Sunday, November 19, 2023

Action Mysteries

Mystery scenarios are probably the second-most popular genre of gameplay in RPGs after dungeoncrawling. Despite this, quite frustratingly, most detective-y games don't provide much support for facilitating the actual act of investigation. Call of Cthulhu, for example, just has a skill system for resolving basic tasks, much like any game would. Same with Delta Green, same with Blade Runner, same with Liminal Horror.

But even if the most popular options dodge such a big question, there's actually a lot of existing literature on the subject of running mystery games. Tools, techniques, and advice abound (mostly in the blogosphere).

Some offer techniques for robust level design. The Justin Alexander famously has the Three Clue Rule and Node-Based Scenario Design. The disgraced Zak S wrote about Hunter/Hunted and Investigations-as-Dungeons.

Others give advice for refining the act of inspecting and uncovering information itself with smarter adjudication. Alexander also described his Matryoshka Search Technique which is a simple trick. Mindstorm wrote a post called Ransacking the Room which I find utterly brilliant. DIY & Dragons gave us Landmark, Hidden, Secret which I'm pretty sure Nintendo must have studied very closely to make the last couple of Zelda games.

Sean McCoy has argued that the answer lies in smart visual information design. Give the player a literal tool that helps them solve a mystery. I took a class in college called "intelligence analysis techniques" that had a lot of very gameable things I think could be a great foundation for a system (timelines, network diagrams, cross-impact matrices, analysis of competing hypotheses, etc.).

Still others reinvent the genre entirely by way of novel game design. Robin Laws built the GUMSHOE system to bypass the issue of players missing clues, which was further iterated on in Cthulhu Dark by adding some dice. The game Brindlewood Bay relies on "quantum mysteries" that everyone co-authors as they go along, which Prismatic Wasteland has also described.

Alice is Missing is a totally unique example because it's built around gamifying one specific mystery and set of ingredients that go into it, having players draw cards from preset decks in order to form the truth as they're discovering it.

The world of video gaming has plenty of insights, too. Game Maker's Toolkit has a really nice video identifying three types of detective challenge: investigation (uncovering and collecting information), contradiction (noticing inconsistencies and flaws in information), and deduction (interpreting available information to extrapolate new information).

These are all perfectly cromulent additions to the collective body of RPG detective theory. I am here today to offer a modest contribution of my own to that corpus. I'm going to refer to it as an Action Mystery.

Redefining victory

The conventional way of running an investigation is to have players acquire clues in a sequence, following a breadcrumb trail from scene to scene until they arrive at the truth. And we take it at face value that solving the question is the goal, and doing so means victory.

But it's important to recognize that the truth is not an end unto itself.

The whodunnit? story is a pretty familiar and simple formula, structured in three acts. Many of them focus on the big reveal as the climax, and save it until the very end. But notice how commonly others will have answered the question of "whodunnit?" by the end of the second act. This may sound counterintuitive. After all, if you're told that the main question of the story is "whodunnit?" then it only makes sense for the story to end once that question has been answered, right? And yet it doesn't. The classic formula requires a third act after that main question has been answered, because that answer alone doesn't actually satisfy the main tension of the narrative. Finding out whodunnit prompts an inevitable follow-up question: "now what?" What do you do with the truth?

Like, great. You know whodunnit. Congrats. So are you going to arrest them? Capture them? Kill them? Expose them? Stop them from killing again? And how do you go about doing that? What's your plan? What could go wrong? Are they going to resist this somehow? There's gotta be a showdown of some kind.

Thus, detective fiction pairs quite naturally with action. And when it comes to designing a mystery scenario, it's the action bit where I think you should begin.

The basis of my theory is for players to focus on solving a problem rather than them merely acquiring the truth. Piecing together a complete picture of something that took place off screen before the session began is cool and all, but it's not "victory."

In my model, every piece of the puzzle serves a very pragmatic function. Information should be useful. Finding clues and learning facts about the scenario is only relevant in as much as it helps you to solve the core problem.

