Investigations and the Frame Problem

One thing which perpetually gets discussed in relation to Call of Cthulhu and many other investigation-oriented games is that of skill checks to find clues. The argument goes something like this: players need to find clues to progress through an investigation. The rules-as-written indicate players need to pass skill checks (often Spot Hidden and Library Use in CoC, but many others are possible) in order to find the clues. So what happens if they fail their skill check? In short, the progress of the players can potentially be stopped dead by a bad string of dice rolls. Is this not a flaw in the structure of the game?

(To be fair, the current edition of CoC attempts to redress this in a couple of ways, firstly by allowing players to push a skill, gaining a reroll at the cost of a potentially negative consequence, and also by the provision of the optional rules for spending Luck. But the general point still stands.)

Personally, I’m finding myself increasingly of the belief that the sweet spot of this kind of game-play isn’t the finding of the clue, but what happens next: the interpretation of it. My main group has recently started playing Night’s Black Agents, one of the Gumshoe family of games, where clue discovery is essentially automatic, and feedback from the players after two or three sessions has been overwhelmingly positive: it’s interpreting the clues and then deciding how to act on them that they enjoy, not rolling the dice to get the clues in the first place.

This has led me to think more about how to handle this kind of situation in other games. Some of the younger members of the family recently expressed an interest in D&D, so I’ve been playing a slightly-simplified version of 5e with them. We’ve mostly been playing traditional crawls so far, and very quickly they figured out the wisdom of listening at doors and cave-mouths before deciding which way to go. This is, obviously, smart play, and as such should be rewarded. Giving players clues about what they might find in a certain direction is also one of the things that stops a crawl from turning into a random amble around the map.

So far I’ve been calling for Perception checks whenever they listen or search for clues, but I’m now thinking of just giving them clues automatically, and save an actual check for when there’s something important in an area but they’re not actively looking for clues. As I say, it rewards good play and can potentially dispense with up to a dozen or more incidental dice checks.

Clue-interpretation in games like 5e is a little trickier than in a game like Call of Cthulhu or Night’s Black Agents, which share an ostensibly real-world setting. All my players to date, so far as my research has been able to uncover, are from the real world and have grown up there, and as a result have a vast intuitive knowledge-base of how it works. This same knowledge-base doesn’t exist when it comes to other settings, which can particularly impact investigative games.

This has some similarities to what AI researchers call the Frame Problem, which is one of the things they have to deal with when trying to get an intelligent system to interact with the real world. Even the smallest of children has an acquired understanding of how the world works – things like gravity and basic laws of physics, down to principles as basic as solid objects not being able to pass through one another. An AI lacks this fundamental frame of reference, so all these facts have to explicitly incorporated into whatever instruction the system receives.

The same really applies when you have players running characters in a setting they’re not especially familiar with. Maybe you’re running a Marvel-themed supers RPG, and most of the players have only seen a couple of the movies and not really read the comics: they may not recognise the clues suggesting the modus operandi of a particular villain, even though their characters probably would. We did contemplate playing Modiphius’ Star Trek Adventures a little while ago: I think this is a really solid and fun game, and my players were open to the idea, but the fact that none of them is really that familiar with Star Trek and its lore was a real concern to me, for just this reason: I anticipated many slightly confused expressions and queries along the lines of ‘The Prime what?’ and ‘What’s a Cardassian?’

Going back to Gumshoe for a moment, one system I own but have never played is Mutant City Blues, in which the PCs are police detectives in a city where a significant minority of the population have developed superhuman abilities of different kinds. To add an extra level of detective work to the game (and presumably also as a balancing mechanism), there are limitations on which combinations of mutant powers are possible (tough skin and superhuman strength often go together, tough skin and flight much less frequently). This is the kind of thing that the PCs would instinctively know, but for the players it involves either frequent reference to a moderately complex diagram, or the GM just telling them this stuff.

The other alternative, I suppose, is to call for some kind of Knowledge check, but once again we have the potential of bad dice rolling producing – for instance - a situation where a Starfleet Captain doesn’t know the name of the Klingon homeworld, which just seems to me to be silly and unhelpful.

There’s a danger here, though. Taking dice-rolling out of clue-acquisition can help a game to flow. Taking dice-rolling out of clue-interpretation as well – which is what removing Knowledge checks can mean, in an otherworldly setting – creates the possibility of the players having virtually nothing to do beyond letting themselves be railroaded through the scenario.

I’m talking about a situation where the GM is providing not only the clues but also their context and significance. This needn’t necessarily be fatal to a game or scenario, but the thing to be wary of is where you find yourself not only providing information but also providing the players with the conclusion they draw from it.

Imagine the following game situation: the character are called to a crime scene where someone has met a weird and untimely death. ‘We investigate the crime scene,’ say the players. No dice roll is called for, for reasons previously discussed. ‘It looks like a guy called Marcus killed him,’ says the GM. This is an extreme example, obviously: what would more likely happen is that the players would ask how the victim died and the GM would describe this, adding that it fits the MO of a known associate called Marcus.

I hope you’ll agree that this is still not very good GMing, whether or not Marcus has actually been framed: there are no real clues, no interpretation, just the GM presenting the PCs with an answer. Off they go to find Marcus, not actually having had to do anything but make the most obvious choices and pay attention to the GM, hardly needing to engage their brains at all.

How could we make this better? Probably by making it more granular and specific and factual, and leaving the players to make their own inferences. How did he die? Stabbed in the back with some kind of punch-dagger, apparently. (Whether or not you reveal that the placement of the wound indicates a left-handed killer is probably a judgement call, though which rules you’re using may also inform this.) Hmmm, think the players. Have we heard of any left-handed villains using punch daggers? Yes, says the GM, and (ideally) gives a hand-out listing three possible perps. Hmm, think the players, again, and check the guy’s phone…

You see how this works. In a real-world setting this kind of thing is really a no-brainer, but in an otherworldly context – let’s go with Star Trek – players may not be aware of the capabilities and limitations of their characters, equipment-wise, in addition to not being familiar with the setting. Starfleet are investigating a strange murder. ‘We check his phone’ is a no-brainer for a real-world game, but what’s the Trek equivalent of this? Are the players aware of it? Do they know they can check for residual transporter traces? Do they know that different kinds of transporter leave different energy signatures? If this kind of thing is essential to the plot, and the GM has to keep informing the players of their options every step, then they’re simply being led through the plot by the hand, barely doing any actual investigating of their own.

I’m not sure what a satisfying solution to this would be, but it may explain why investigative scenarios tend to either have a real-world setting, operate on a fairly basic level, or rely on players having an in-depth knowledge of the setting. The key thing to ensure is that players have a solid frame of reference in which they can confidently apply whatever knowledge they have, make inferences, and draw their own conclusions before deciding what to do next. That’s where the fun is.