get more information than they needed. It’s not all yes/no stuff; they find out that the victim did not have rat poison in his stomach, but in fact he had nothing at all in his stomach, despite the fact that he apparently ate dinner just an hour before his death. Huh? What? How is that possible? Another example, and a better one: they want to check whether Mrs. Grundy saw someone enter the building at 8:45 am. She might have been asleep or not looking out the window, of course, but if the PCs are on the right track you might have her be called to the phone just then by the killer’s accomplice. A completely pointless, annoying call that went on a strangely long time. It’s been bothering Mrs. Grundy, and she’d like to tell the nice young investigators all about it, in great detail, and incidentally don’t they think her poodle is very nice too? And won’t they have some more tea? And on and on and on. Silly trivial characters like this spice up the game immensely, especially when they do have one really important nugget of gold and are happy to give it up as soon as the investigators get anywhere near it.

This stage is the main part of the game, and it takes quite a while. Lots of investigation, asking questions, arguments, charming servants, and so forth. If there are creepy underworld types involved, maybe people strongarm them or think they’re police. If everyone’s very high-society, maybe they spend so much time sneering that they give unconsciously deceptive answers, or simply refuse to talk until coerced; on the other hand, the servants hate them, and will tell all their dirt and gossip immediately – much of which is irrelevant.

At the end of this stage, the PCs should be narrowing down to a Case that is very close to correct.

Stage Four: The Case
Now the PCs discuss it all and get the whole thing very clear in their heads. They check every detail and make sure it’s all watertight. This can take little time or a lot. They may have to go back and investigate new things, or they may realize that there’s a whole other Case that is really the right one, and so on. But this is mostly a matter of talking it through to be sure they’ve got it clear.

The GM should not let them get through Stage Four with a totally wrong Case. If that happens, kill off their suspect, or do something equally drastic that sets them back on the right track. In murder cases, killing the suspect is the usual method. And the more dead bodies pile up, the more the PCs know they have not done well.

Stage Five: The Plan
The PCs must come up with a plan to expose the killer and bring him to justice. This will depend a lot on their style and the type of mystery. Do they confront him head-on? Do they try to bait a trap? Do they convene all the suspects and grill them by presenting the whole Case as a hypothetical?

And then they actually go do it. Let it work, since the Case is perfect or close enough. But make them work for it a bit. This is the climactic moment, and you need some tension and power. If this is a tough-guy gritty thing, the climax should probably happen at gunpoint, with the clock ticking down. If this is an upper-crust servants-in-the-old-manor thing, restaging the crime is a classic. The players should lean on the genre here, and the NPCs should react accordingly.

The Whole Session
This will take a lot of practice on everyone’s part. You will need to keep it simple at first, but remember that this means having the Case itself be very weird but with as few extraneous facts as possible. The more times you do this with the same group, however, the more you can interweave the Case with extraneous information and complication.

Read a lot of really good mysteries. You want things like this that are puzzles: John Dickson Carr, Sir Arthur Conan Doyle, Rex Stout, etc. I find Agatha Christie mysteries are often very dependent on wild coincidences at exactly the right time, but you may disagree. Carr is the master of the locked-room mystery, if you don’t know him. Rex Stout is a truly fine writer, and his characters are wonderfully complicated and rich; his mysteries are also often very simple at base, but complicated by a lot of people being self-centered and stupid, which is great stuff but something to be a little cautious about in gaming. Holmes mysteries are wonderful models, but remember that you have to work backwards: you have to make the players create all those weird clues he sees immediately.

To reiterate my original point, I think the most important thing is to have the players all read a lot of such mysteries and, quite possibly, this whole thread on the Forge. They simply must understand that you are asking them to do something they may not be used to: I have never seen a gaming group that is used to doing this. If they don’t get how this is supposed to work, they will dither and wait for clues to present themselves. They must invent hypothetical clues and go seek them out.

