How I run investigation games (part 1: prep)

A really juicy mystery, with the cool feeling of piecing together clues and coming to the correct conclusion, is one of my favourite things in roleplaying. It’s also something that I feel isn’t well delivered by existing RPG systems. Here I’m going to talk about my approach to building a mystery and enabling real investigation.

This isn’t the first time I’ve explored this terrain. Back in 2013 I talked about how mysteries are like stones falling into a pool, creating ripples. And I went on to talk about how investigation isn’t just about clue detection, but about deduction and reaching conclusions. But I stopped short of talking about how to construct a satisfying mystery, which is what I want to do now.

Just for the moment, let’s assume I have an ok system that will cover the business of discovering clues, and an ok premise that make sure the players want to investigate this mystery. I may come back to these later, but let’s imagine they’re solved problems for the purposes of this article. Let’s also assume I’m running something that has a substantial investigative focus, so there’s more than just one simple mystery to solve.

I then create my mystery in a number of fairly discrete steps:

  • Decide what the fundamental driver of the mystery is. Something like “There’s a cult trying to summon a demon through a series of ceremonial sacrifices”, or “House Rukh are planning to assassinate the governor and take over the planetary government”
  • Generate from this driver a series of events. These can be past events which the players are (presumably) going to be investigating, or future events which the players are (presumably) going to be trying to avert.
  • For each past event, I generate a footprint, that is, a set of clues which are out there waiting to be discovered by the players.
  • The footprint breaks down into physical clues and witnesses, which are obviously investigated in different ways. Each of these is amenable to assigning a location and/or time. I’m also thinking about the ways in which the players might discover the clues, though I’m leaving myself open to other ways as well.
  • For future events, I generate a timing and/or trigger, some consequences, and (in case the players don’t find out about it until after it happens) a footprint, exactly as for a past event.

For instance, let’s look at the cult example:

  • For events, I decide that the cult has already sacrificed two victims. One of them was pursued through a particular district in the city in the night, and then murdered in a junkyard. The other was killed previously and more quietly, in their apartment.
  • The pursuit generated some witnesses along the route it followed – people who heard screams for help and some who looked out of their windows to see a group of figures pursuing the victim.
  • Both the murders generate a corpse, some messy bloodstains, perhaps a footprint. They also include the identity of the corpse – for the junkyard murder that may not be obvious, while the apartment victim (if the players discover it) is in their apartment so probably can easily be ID’d.
  • The junkyard murder will be reported, which is the trigger for the players’ investigation. The apartment murder will likely lie fallow for a while, but might show up later.
  • I also create three future events: a near miss where someone is cornered by the cult and nearly killed, but escapes by jumping out of a window; and a murder that involves an initial kidnapping and the victim being brought to a specific site for the final sacrifice. Perhaps the near-miss will report in to the authorities and the players can find out about it that way. Perhaps the kidnapping will be reported, perhaps not.
  • At this stage I might also add in some kind of link between the various murders, be it geographical (the locations form a shape on the map, with the final sacrifice in the middle) or social (the victims are all highly religious people, say), or whatever.
  • If the final sacrifice is completed then the demon will be summoned and a whole new set of events will be generated after that (but I don’t bother thinking about that right now, because I’m expecting that the players will stop the sacrifice happening and/or kill the demon after it’s summoned.

Once I’ve planned all this out, I’ll review what I’ve got to make sure there’s enough there to give the players a fighting chance of cracking the mystery, but not so much that they’ll solve it in five seconds flat. I can add or remove witnesses and clues until I think I have got that right. Of course, my future events ensure that, no matter what happens, the players will have something to do. If time passes and they haven’t made progress, the next event happens.

I’ll then break the information down into a number of components I can use:

  • A timeline of events
  • A list of locations with clues that can be found there
  • A list of characters with motivations, information they might have and any key abilities

Once I’ve got all that in place, the game more-or-less runs itself. The players move from location to location as prompted by clues and/or a future event becoming a present event. Perhaps they discover clues which help them to get ahead of the timeline, perhaps the timeline runs ahead of them and they’re forced to confront a scary situation unprepared.

I’ll talk in a future article about how I use this prep in practice.

This article is supported through the Black Armada Patreon.

Become a Patron!

Investigating investigation

Due to popular demand (well, Blackrat demand), I am going to write a bit more about investigation and how it can be systematised.

Fundamentally, investigation in roleplaying is about searching for and discovering clues which can be used to draw conclusions about something happening in the fictional world of the game. If you over-mechanicise searching for clues (for example by making discovery automatic, as with Trail of Cthulhu) then you end up with something that feels like railroading. If you elide the discovery of clues and drawing of conclusions (something it’s easy for a GM worried about whether the players will be up to the task of deducing what is going on) then you end up with exposition rather than true investigation.

