Thursday, 20 March 2014

Monitors: a tangent on traits and difficulties

On the current working model for Monitors, characters use a combination of generic attributes and specific traits.

My original idea for traits was that each would provide a flat modifier to relevant attribute rolls. Dan mentioned a few concerns about this when last we met, which I will try to recall here. This post naturally presents my point of view, which I actually remember and can develop here, far more favourably than the stuff Dan said that I can't really remember.

  • It can be difficult to make modifiers reasonable. With a large modifier, traits will be all-important and nobody without a trait is likely to succeed. With smaller modifiers, someone with a trait will only succeed slightly more often than someone without it. I'm not yet sure how far I agree with these, but worth noting.
  • He also noted that with (say) a +5 training trait, there's no mechanical difference between Wits 5 + Mad Scientist (+5), and Wits 10. This is true enough, and he argued that the psychological difference is important.
  • The combination of difficulty levels and training gets tricky (is solving a differential equation easy, which a mathematician might say, or impossible, which someone without maths would say?). I was originally aiming for an objective difficulty system, but that may be problematic. In fairness, I think all difficulty systems are problematic, so it's a question of finding the least annoying one.
  • He also stated a preference for avoiding maths, which is reasonable enough but I'm less concerned about that. Doing arithmetic is a common enough aspect of games that I don't feel including it is a major drawback.
  • He thinks training should mean auto-success on routine tasks. I also think this, and attempted to design the system to ensure it, so no issue there.

His suggestion was that instead, traits allowed you to roll two dice and pick the most advantageous. Worth investigating. My immediate thought here is that this is less friendly to trait-synergising than modifiers, although you could of course roll multiple dice (this would be getting dangerously close to a dicepool). It also doesn't allow any differentiation between trait size, which modifiers easily do.

It's arguably equally friendly to combinations of helpful and unhelpful traits, since you can work out the difference, then roll that many extra dice and pick the most or least advantageous one as appropriate. So if you had four negative traits and two positive, you might roll three dice and pick the worst.

I could, of course, distinguish training from other kinds of traits. I'm a bit reluctant to do that, though. It seems clunky.


The issue with difficulties is this. Let's stick with training offering a +5 bonus for now. A character with training in Bureaucracy and a fairly low Wits 5 has a 50% chance of success in moderate Bureaucracy tasks. On an Everyday task (+10) they will always succeed unless something happens to make it difficult. However, someone without training would also succeed at 75%, even if the task is only routine for a trained bureaucrat.

One possibility is to make the training bonus bigger, say +10. This is mechanically reasonable, as long as we remember to set difficulties for untrained characters. This way, having training would offer a huge advantage (auto-success in many cases) but difficulties would still allow untrained characters to succeed a significant proportion of the time.

The problem here is, as Dan highlighted, more one of logic. There are many tasks where the difficulty basically comes from not having the requisite information, not from any lack of skill. If you don't know any Chinese, you cannot possibly write even the simplest message in Mandarin. If you have never read Macbeth, you can't write an essay about the play, no matter how gifted you are at literary criticism. If nobody ever explained what steps are needed to make cheese, you will not make satisfactory cheese. If you don't know the fifty-two steps for overhauling a SG-9 starship, you won't do it right. These aren't problems that can be solved by reasoning from first principles, or analysing what's in front of you, in any kind of reasonable timescale. However, some of these tasks are trivial if you do have the right training.

Dan's suggestion is that difficulty not be used at all. This is a bit of a radical departure, and a case where we seem to have different approaches. I can't, sadly, remember what he actually said, which is a shame as it was long and eloquent. Basically he didn't think it added much to the game. I think his argument was something along the lines that you should aim for a system to make things interesting, and that variable difficulty doesn't do that. If I could remember the argument, I'd be better able to decide whether I agree with it. I have the sense that I don't entirely.

We do seem to agree that a character should always succeed at a task that's routine for someone with appropriate training when there's no circumstance to create unusual pressure. Our discrepancy there is basically that I would like that situation to arise naturally as part of a difficulty continuum within the system, whereas Dan I think has a more pragmatic attitude and isn't bothered about difficulty.

At some point a game with die-rolling comes down to rolling dice, and if you don't have variable difficulty mechanically, then the variation basically comes in whether or not you need to roll. This seems to me to create a far sharper distinction that a graduated difficulty system, bearing in mind that in many cases characters will not have relevant traits, and that "no trait" and "needs to roll" are going to correlate pretty strongly. It doesn't really get rid of the problem of establishing what difficulty represents, since the GM still needs to determine what, for a given character, counts as difficult enough to require a roll. The stakes are pretty high (auto-pass, impossible, or roll on 25% chance?) so it's a relatively important decision. Oh, and there's also very little flexibility to reflect PCs' efforts to make their tasks easier, since your options are to either ignore those efforts as trivial or allow the auto-pass. The first discourages sensible in-character actions, and the second is going to undermine the importance of training. A bit of a blunt instrument there.