Let me talk about an example before elaborating further on the theoretical principles.

The mystery of the mutant rat monsters

Tricks & Treats is an ongoing project of mine that I talk about a lot. It's a game about having spooky adventures on Halloween. You roleplay as kids and teenagers celebrating the best holiday ever, taking advantage of a rare night of unsupervised freedom, navigating the complex and unforgiving social landscape of adolescence, and investigating a mysterious horror that needs to be thwarted in order to save the day.

It's built for one-shot mini-sandbox scenarios, each one revolving around a major Halloween activity (trick-or-treating, going to a haunted house, attending a costume party, etc.). They always prominently feature a cast of NPCs thoroughly stocked with conflicts, rumors, and various hooks, and a unique "puzzle monster" that can't be defeated without gathering clues and forming a clever strategy.

The way I construct the scenario involves first making a monstrous problem with lots of traits and then seeding the sandbox with opportunities to learn about those traits. I call them "puzzle monsters" because they're more substantive than just a sack of hit points you randomly bump into and then bludgeon to death. They have immunities, weaknesses, unique attacks and side effects, agents, and a whole modus operandi. Crucially, their actions will inevitably result in doom by the end of the night, creating a ticking clock.

For example, let's look at the fourth scenario, Downtown Doomsday. Here's the problem: below the city is a colony of hungry, man-sized, mutant sewer rats numbering in the hundreds. To pass an EPA inspection last week, the local power plant dumped several tons of radioactive coal waste into the sewers. Now as rats swarm the streets, they threaten to infect the populace beyond containment levels.

What are some of these rats' traits?

  • They travel in swarms. Fighting even one alone is hard enough, but they appear in big groups.
  • They live in the sewer system.
  • They infect anyone who gets near them.
  • They're foreshadowed by clouds of flies and the smell of sewage.
  • They seek out candy and sugar.
  • Music agitates and confuses them, drawing their attention.

From this, I extrapolate lots of clues for each trait. There are trails of candy wrappers in some obscure locations, which lead to rat corpses if followed. Many NPCs are listening to the local radio station to participate in a radio contest, meaning that music is frequently playing out loud. Sometimes, lots of clues can be packaged together. Early on in the scenario, there are a handful of attacks against NPCs in their places of business. Curious players will investigate the aftermath and find a sewage smell, missing candy, and an unconscious victim who's come down with some illness.  So on and so forth.

Some information is easy to stumble across just in the natural course of celebrating Halloween. Some requires active probing, but the scenario has lots of rumors and indicators already pointing in the direction of interesting things to check out. As a player, you're expecting those to point towards fun Halloween-y activities. And sure, plenty of them do. But some instead lead you to evidence of monster activity. And of course, some information is learned by having an actual confrontation with the monster, which can be triggered by a random encounter roll.

As the night goes on, you gradually accumulate facts about the problem. The stakes steadily rise as the situation worsens. There's nowhere to run, there are no other NPCs capable of solving the problem, and attempting to just hide and wait will inevitably result in doom. So at some point, the party forms a plan based on everything they've learned so far and puts it into action. 

What makes this work so well?

  1. The investigation feeds into itself naturally. In a conventional mystery, the whole chain of clues is lying in wait from the moment you begin playing. The situation is inert, waiting to be discovered. Acquiring new information only happens when you correctly figure out how your current information can lead you to it. If you can't come up with anything new with the information you've collected so far, you're stuck. And in this situation, most GMs will instinctively resort to the clumsy "solution" of just handing them a hint, realizing that they must not have given enough info to begin with.

    But in my version, the situation is active. The rats are attacking more and more and the crisis escalates. The players want to disrupt this of course, but it also creates more and more evidence for them to learn from. There's still a strategic incentive to be proactive, of course. The group that learns crucial information quickly has more time leftover to make a plan, gather resources, and put it into action. More time means more possibilities for what you can pull off. But the group that isn't proactive, that's taking all night to learn the crucial information, isn't going to ever get stuck. Because new clues are being generated with each rat attack, there'll always be more and more opportunities to learn that info as the scenario goes on.

