![[community profile]](https://www.dreamwidth.org/img/silk/identity/community.png){kind=small}
Fixing Gumshoe
Jan. 9th, 2013 04:49 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
mnemeSo, I've long had some problems with the Gumshoe system, which underlies Gumshoe, Ashen Stars, Night's Black Agents, and Esoterrorists.
It's not the investigative subsystem -- despite the megabytes devoted to defending that system, -that- system is well set up for exactly what it wants to do, both in handling "this clue gets found; who found it" questions as well as "this clue might get found; is it worth it to people, or do they want to conserve their resources for other clues they find more interesting?" I like it, and engage with it. Sure--to anyone who has played a lot of investigative games with good GMs, it duplicates what a good GM will do in an investigative story anyway--making sure the PCs maintain a route to the story and adding extra redundancy to core clues (on the fly, if necessary) while having other clues affect the story, but ringing changes depending on how much the PCs do further than the necessary investigation. But there's nothing wrong with codifying what "good GMs" do, and everything right with it, as that's one of the better ways to train potentially good GMs into actually good GMs.
No, my issues are all with the General system. The approach seems good enough on the surface, at least for those who can stand some resource management. Whenever you make an action, determine a target number (between 2 and 8), then spend as many points out of a relevant skill as you like (you don't always know the target number, but you usually know it within a point, and the system gets wonky when you don't, so let's just say you do). Then roll a d6 and add it to your spend; if the result equals or exceeds the target, you succeed. If not, you fail and the action is wasted.
Now, there are some issues here that are a matter of personal preference. I'm not fond of systems where being good at a particular thing is only valid if you can spend resources being good at it (but here, the sting is somewhat removed by having many skills grant a special abiltity if you have more than 8 ranks in them). And on principle, I I'm not fond of systems where to get an extraordinary result, you have to bet that an extraordinary result will happen--I have the issue here, in Deadlands, and in L5R; I think one of the joys of rolling dice is that unexpected and extraordinary things happen -- but here, you decide what "unexpected" is before you try to roll the dice, and they never exceed your expectations unless you aim high (and you have little reason to do so much of the time, but see below). Neither of these are issues that need to be fixed -- they're just ways the system doesn't appeal to my particular tastes, and it -does- appeal to my tastes in other ways.
No, my issue is that the system is just incredibly binary. For the most part, the Investigative skills are the ones you use to find the adventure; the General skills are the ones that make sure things happen as you desire once you do--including your characters not dying. This means that for the most part, you can divide General rolls into two categories: Ones that matter, and ones that don't really matter.
The other important factor is that linear "1 point equals +1 on the roll" mechanic -- with (with the exception of points you spend to get success onto the d6 if the target is higher than 6, which are worth less) every point you spend out of a pool being worth 1/6 of a success (and, if you do, say, an average of 4 damage per strike, each point you spend is worth 2/3 (that is, 4/6) of a point of damage). Essentially, every point you spend is worth an exactly equal chance of success with everything -- in fact, the optimal strategy would be to spend enough to make every roll a guarunteed success, if it weren't for the healing mechanics (which let you spend a set of different general pool items to heal yourself, or more efficiently, your allies between combat) and the fact that not all tasks are equal in importance. Instead, tasks in Gumshoe divide into three rough categories:
1. Tasks that are crucial, which you spend enough to guaruntee a success (if it's worth that first point, it's probably worth the one that gets rid of the chance of failure, too).
2. Tasks that are fun/useful, but not worth spending any points on. You don't spend any points on these. (there are some cases where multiple PCs are trying the same thing where you might spend a point or two here to make the math work out better, but...usually not so much).
3. Tasks where you'd -like- to guaruntee a success, but you don't have enough points to do so. So you spend what you can and go from there. Or more likely, you don't try that task; you try a task using a pool you still have points in, reducing us to the first two options.
The problem I have might be obvious with that breakdown, but to put it plainly: In Gumshoe, played efficiently, I find that the only results that have any variability are the ones you don't care about -- the exception being if you are out of all relevant pools, in which case you're at the mercy of the dice and will probably lose/die unless there isn't much adventure before another refresh.