Monitors is intended to be an actiony, adventury kind of game. I'm not intending to aim for perfect simulation, only to give things a flavour of plausibility when it seems appropriate. Moreover, I've already noted repeatedly that Monitors are given a strong all-round grounding to prepare them to work as flexible, multi-skilled field agents who can turn their hand to anything. It's also a shiny optimistic sci-fi setting where you can reasonably assume that information is easy to get. As such I'm not particularly worried about sometimes handwaving that specific knowledge might be required for a task. Generally I think we can assume that most activities fall into two categories: stuff that Monitors can have a go at, and stuff that requires very specific plot-derived knowledge that they actually have to go and get. On the rare occasions that it seems important for only a dedicated astronomer to be able to try something, you can just rule that way. If something should be trivial for a trained character but genuinely difficult for anyone who isn't a specialist, and that bothers you, ditto.

Personally speaking, I definitely think that letting your character ace a task without rolling can contribute to feeling awesome and that their background and skills are relevant. I also feel like there is value in a game distinguishing stuff that your character can do effortlessly, stuff that might go wrong, and very long shots, as well as what's just not achievable. Pulling off a huge risk feels very different (better, typically) from something that's merely challenging.

Slightly inarticulate due to illness. Apologies. Dan?


  1. I can only half remember what I said myself (and I'm also feeling pretty grotty).

    Roughly speaking, my argument against variable difficulty (or, more specifically, against variable difficulty represented by modifiers in a roll-under system)is that I think modifiers can be lumped broadly into two categories:

    1) Large modifiers. Things like the +60 you can theoretically get for a Trivial task in Deathwatch.

    2) Small modifiers. Things like a +2 bonus in D&D.

    I basically have objections to both types of modifier, for slightly divergent reasons.

    Large modifiers bug me because adding a large modifier to a roll (particularly a large positive modifier, that is, one designed to make things easier) creates an expectation that things will go the way the modifier predicts, and if they don't it can be jarring or anticlimactic. If what I'm doing is so easy I get a +10 to my roll, why bother rolling at all? And conversely if what I'm doing is so hard I'm getting a -10, perhaps it doesn't make sense for me to succeed at all. I think often GMs use large modifiers as figleaves because they can't quite bring themselves to just say "yup, that works" or "nope, that doesn't work" and as a GM I would often find myself getting frustrated when players failed despite large positive modifiers or succeeded despite large negative ones, so I've taken to just making it automatic.

    Small modifiers, by contrast, just seem too fiddly to be worth it unless it's something long-term (like combat). On any given die roll a +2 modifier has a 90% chance of having precisely zero effect, and as such strikes me as not really worth implementing.

    Somewhat confusingly, I have less of a problem with variable difficulty, even through it winds up being mathematically identical. Think the difference is that in a variable difficulty system, your dice roll directly translates into a measure of how well your character performs, whereas in a roll-under system it doesn't necessarily. So difficulty becomes "your character would have to perform well to do this" while modifiers are "your character's chance of doing this is reduced", which is a different thing.

    1. Okay, I think I got that, and I take your points. I do tend to agree on very large and very small modifiers, although I think it's worth noting that small modifiers that crop up frequently will end up having an effect, in the same way that a slight change in stats is significant.

      I think where we diverge is on things that are Quite Hard or Quite Easy. One of the things that grates on me about Call of Cthulhu sometimes is that all tasks with a given skill are equally difficult. This means that in many cases you only ever have a long shot, no matter how difficult the task ought to be.

      Tangent: 3.5D&D's skill section is pretty humungous, but it does at least give the sense that things do vary in difficulty, and that taking sensible options or precautions is useful. It also has the (for me) attractive feature that you actually get better at stuff as you succeed, with things that used to be a challenge becoming easy. In contrast I dislike the 4E skill system because so many tasks scale in difficulty with level.

      If what I'm doing is so easy I get a +10 to my roll, why bother rolling at all?

      I can see what you mean (at least, in d20), but then if you have a lousy skill of 1 in the first place, that is still a very big failure gap. This argument doesn't seem to hold up well to variability in skill. This is important in the system I'm proposing, because you might be an amazing pilot, but when you're trying to impress people at parties with your exploits, you're still rolling on Pilot + lousy social attributes. If this were not the case, then a character would succeed at anything by finding a way to make it about their training, and this seems likely to skew the playstyle massively.

      A thing that I think difficulty modifiers do well is handle low and high skills without trivialising the differences. Bonuses mean there are some tasks a low-skill character can usefully attempt. Penalties create challenges for high-skill characters, but also offer them an opportunity to be particularly awesome by doing things out of the reach of lesser experts.

      As a GM I would often find myself getting frustrated when players failed despite large positive modifiers or succeeded despite large negative ones, so I've taken to just making it automatic.

      Conversely, when PCs succeed at something despite the odds, that can be hugely enjoyable. That's not less true when the odds are from penalties than when they're from 1% skills. I think here it partly depends where the odds are coming from - if you've put in effort to create a substantial bonus than it can be frustrating to fail anyway. This operates kind of like a skill challenge where the objective is to avoid the original roll. In contrast, if you're just taking a punt on something then I think it's less annoying.