Eventually, if the group likes this sort of thing and you are all willing to work together for it, you will end up being able to generate wildly complicated Cases, with insane complications, and have them walk right through being cool and discovering things right and left. The police will be stunned, the ladies will sigh, and the men will say, “My gosh, how terribly clever!” And with that in hand, you can make such a mystery tell a deep and meaningful story as well as a great puzzle. That’s a mystery worth playing, but it takes skill and practice.

As Nero Wolfe said to Julie Jacquette, “I wish you well.”

clehrich wrote:
I guess what I'm saying is that if you want to tell a solid Premise-oriented story -- Narrativist Story Now stuff -- then narrative power to manipulate proto-clues to clarify the Premise and keep it focal is probably worthwhile. You can have all the NPCs have subtle involvement with the same Premise, for example, which wasn't required by the GM's Case. For example, to keep with the race Premise, all the other people in the story can be made to have important perspectives on the issue, which complicates the apparent Case wildly without much changing what has really happened. But on the other hand, since the players cannot know which things are really clues and which are incidental, that same narrative control is likely to produce a great story and a Case that probably isn't much like the GM wanted and quite possibly doesn't make a whole lot of sense.

I'm nearly finished designing an adventure that is a mystery, and I'm probably going to pour over Chris' contributions here to refine a few points. However, I'm going to suggest a few things that he has not covered.

The first thing that's not been discussed is the critical question: why do the players' characters care about the answer to this mystery? There's been an assumption from the beginning that "this is what they do"; but why do they do it? Dagliesh does it because he's a police detective; Poirot and Holmes are private detectives. Jane Marple just does it because she understands people, and when friends of hers are in trouble because of something like this they call her for help (and a number of police officers have learned to trust her instincts, so they, too, look to her for insights when she's around). Tommy and Tuppence just do it because they love mysteries, usually, so when confronted by one they leap into it.

But if you're talking about introducing a mystery as an adventure in an ongoing game, you need to think about why the players are going to have their characters take an interest in solving it. A call for help from someone who thinks they could be helpful is a good possibility, provided there's reason for the players to want to help and it makes sense for the person in need to care. One motivation that has always worked for me is putting the players' characters in such a position that they will be suspect, so that they will want to solve the matter to clear their own names.

Chris has several times suggested that you kill the character the players have come to suspect if it's the wrong character; that can work, but it faces a second set of complications. Why did the killer kill the person on whom suspicion has fallen? It's much more common for the second death to be someone who knows something but hasn't told anyone yet--whether because they were thinking of blackmailing the killer or because they haven't yet realized what it means. Christie's Thirteen At Dinner made good use of this, as the killer heard the second victim make an appointment to talk to the detective, and although the killer didn't know what mistake she had made, she knew she had to silence the witness. This led to the detective trying to work out what the second victim must have known that would have helped solve the first case. That, though, points to the problem with killing suspects: the killer has to be threatened by them, and that means they have to know something, or at least have been in a position where the killer thought they probably did.

Another technique that works fairly well is the limited suspects approach. In my mystery, there are only nine people who might have been in the building, including the player character who is suspect (and knows that he is innocent). This particular adventure is keyed such that the referee can choose which of the suspects is guilty, and that choice alters the clues in subtle ways such that other suspects are eliminated. For example, the crime might have occurred in one of two rooms; there is convincing evidence that three of the suspects never went near one of those rooms, so if it is established that it occurred in that room, they're eliminated. This elimination approach may work best for a simple mystery. In a complicated mystery it becomes too much--Ten Little Indians works because the doctor is in cohoots with the judge, so when the doctor declares the judge dead we never question whether that was a lie. The remaining suspects/victims worry about whether the doctor might still be alive and killing people, but they never wonder whether the judge is still alive. Thus they, and we, have used the elimination process to get rid of the suspect, and in a game you probably don't want to do that.