The reason this is an itch I feel compelled to scratch is that as problems go, it's an annoying wart in an otherwise servicable system--and one that once you've seen, can be hard to ignore.
The core problem is the linaarity of dice rolls/spends. That's the thing that means that once you're "in for a penny", you might as well buy the whole bank--why make a favorable bet when you can make a pile of bets and reduce any loss chance? I've in the past, looked at various ideas for solutions--you could use a bigger die and have different spends grant different bonuses, for instance. The problem is, any solution I've tried before now either fails at some crucial points (throwing off basic chances of success in one fashion or another), requires too much manipulation of the existing numbers in the game, or is simply too complicated. Often all three.
However, I think I've managed an elegant solution that handles the whole issue nicely: Instead of "1 spend = +1," replace this with "1 spend = +1 or roll +1 die, keeping the highest". The nice thing about this is that +1 die is actually strictly worse than a straight +1 (adding a die and dropping the lowest increases the likely result by +35/36). But Gumshoe isn't about your maximum total; it's about your chance of failure, and there the first +1 die (but not so much the rest; extra dice have a steep diminishing return) drastically drops your worst cases -- changing the chance of a 1 from 1/6 to 1/36 (although there you'd rather go from 1/6 to 0/6), a 2 or worse from 1/3 to 1/9, and a 3 or worse from 1/2 to 1/4.
This means that with my hack, there's a strong incentive, if you're spending on a roll, to spend first point on +1 die rather than +1. Sure, it doesn't help you guaruntee success at all (and if you're going to do that, you're not going to buy the die), but on a non-guarunteed roll with a 2/6 or better chance of success, it's got more bang for the buck than just buying a +1 (ok, for 2/6 chance, it's exactly the same, but spending one on a +1 and one on an extra die is much more efficient than spending both on a +1). And once you're buying at least one extra die, spends stop being linear--on the same roll where you'll avoid a 1/6 chance of failure, it might not be worth it to mitigate a 1/9 chance of failure instead. Of course, in theory, this significantly (but not massively) favors the players -- they're spending less but getting more bang for the pools they're spending. On the other hand:
1. They're actually accepting chances of failure, and occasionally failing.
2. The big bonus for that first spend encourages spending on rolls where you might not spend for a +1. So while they have more effective points, players are likely to not keep them for much longer.
So, what you you think?
It's not the investigative subsystem -- despite the megabytes devoted to defending that system, -that- system is well set up for exactly what it wants to do, both in handling "this clue gets found; who found it" questions as well as "this clue might get found; is it worth it to people, or do they want to conserve their resources for other clues they find more interesting?" I like it, and engage with it. Sure--to anyone who has played a lot of investigative games with good GMs, it duplicates what a good GM will do in an investigative story anyway--making sure the PCs maintain a route to the story and adding extra redundancy to core clues (on the fly, if necessary) while having other clues affect the story, but ringing changes depending on how much the PCs do further than the necessary investigation. But there's nothing wrong with codifying what "good GMs" do, and everything right with it, as that's one of the better ways to train potentially good GMs into actually good GMs.
No, my issues are all with the General system. The approach seems good enough on the surface, at least for those who can stand some resource management. Whenever you make an action, determine a target number (between 2 and 8), then spend as many points out of a relevant skill as you like (you don't always know the target number, but you usually know it within a point, and the system gets wonky when you don't, so let's just say you do). Then roll a d6 and add it to your spend; if the result equals or exceeds the target, you succeed. If not, you fail and the action is wasted.
Now, there are some issues here that are a matter of personal preference. I'm not fond of systems where being good at a particular thing is only valid if you can spend resources being good at it (but here, the sting is somewhat removed by having many skills grant a special abiltity if you have more than 8 ranks in them). And on principle, I I'm not fond of systems where to get an extraordinary result, you have to bet that an extraordinary result will happen--I have the issue here, in Deadlands, and in L5R; I think one of the joys of rolling dice is that unexpected and extraordinary things happen -- but here, you decide what "unexpected" is before you try to roll the dice, and they never exceed your expectations unless you aim high (and you have little reason to do so much of the time, but see below). Neither of these are issues that need to be fixed -- they're just ways the system doesn't appeal to my particular tastes, and it -does- appeal to my tastes in other ways.