  2. There's no correct "order" to the clues. In a conventional mystery, there's a sequence of scenes you visit in order. One piece of evidence points to the next, which points to the next, etc. You have to hold back the big revelations for later, gating them behind smaller reveals first. If the players learn the big truths too early, your mystery falls apart!

    But in my version, the clues are merely sprinkled throughout the sandbox all over the place. Find them in whatever order, I don't care. My players found a lone dead rat first. By they just as easily could have checked out the closed construction site and come across the open sewer pipe with all of its radioactive waste barrels visible. Or they could've talked to the nice lady at the city mission and found out how many homeless people have recently been hospitalized. Any order is valid and no information could "spoil" the mystery by being learned.

    Clues in an action mystery are free to be completely disconnected from one another. One of the reasons this works is because clues are serving a different basic function. In a conventional mystery, the main purpose of a clue is to generate a lead. It's an arrow pointing in another direction. My clues almost never generate leads. Those come from elsewhere (the aforementioned rumors, hooks, encounters, etc. Oftentimes, just deductive reasoning). The main purpose of a clue is to instead generate a solution. Or at the very least, inform an eventual solution. They have very practical benefits. Learning that the rats want candy or that they're agitated by music are both directly applicable when formulating a strategy.

  3. There are still layers to the investigation despite this. In my experience, the first half of the "investigation" is usually just a matter of figuring out what the hell is even going on. You're minding your own business and trying to celebrate the holiday, after all. When weird stuff starts happening, it takes awhile to get a handle on the situation.

    Thus, early on you're only picking up on basics. "There are attacks happening," and "there are giant rats" and "a lot of people are getting really sick." Early on in the scenario, a biohazard barrier is erected in a perimeter around the entire downtown area, which players could stumble onto and get a hint that something funny is going on. That's still helpful to getting them closer to solving the problem, even if you can't "exploit" that information exactly.

    So there are some clues that are a little less practical. Not everything has to be entirely solution-oriented. Answering "what exactly even is the problem?" can also be a good purpose for a clue to serve.

  4. The players don't need to learn everything. In a conventional mystery, you want your players to follow all the clues and assemble the whole puzzle. That's how you "solve" it. If you're missing a major detail, then you haven't fully answered the question.

    But in my version, I don't care how much you learn. There's no "minimum" amount of information you're required to uncover before you can win the game. I suppose there is technically a "maximum" amount of information to learn, if you take your time to be thorough and investigate every last nook and cranny. But that's also usually unnecessary, and could even be detrimental!

    This is the key: you only need as much as info as it takes you to beat the monster. Hypothetically, the more investigating you've done, the more robust your strategy will end up being. But the clock is always ticking. You have to decide for yourself when you're ready for action. When you've gathered enough intelligence to make a strong plan.

  5. Despite all this, investigating clues is the core action defining the game. It's entirely possible, albeit unlikely, to speed run the scenario and go straight to the monster without taking the time or care to probe anything for answers. But how likely are you to succeed if you do that? And as previously mentioned, you still probably need to do a good amount of investigation even during a "speed run" anyway. You can't rush straight to the monster if you don't know what the monster is or where to find it. You'll always have to first go through the early stage of "what exactly even is the problem?"

    In a conventional mystery, the main thing you're asking yourself in every scene is, "how do I find the next clue?" or maybe "where is this clue trying to lead me?" You feel like you're supposed to be following a thread that's been left out for you, because you literally are.

    But in my version, your driving question is "how do we fix this crisis?" It's up to you to decide what kinds of answers would be helpful to have in the pursuit of that goal, and then infer logically where you'd get such answers. It's fascinating how much easier it is to focus on finding information when your goal is to use information.

[EDIT: now's a good place to plug a blog post by my friend Nick at Papers and Pencils which followed a similar train of thought]

You can make an action mystery!

