How To Run OSR-Style Investigative Adventures

People love the idea of investigative adventures, but there's a lot of confusion about how to do them, or whether they can be done OSR-style– that is, sandboxy, player skill-based, difficult, failure is an option, etc.

The only OSR game built around investigative adventures that I know of is Silent Legions, and as much as I love Kevin Crawford, he seems to think the answer is no, you have to discard a lot of OSR principles to make investigations work, because without some rails most parties just won't solve the mystery.

I think he's wrong.  Here's how you can make investigations sandboxy, challenging, and OSR-style while still making it highly likely that the party will (eventually) complete the investigation.

Artist: George Doutsiopoulos

Four Things Not To Do

1.  Don't Railroad

It's generally agreed that railroading is bad so I won't belabor that point, but a lot of people think that you have to throw this rule out the window when designing investigative scenarios.  You don't, as you'll see.

Let's be clear about what railroading means in this context though.  A mystery has one definite solution– that is, a definite truth about what really happened.  Railroading means a) having a clear plan for how players reach that solution, or b) a clear plan for what they do with the information when they get it.

In other words, a good investigative adventure should be a bit like a hexmap or sandbox setting in which the players have a clear destination in mind, but any number of ways to get there, and multiple options for what to do once they arrive.  

2.  Don't Organize Things By Time (Mostly)

This is generally just good sandbox design advice, but again it gets thrown out a lot where investigations are concerned (See: Silent Legions).  Organize your scenario by things to be investigated– locations, people or organizations. Not by time, or by "scenes."

There are exceptions though, where a certain sequence of events may be logically required within the fiction.  Like, the cult has to murder 7 people in ways that match the 7 deadly sins before it can summon Balphegor.  But note that even then, it's only dictated that things have to happen in that order if they happen at all, and not that they definitely will happen.  Preventing them from happening is often the point.

It's also fine to have some if-then statements about stuff that might happen in a certain sequence– like what happens if it gets out that the king's second son is the killer.  Still not railroading as long as the players aren't forced into either solving the mystery, or publicizing the truth when they find it.  

3.  Don't Make Players Roll To Find Clues

Don't let the whole adventure come to a screeching halt because of bad rolls.  Either make finding the clues more or less automatic if they're out in plain sight, or have players find them using player skill.  

Note the use of the word "make" though.  It's fine to allow players to roll to find clues as a backup to using player skill, the same way they can make search rolls instead of manually searching.  Like search rolls, it should just take more time and be less certain to succeed.  

So like if there are fingerprints on the windowsill, you could allow any player to say "I search for fingerprints on the windowsill," but also allow a player with a forensics skill– or any player if you don't specifically have a skill for that– say "I search the room for fingerprints," take a turn or an hour or however long it takes, and see if they find the ones on the windowsill. 

4.  Don't Put In Red Herrings Or Dead Ends

Your players will create their own red herrings by misinterpreting clues, or seeing some random flavor detail as a clue.  Deliberately creating red herrings is redundant.

That said, if the investigation is part of a broader campaign, you can have some side branches that connect to other things in the game world, but don't help solve the mystery.  This makes the investigation take longer, but these side branches at least have a reward so they're not a total waste of time.  

Five Things To Do

There's a general theme here: redundancy.  Having a massive amount of redundancy makes it unlikely that the investigation will ever come to a dead end.

1.  Have Multiple Pathways To The Conclusion

Think of investigations as consisting of a series of "nodes" that players can investigate.  These are usually locations or people, but they could be organizations or objects too.  

Justin Alexander's series on node-based design does a great job of illustrating how this looks; here's one example of how an investigation could be plotted.

Let's assume that the blue node is the beginning of the investigation, and node H is the conclusion.  Every connection between two nodes indicates clues, so node A has clues leading to nodes B and D, and so on.

That means there are multiple ways to get from the blue node to node H.  If you're in node A you could find clues leading either to B or D, and so on, so no particular clue is absolutely crucial, and neither is any particular node other than the first one (which you don't have to work to find) and the last (which you have many ways to get to).  

2.  Have Multiple Clues Connecting Any Two "Nodes" In The Investigation

Justin Alexander also has something called the Three Clue Rule, which says you should have three clues pointing to any conclusion you want the players to draw.  

I like that as a rough guideline, although the exact number of clues can depend on how vital a particular conclusion is.  It's alright to have a non-critical "easter egg" with only one clue leading to it.  

I disagree with how Justin combines the three clue rule with node-based design though.  He thinks you can split the three clues up between every connection leading out of a given node, so like node B would have just one clue each leading to nodes C, D and E, and so on.  I think that's too little, plus it's pretty easy to rapidly come up with a few ideas for clues, so I'd sprinkle more around.

