I have a love/hate relationship with the Gumshoe system. What I love about it I love deeply, and the flaws in it can be overcome with some minor tinkering on my part.
Gumshoe is an engine for investigative games. So far Pelgrane has published four flavors of the system-- Esoterrorists (a modern supernatural conspiracy game), Fear Itself (a more classic horror game), Mutant City Blues (police procedural in a world of superbeings) and Trail of Cthulhu (a Cthulhu game, as you might guess). The engine is easily adapted to other settings. I've run MCB and a homebrew version trying to simulate spy movies). I've also played in a Victorian England take on the game. I can imagine it being adapted to anything from Roman Mysteries (like the Didius Falco novels), to more science-fiction-y treatments (maybe something like Flash Forward) to fantasy (perhaps something like the Lord Darcy stories from Randall Garrett).
Gumshoe begins with a couple of key premises. First, to quote the book:
Investigative scenarios are not about finding clues-- they're about interpreting the clues you do find.
That may seem like a pretty simple idea, but it has pretty broad-ranging implications for this game, and actually for games in general. We come to the classic problem in a dice/challenge based game. A player searches a scene for a clue, a necessary one to move forward. They botch their roll. Suddenly forward motion in the story halts. You can have the player roll again, but then you're just rolling until they get it. Or perhaps you have more players roll, which is just another path to that. Or maybe you have to have a deus ex machina appear and hand them the clue. Or perhaps you just cheat the system as written and hand it off to them.
The basic problem is that you have to jump through these hoops. Laws has a great analogy for this. Imagine a dungeon crawl where you had to make a roll to get to the next room. If you fail you can't go any further until some system jerry-rigging gets done. An investigation is a series of rooms, which in this case are scenes and locations.
Gumshoe gets around this by having two sets of skills (called abilities, but I always call these kinds of things skills): Investigative and General. I'll come back to the latter in a bit, but the former is what makes the game work. In character creation, the players have a pool of points to put into Investigative skills and the rules encourage the party to discuss so that the group has at least one person with a point for each skill.
And one point is really what you need, the rest is gravy.
Basically when a group comes to a scene- if they ask to use a particular ability at that scene and it is relevant, the GM tells them something. No rolls-- they're assumed to have basic competency in that field. The goal for the GM is to provide to the players what are called core clues-- essentially the keys to move on to the next scene or place. There may be several core clues, meaning that the investigation may move in different directions based on player choice.
To get those clues, a player need only describe applying the skill to the scene. Having a score of “1” means you know what you're doing. However players with higher scores may additional spend points from that skill in order to gain more information. These points refresh from session to session. The rules suggest that early on the GM suggest when players can spend those points, but later allow the PCs to make those choices on their own.
Clues/additional information gained from this use can be used to give more depth to elements, give an edge for later investigation, or provide other intangible benefits (like impressing someone). The rules suggest that these clues should never be necessary to solving the question or moving the scene forward.
All fine and good and you can certainly find this summary in various other versions in any review of the system on the net. The real question is: how does it actually play?
Very well. It takes players a scene of two to get the rhythm of the system, but once they get it, they find it liberating. They enjoy finding the core clues and they enjoy hunting around for opportunities to deepen their understanding by spending points to draw out more information. In short, the key concept of Gumshoe works, and quite frankly I can't imagine going back to anything else for running a straight investigative game. As well, seeing this kind of structure and having a sense of what core clues are has informed my other games, those not entirely focused on investigation, but having that as a recurring trope (Changeling or my H. Potter influenced Steampunk game).
The basic idea is that a system can find a way around these kinds of potential stumbling block moments in a game. They can be built into the system rather than having to resort to mechanical gymnastics to get the players moving forward. I like that idea-- and I think it is worth applying in a general way. If you find yourself consistently contorting something in the rules or in play, shouldn't you examine it to see how it can be done better?
I have only two small problems with this part of the game rules, and those are minor. First my general preference is to reduce the number of variable resources a player has to keep track off on their sheet. For things like Hit Points, Wounds, Willpower, Karma, etc-- I usually don't want more than three of these items. That's, however, really an aesthetic judgment on my part-- because in play it doesn't seem to cause problems.
The second problem is more about the nature of an investigative scenario-- regardless of whether you use Gumshoe or not. They're much harder to come up with and run on the fly. Gumshoe provides tools for helping you structure those stories and advice on how to construct clues and scene flow. But the bottom line is, these kinds of game require more prep than others. Of course this is coming from someone who plays fairly rules light and doesn't have write ups or stat blocks for the bad guys usually. If you're comfortable with that kind of prep, then this shouldn't be an obstacle. But again, that's the nature of a mystery-based game, and certainly when well-done the pay-off in play can be enormous.
OK, so what don't I like about Gumshoe.
Basically I dislike the way it handles all of the other skills, called General Abilities. These are skills used in circumstance where there's risk and challenge. Tests on these skills should only happen in dramatic situations, where the outcome is in doubt or opposed. So General Abilities include things like Athletics, Scuffling, Preparedness, and Shooting.
For any particular task, the GM sets a difficulty (from 2-8). The player rolls a single d6, and if the roll is equal to or higher than the difficulty, they succeed. The rating a character has with a General Ability is considered to be a pool. Before a roll on a particular skill, the player may opt to spend points from their pool on that roll and add that value to the total. If the declaration is not made beforehand, no bonus can be added.
In play our groups disliked this system immensely. They felt the die range ended up being too narrow and the mechanic of spending from a pool of points felt “gamey” to them. This was consistent across the three games (two one-shots in different settings, and one twelve episode campaign). Your experience with this system may well vary-- if your group is more used to these kinds of mechanics they may find no problem with it. Our group- made up of players who know Storyteller, Unisystem, our homebrew card resolution system, and True20- found it got in the way of everything else they enjoyed about the system.
I don't think that's a deal-breaker at all for me. The General Abilities rules and the Investigative Abilities rules exist in two different universes. It would be no problem to rip out the former and replace it with a mechanic from a system your players are more comfortable with. I'd avoid something heavy (like full-on nWoD), but other resolution systems where players know they have a skill and can see a concrete value to it could work better.
Overall I think Gumshoe has to be one of my favorite gaming products of the last several years. It has helped me see some of the systems of stories in a new way and given me a toolbox for playing out investigations in a satisfying way. I'd recommend any GM thinking about running mystery games pick up one of the Gumshoe books for a read through.
I'd also point people who want a deeper read on Gumshoe's ideas to look at this recent interview with Robin Laws.