The example I gave has a lot of constraints specific to its situation. It's Halloween-themed, the players are trapped in a bottle, it's made for one-shots, the scenario takes place over a single night, and there's a specific doom that the situation moves towards. But none of those are required for an action mystery.

Let's imagine that you want to run a more traditional mystery story. The players are detectives trying to track down a serial killer in the big city. That's as classic as it gets, right? You can still build this to be a solution-oriented scenario. Let's assume that the end goal is to arrest and prosecute the killer.

Therefore, all clue finding along the way should serve part of that goal. Every piece of information should help you to either...
  1. Get a handle on the scenario. Just establishing the basic facts is itself an important part of early gameplay. Instead of handing the PCs a case file of this known serial killer, start off with a mysterious death followed soon after by another. Both are weird and they have a compelling connection. This category of clue would be the kinds of things that are only there to show "these people were murdered" and "they were murdered by the same person" and "they weren't murdered because of, like, a personal crime of passion or a gang beef or any other 'mundane' reason people get murdered."
  2. Arrest the killer. Where is the killer located? Are they on the move? Can you anticipate their next move? Do they know you're tracking them? How will they respond if you confront them? What will it take to capture them without getting killed yourself?
  3. Prosecute the killer. You also have the extra burden of building a case against them. This is more difficult than if the solution could just be "kill the killer" like with a werewolf or something. So you should collect evidence that tells the story of their crimes, exposing their actions. How did they commit the murders? When? Where? Why? How can you prove all that?
Don't make a trail of leads that each point to the next lead. Instead, start with the killer's actions. Think logically about what sorts of signs indicate those actions. Embed it in locations and objects and NPCs and rumors circulating and events taking place and so on. Don't put them in a sequential order of any kind, don't make any of them essential or necessary to find, and put clues even in the places where players aren't expecting it so they can constantly be learning new information.

You should still be following a lot of the advice in those other resources I linked to at the beginning of this post. Come up with three clues for every conclusion you want the players to draw, have extra information that rewards players who inspect things thoroughly, etc.

The ticking clock is probably a necessary component for this formula, but the good news is that it's very flexible. The serial killer probably won't cause some apocalyptic doomsday or anything, but they don't need to. As long as there's the looming threat that they'll kill again, that's sufficient. It has the dual-benefit of 1) pressuring the players into taking action instead of slowly and thoroughly investigating every angle, and 2) steadily providing new batches of information in a way that doesn't seem contrived.

And of course, as in the case of puzzle monsters, you should come up with a good reason the players can't just go straight to the bad guy and thump them on the head. What's getting in the way of them doing that? Puzzle monsters of course often rely on immunities, strength in numbers, being part of a magical curse that has no head to thump, and so on. But it's perfectly fine if the reason is as plain as, "because you gotta find the bad guy first" or "because they have hostages." Hell, it could even be, "because there's more than one bad guy and each one needs to be tracked down and thumped separately."

So try coming up with a basic action mystery premise. Start with the problem that needs to be solved. One with urgency, and which can provide a steady stream of clues as the situation actively develops. The kind of problem that doesn't get solved unless you've taken the time to learn about it before jumping right in and trying brute force.
  • A cult is trying to summon Cthulhu with human sacrifices.
  • A house has become haunted and is tormenting its residents.
  • Agency operations keep getting expertly sabotaged by an enemy spy.
  • A series of heists are targeting... uh... I dunno, something more meaningful than just bank vaults. Maybe museum artifacts. I think that's easier for most regular people to feel invested in thwarting. Or, like, dogs.
Either you're trying to prevent the next time the bad guy strikes or the bad guy's strikes are building towards something you need to prevent.

If you don't mind the idea of moving further away from the "mystery" aspect, I would look to emulating The Wire. It's very gameable and is still very much an investigation adventure. The default action in the game is still "acquire information" and the main way that progress is made is by using information, so in gameplay terms you're still "playing detective." But the conflict itself is structured completely different from a Sherlock Holmes story. 