No, my issue is that the system is just incredibly binary. For the most part, the Investigative skills are the ones you use to find the adventure; the General skills are the ones that make sure things happen as you desire once you do--including your characters not dying. This means that for the most part, you can divide General rolls into two categories: Ones that matter, and ones that don't really matter.
The other important factor is that linear "1 point equals +1 on the roll" mechanic -- with (with the exception of points you spend to get success onto the d6 if the target is higher than 6, which are worth less) every point you spend out of a pool being worth 1/6 of a success (and, if you do, say, an average of 4 damage per strike, each point you spend is worth 2/3 (that is, 4/6) of a point of damage). Essentially, every point you spend is worth an exactly equal chance of success with everything -- in fact, the optimal strategy would be to spend enough to make every roll a guarunteed success, if it weren't for the healing mechanics (which let you spend a set of different general pool items to heal yourself, or more efficiently, your allies between combat) and the fact that not all tasks are equal in importance. Instead, tasks in Gumshoe divide into three rough categories:
1. Tasks that are crucial, which you spend enough to guaruntee a success (if it's worth that first point, it's probably worth the one that gets rid of the chance of failure, too).
2. Tasks that are fun/useful, but not worth spending any points on. You don't spend any points on these. (there are some cases where multiple PCs are trying the same thing where you might spend a point or two here to make the math work out better, but...usually not so much).
3. Tasks where you'd -like- to guaruntee a success, but you don't have enough points to do so. So you spend what you can and go from there. Or more likely, you don't try that task; you try a task using a pool you still have points in, reducing us to the first two options.
The problem I have might be obvious with that breakdown, but to put it plainly: In Gumshoe, played efficiently, I find that the only results that have any variability are the ones you don't care about -- the exception being if you are out of all relevant pools, in which case you're at the mercy of the dice and will probably lose/die unless there isn't much adventure before another refresh.
The reason this is an itch I feel compelled to scratch is that as problems go, it's an annoying wart in an otherwise servicable system--and one that once you've seen, can be hard to ignore.
The core problem is the linaarity of dice rolls/spends. That's the thing that means that once you're "in for a penny", you might as well buy the whole bank--why make a favorable bet when you can make a pile of bets and reduce any loss chance? I've in the past, looked at various ideas for solutions--you could use a bigger die and have different spends grant different bonuses, for instance. The problem is, any solution I've tried before now either fails at some crucial points (throwing off basic chances of success in one fashion or another), requires too much manipulation of the existing numbers in the game, or is simply too complicated. Often all three.
However, I think I've managed an elegant solution that handles the whole issue nicely: Instead of "1 spend = +1," replace this with "1 spend = +1 or roll +1 die, keeping the highest". The nice thing about this is that +1 die is actually strictly worse than a straight +1 (adding a die and dropping the lowest increases the likely result by +35/36). But Gumshoe isn't about your maximum total; it's about your chance of failure, and there the first +1 die (but not so much the rest; extra dice have a steep diminishing return) drastically drops your worst cases -- changing the chance of a 1 from 1/6 to 1/36 (although there you'd rather go from 1/6 to 0/6), a 2 or worse from 1/3 to 1/9, and a 3 or worse from 1/2 to 1/4.
This means that with my hack, there's a strong incentive, if you're spending on a roll, to spend first point on +1 die rather than +1. Sure, it doesn't help you guaruntee success at all (and if you're going to do that, you're not going to buy the die), but on a non-guarunteed roll with a 2/6 or better chance of success, it's got more bang for the buck than just buying a +1 (ok, for 2/6 chance, it's exactly the same, but spending one on a +1 and one on an extra die is much more efficient than spending both on a +1). And once you're buying at least one extra die, spends stop being linear--on the same roll where you'll avoid a 1/6 chance of failure, it might not be worth it to mitigate a 1/9 chance of failure instead. Of course, in theory, this significantly (but not massively) favors the players -- they're spending less but getting more bang for the pools they're spending. On the other hand:
1. They're actually accepting chances of failure, and occasionally failing.
2. The big bonus for that first spend encourages spending on rolls where you might not spend for a +1. So while they have more effective points, players are likely to not keep them for much longer.
So, what you you think?