Given that it's possible to move towards, perpendicular to or away from the solution to a mystery, I'd bias the number of clues towards going closer to the solution.  So like in the above chart, I would have 2-3 clues for every "downward" connection like from A to D, 1-2 for every "sideways" connection like from B to C, and only one for every "upward" connection, like from D to A.  

This means the players can potentially find a lot of clues and maybe get lead in the wrong direction, but on the whole those clues usually take them in the right direction.

Bear in mind that players often will have some ability to make an educated guess about which clues are more likely to lead closer to the heart of the adventure.  Like if you're investigating a murderous cult and you find one clue leading towards a pizza parlor and one clue to a cavern deep under the city, you can infer that the first is likely a front and the second might be a major cult hideout.  

3.  Make Some Clues Really Easy To Acquire

The GUMSHOE system, which is billed as the ultimate investigative system, has this rule that clues are always found automatically and the challenge is purely in interpreting them.  It's fine to do that for some clues, but I don't like that as a rule for all clues,  since it negates half the challenge of an investigative adventure.  

Clues can be rated on two dimensions of difficulty: how hard they are to find, and how hard they are to interpret afterward.  

A clue can be obvious, like message written in blood on the wall.  Obvious clues generally take no effort to find.

It can be inconspicuous somewhere you probably would think to look if you thought long enough, like a fingerprint on the doorknob.  Inconspicuous clues can be found either with a highly focused player skill-based search (I check the doorknob for fingerprints) or a moderately focused character skill roll (I roll forensics to search the room for prints, or search the door for whatever is there.)

Or it can be hidden somewhere you might truly never think of, like a drug stash hidden in an empty protein powder jug in someone's pantry.  These can only be found through player skill, unless a truly exhaustive amount of time and effort is put into searching, and even then a roll might be called for.  

A clue can be clear in its meaning, like a record showing that suspect A ordered Uber Eats the night of the murder.  This takes little interpretation; the person was home when they ordered it (assuming there's no message saying like "we left it at the door because you didn't answer")

It can be vague and in need of some figuring out, such as a book about human anatomy.  Was the person in med school?  Are the pages related to the liver, which was cut out of the victim, dog-eared?  This has to be figured out with player skill; character knowledge may sometimes help in interpreting clues (you realize this book is geared towards first-year med students, not surgeons) but can't do all the interpreting for you.

Or it can be cryptic with no obvious interpretation, or at least the correct interpretation isn't obvious.  Like, a suspect has a lot of oddly-colored  red dirt under his nice loafers because he's been sneaking into the underground from his office, but the dirt probably wouldn't mean anything to you unless you had already seen the cave or tunnel it came from.  

Cryptic clues usually can't be interpreted on their own; they only make any sense when combined with other clues.  Or alternatively, they might have an obvious explanation that's wrong, and a less obvious correct one.

So combine these possibilities and you have a 3x3 chart of clue types, not unlike an alignment chart

The numbers are very rough guidelines for how relatively common each type of clue should be.  So the moderate-difficulty clues are more common by default.  If you do a lot of investigations and your players get good at them, shift this breakdown towards the hidden-cryptic corner of things a bit more.   

Have one path from the start to the end of the investigation that can be followed entirely via relatively easy clues; clues in the upper left four boxes and not the other five.  At least, if you don't have much confidence in your players' abilities; this gives them one path of least resistance they can follow.  

4.  Have a Fail-Forward Mechanism For When The Investigation Stalls

So if the party still can't figure out the clues and keep the adventure moving forward what then?  Well, the bad guys aren't just static fixtures in the world; eventually they get to take action too.  At any given stage of the investigation, have an idea of what the opposition will do if the party doesn't keep things moving forward.

Early on, if you don't make any progress, the cult kills another person.  Once you've started to worry the cult, any time you give them enough time to react, they send a hit squad after you.  If the party doesn't make something happen, the bad guys will.

And this is a fail-forward mechanism because it provides more stuff to investigate.  A new victim means a new crime scene with more clues.  A hit squad coming after you means you can investigate the people in it after they die, maybe interrogate a survivor, or follow one back to their hideout. 

5.  Be Willing To Let The Players Fail (Or Quit)

Hey, it's OSR.  Failure is an option.

Let the party fail or quit, but if that happens, do dig into it and figure out why.  Were your clues too hard to find or too hard to interpret?  Was the adventure not that difficult, but just too slow to keep their interest?  Are they just not into investigations?  Would they have liked more combat?  Was there no clear reward for taking part in the adventure?

Let the chips fall where they may, but learn what you can to design a better adventure next time.  That said, as long as you a) verify beforehand that the players actually are interested in investigative adventures and B) combine all four of the above rules to have massive redundancy against failure, this shouldn't happen much, if ever.