It features a team of investigators (just like the party of PCs!) trying to systematically take down an entire faction. There's no great big mystery that they're trying to get to the bottom of. No reveal of some hidden truth. They just need to gather the evidence necessary to build a case that would be strong enough to defeat the faction. Could they easily charge a bunch of the low-level street dealers for drug crimes? Sure, but that wouldn't affect the operation much as a whole. Could they put in a little more legwork to charge one or two of the enforcers with a murder? That'll make a bigger impact, but they'll train a new enforcer. What would it take to cut off the head of the snake?

The protagonists get a lot of use out of the classic "corkboard with strings" technique that's associated with the genre, but notice that it isn't just a trail of clues they're tracking. It's a plan.


I think it's pretty clear by now that the conventional approach to running mystery games has a lot of pitfalls. Each little snag has been targeted by great thinkers and designers trying to patch that one vulnerability.

Players might miss some of the clues? Justin Alexander says to add more clues and have a lot of redundancies. GUMSHOE says to eliminate the possibility entirely by simply handing the clues over automatically.

Players might not guess the right answer? Brindlewood Bay says to allow all kinds of plausible answers and then just go with the one they come up with.

Players might skip too far ahead in the plot? Alice is Missing spontaneously generates the answer during the last few minutes of the game by combining some of the cards that have been drawn leading up to that moment.

But the fundamental problem is that the victory condition is just too narrow a target. Finding 100% of the information the GM has included and interpreting all of it perfectly is incredibly unlikely even for a group of really skilled and intelligent players. Some players are happy to receive support to make this easier to achieve. But many are instead going to feel like the challenge is being gutted and their abilities of investigation are being undermined.

Solving crises, stopping villains, rescuing innocents, and thwarting plots are the kinds of things most players know how to do. Let them do that! It's easier to make such a goal open-ended and it invites players to strategize and take action rather than simply follow a path and let it lead them to an ending.

The players should be opposing the villain, not catching up to them. They should have the burden of creating the climax rather than reaching it. And if they're going to fail, they should fail because they couldn't prevent the consequences, not because they just "got stuck" and the plot couldn't move forward.



  1. Insightful post as always Dwiz. I just recently posted on genericizing common wisdom on good scenario structure (at in terms of Points (of interest) and Lines, and this complements my thoughts nicely.

    Many mystery scenarios are framed in a rather uniform way: each clue is either a lead to another node/scene, or adds to understanding the overall solution. Yet varied (multifarious) Points and Lines are important, not only for building to an action-driven third and final fact. If you're playing a game with swords, spells, or guns the players will want to use them.

    I want to highlight your reference to the 'small target' of standard mystery scenarios: that's a great framing for the problem. The role of mystery in play is interconnected with the central premises of playing to find out what happens, and that games are a series of meaningful (interesting and impactful) decisions. We can't be facilitating player agency if the scenario structure ultimately has a tightly prescribed 'success' (or indeed a single 'failure') state. Likewise as you allude to, this lacks robustness.

    In a way, an entire mystery scenario could be seen as the Information part of the Information, Choice, Impact doctrine of adventure gaming writ large. On this basis, a mystery is intapt play if it DOESN'T build to Choice and Impact. Perhaps the Action Mystery is the one true way?

    1. I must say I probably wouldn't buy into a "one true way" for /anything/, but that is a really interesting connection. I would say that McDowall's ICI doctrine aligns well with /my own/ preferences in gaming, but I try to be mindful of all the many things that different people get out of RPGs (even when it doesn't make much sense to me). My previous post was about the nebulous and mysterious kinds of intrinsic fun that players get out of just "playing pretend" in any form, even if there's no information, choices, OR impacts.

      I think for this particular example, the name "action mystery" tells you that it must not be the be-all end-all of anything. It seems almost antithetical to the subgenre of "cozy mysteries" ( which Brindlewood Bay is seeking to emulate. But then again, I /personally/ don't see much appeal in playing out that subgenre... so there you go.