You should also distinguish in your mind the whodunnit from the howtogettem. Most of the mysteries we usually discuss are whodunnits. Columbo shines as an excellent howtogettem: he knows who did it from the beginning, and so do we; the question that is out there is how do you prove the guilt so someone can make an arrest. I've noticed that Monk has been doing both sorts at times--most are whodunnits, but the episode with the airplane pilot who murdered his wife was a howtogettem.

I think, though, that Chris is right on most points. To create a mystery, you must start from the perspective of the criminal. He plans this crime and carries it out. Perhaps something goes wrong, and he attempts to cover it, or perhaps there is something he did not consider at the time. Somehow the player characters are made aware that the crime has been committed, and in a manner which demands their involvement in solving it. They proceed to solve the crime by uncovering the evidence incidentally left by the criminal. In creating the mystery, you must work backwards from what actually happened to what clues will exist because of it; then you have to make them available to your detectives beginning with the most evident clues and working in to those which can be found with a bit of digging.

Also, this discussion seems fixated on the murder mystery. Murder mysteries are good, but sometimes a good theft makes a good investigation, too. Banacek solved some wild insurance cases in which someone or something was taken under impossible circumstances. The mystery I'm doing is the theft from a museum of an ancient relic thought by people in the past to have had mystical powers. It works quite well, and has the advantage that recovering the object becomes the final proof of the solution.

Hope this helps.

--M. J. Young

I agree with pretty much everything M.J. wrote. A few points might be worth extended discussion, which is to say, are things we have both touched on in one way or another but really deserve a broader conversation:

M. J. Young wrote: The first thing that's not been discussed is the critical question: why do the players' characters care about the answer to this mystery? There's been an assumption from the beginning that "this is what they do"; but why do they do it?

Chris has several times suggested that you kill the character the players have come to suspect if it's the wrong character; that can work, but it faces a second set of complications. Why did the killer kill the person on whom suspicion has fallen? It's much more common for the second death to be someone who knows something but hasn't told anyone yet--whether because they were thinking of blackmailing the killer or because they haven't yet realized what it means.

Another technique that works fairly well is the limited suspects approach.

You should also distinguish in your mind the whodunnit from the howtogettem.

In creating the mystery, you must work backwards from what actually happened to what clues will exist because of it; then you have to make them available to your detectives beginning with the most evident clues and working in to those which can be found with a bit of digging.

Also, this discussion seems fixated on the murder mystery.

Lots of great ideas for me to process here.

Although, this one bit stuck out at me;

clehrich wrote: My point was simply that you shouldn't usually let the players get from stage 4 to stage 5, i.e. from the end of whodunnit to howtogettem as it were, unless they are correct about whodunnit.

Technocrat13 wrote:
Or perhaps it's just that I'm imagining that, giving advice to a GM that reads "If the players are ready to make a move on the wrong suspect, then make sure to throw something in the way to show them that they're wrong." would lead to some railroady, less than satisfying play. I mean,
[..]
It may not fit into the classic model as well, but I think I'd prefer that the detectives have a chance to fail.

Again thanks to Chris for those useful insights.

Regarding the chance to fail, it's a lot more complex and probably violates Chris' proposed stages model, but indeed you could have the players accuse the wrong person. It does happen in real mystery books sometimes, and can be interesting.

The best recourse in this event may be to have the accused immediately (or shortly thereafter) reveal some secret that is life-shattering for him to admit but clearly means that he could not have done it. He was in fact at that moment making love to his boss' wife, or engaged in a torrid tryst with his gay lover across town; or perhaps he was the person who at that moment was robbing the bank in Mudville, and although he hasn't been so much as suspected for that, he can prove his guilt in that robbery which will clear him of this murder.

Also, if we are moving from the whodunnit to the howtogettem and we have the wrong suspect, the obvious solution is that the evidence that should be there just isn't. At that point, in the process of trying to nail down the case, our investigators should find that one piece of data that cannot in any way, shape, or form fit the current suspect but has to force them to consider the true culprit. That might feel a bit like "we blew it"; but then, that's part of gamist play, right?

--M. J. Young