We can break it down further
– Following leads to direct the search for clues towards particular people, places, groups, events and so on.
– Finding the imprint left by the events one is investigating, especially if it is concealed (fingerprints, footprints, CCTV footage, a statement by a witness…)
– Identifying a pattern, an anomaly within a pattern or a definite lack of pattern (for instance, all the victims belonged to the same religion)
– Interpreting the imprints and patterns found so far to draw conclusions about what might have happened
– Making the link between an imprint, pattern or conclusion and a new lead, widening the investigation (a person whose fingerprints, footprints, etc were left at the scene; the local temple of the religious group being victimised, etc)
– Drawing a solid enough conclusion to allow a confrontation of some sort (arrest the murderer, grab the lost artifact, reunite the father with his lost daughter)

Each of these can in principle be broken down into appropriate skills or abilities (forensics, interrogation and so on), if the game system wishes, but most game systems don’t really use the above breakdown. Most systems concentrate almost exclusively on the second bullet: finding the imprint through awareness tests, while some that are more focused on investigation also move on to the third and fourth bullets, allowing skill use to draw conclusions such as “these footprints were made by a very large man”.

To me, at the heart of an investigation game is how you move through these steps, and how the elements highlighted in bold get you from one step to the next. What makes an investigation game enjoyable for me as a player (and vicariously as a GM) is taking those steps myself, not having the system do it all for me; but of course, since I am not in fact a forensics expert or arch-interrogator or whatnot, it must give me just enough information to make it possible for me to draw conclusions and decide on appropriate leads myself, rather than spoon feeding me.

A stone’s throw away from the answer

An analogy I often come across when describing investigative games is the “trail of breadcrumbs”. That is, a linear series of clues each of which points to the next in the series. It’s an approach that reaches its apex in the Gumshoe system, which advises the GM to write a “spine” of scenes, each of which contains a core clue which is necessary to progress to the next scene. The game makes discovery of core clues automatic for any character with the relevant skill, solving a genuine problem with investigative games, which is that they can stall when the players miss an important roll.

This is not an approach I subscribe to. Notwithstanding the fact that the “trail of breadcrumbs” is rather demeaning towards the players, suggesting they are simply mindless birds pecking their way to success – well. That is exactly what they are in Gumshoe, it seems to me. The system deliberately removes any element of challenge in the process of discovering the Truth, leaving the players with the job of describing how they do it. Whether they peck at the breadcrumbs furiously or idly, I suppose.

Let me give you my own pointless metaphor for the process I follow. Imagine the villain of the piece is standing on the shore of a lake, throwing stones. The stones create ripples, which spread out in all directions and persist for a long time.

If the stones are the villain’s actions, and the ripples are the clues they leave behind, then you should start to see where my analogy is going. The players are presented initially with some information about the “ripples” – the key evidence that starts their investigation off. Now they are in a position to look for more ripples. They begin to be able to piece together where some of the stones fell. If they are watching closely perhaps they can even spot some stones landing. But I as GM don’t plan out which bits of the ripple they’re going to find, or which stones they will discover the location of. Or in other words, I don’t know which evidence they will discover or what clues they will deduce from it.

Instead, I make sure that there are plenty of stones being thrown, with a varied size of ripple, so that I can be reasonably sure that they will eventually figure out who is throwing those stones. Finally, the stones continue to be thrown as the investigation is ongoing, generating still more ripples. By which I mean:
– My villain is doing lots of stuff
– He is, therefore, leaving lots of evidence behind for the players to find
– Some of it is really obvious, some less so, so there is room for skillful play
– My villain carries on doing stuff while the investigation is going on, so the trail never goes completely cold, and (this is important) there are consequences to failure

Think about the difference between these two approaches. Under the Gumshoe approach, no matter what the players do, they will uncover the mystery, and without any role for intelligent deduction, clever investigation or even just plain good or bad luck. The game (I read Trail of Cthulhu) strenuously denies it railroads the players, but this feels like a game on rails to me. Even a standard “trail of breadcrumbs” (not Gumshoe) game will feel a lot like this.

Meanwhile under my approach the players are required to use their eyes and ears and brains to piece together what happened. They don’t get any free passes. There is scope for them to solve the mystery slower or faster, and there are consequences if they fail. It is unlikely the players will fail because there is lots of evidence and the villain doesn’t disappear off the radar, but they can still screw up properly, allowing the villain to commit more mayhem, and conversely they can score a roaring success, catching the villain early.

A trail of breadcrumbs investigation means you are roleplaying an investigation, but not actually doing one. The stones and ripples approach means you get to actually investigate, not just go through the motions.

Final thoughts: my approach is not without problems. It’s a lot of work, and it requires the GM to think on her feet about every player action and what clues it might uncover. It can mean an unpredictable game length. The GM must pay a lot of attention to whether there is enough evidence to go on (so they will probably succeed) but not too much (so it just feels like childsplay, ruining the challenge). I think these are a price worth paying to get an approach that feels like real investigation.