Saturday, 6 July 2013

Monitors: progress report and revisiting skills

Since I've now got 20 posts about this silly little project, and I'm wondering where to go next, let's quickly review what progress I have made.

  1. Decided to try and actually make this into a game
  2. Analysed reptilian biology for mechanics
  3. Laid out some basic keystones for the game: skills-based, level-based, adventure-heroic
  4. Roughed out skills
  5. Roughed out races
  6. Spent really an amazing amount of time on a body-temperature mechanic.
  7. Established likely archetypes
  8. Considered ways to make Monitors feel like part of a massive organisation
  9. Spend a huge amount of time analysing and selecting basic principles for magic in the Monitors setting.

Well, considering I've never written an RPG before, that seems like actually a fair amount.

Revisiting Skills

The things I suggested looking at next are:

  • Reviewing the skills/stats system in consideration of more recent posts
  • Deciding on an injury system
  • Thinking about tech
  • Designing some basic spells and tech in light of mechanics
  • Picking some numbers, however arbitrary, to use with said mechanics

On closer inspection, basically all of these boil down to "fixing the mechanics". This mostly means "fixing the skills", which were admittedly a stopgap measure.

The initial game mechanic ended up as a purely skill-based system where you roll under skill on a single die to succeed at tasks, with a block of 14 skills. There was no injury mechanic, and thus no discussion at all of things like combat.

It seems like I need to consider:

  • The skills list: is it the right size, and the right breakdown?
  • The mechanic: is a one-die roll-under system optimal?
  • The skills system: would another system, such as stats-with-traits, be better?
  • If, and how, skills and injury are to be linked.

Skills systems

At the moment, I have a fairly conservative list of 14 skills. This falls far short of substantial games like D&D (well, the editions with skills), Call of Cthulhu and so on, which means each skill is quite broad. However, it's still long enough that I can't readily use it with erosive damage, because a) tracking damage to 14 skills would get unwieldly very quickly, and b) the breakdown of skills doesn't really match up well to types of damage. It's not obvious what would damage Science, for example, and each skill mixes theory and practice. I also notice that none of them represents physical hardiness (though I could co-opt Strength). This all means an injury system is required.

Let's quickly see if there are any decent alternatives.

Stats and Traits

One obvious alternative to all this is an even more basic model (something Dan did mention). Drop skills entirely, and have a tiny pool of basic stats that's modified by fixed traits. This might look something like:

  • Learning: how much information the character has memorised and can apply.
  • Wits: how good the character is at analysing, resolving and reacting to situations.
  • Muscle: how physically strong, fit and healthy the character is.
  • Finesse: how good the character is at deft, fine movements, awareness of their movements and hand-eye coordination.
  • Perception: how keen the character's senses are and how well they attend to them.

In this model, everything would simply be rolled against an appropriate stat. To represent specific training and racial qualities, characters would take traits, which might range from Chameleonic to Ferocious, and from Sneaky to Mechanic to Necromancer to 35th Century Betelgeusian Neo-Punk Enthusiast. A trait would help with a roll either by providing a bonus, allowing a re-roll or something similar. Maybe there'd be choices of Homeworld, Racial, Background and Personality traits to make.

In the simplest possible scheme, all stats might be rated 2-6 and you roll under them on a D6, with a relevant trait granting a reroll. Yes, I think roll-under is simpler than roll-over with fixed target numbers, because then your stats go up as they get better, which is generally more intuitive. A slightly more complex version would involve target numbers, and a stat + trait + roll system. Obviously you could faff about with this for a while (D10, percentage, what numbers to pick...).

This model would probably work okay with erosive damage, especially with fairly low stat numbers. Flash bombs damage Perception, alcohol damages Finesse and Wits, neuroscramblers damage Wits and Learning, tranqs damage Muscle and Finesse...

Well, it'd probably work. Running up a list of traits might be a pain though, and allowing people to devise their own is asking for balance problems. Still, could be worth a try.

Dice Mechanics

Not really my territory, so some mistakes are likely. But let's see, nevertheless. At the moment, I'm down as using a D20-based, roll-under-skill mechanic.

One incidental point is that I quite like the idea of allowing auto-fails and auto-successes, and am a lot less keen on the D&D-style "1 always fails, 20 always succeeds" model. In a fairly bouncy sort of game, it makes sense to me that some things just aren't a challenge to the skilled expert, and others are flat-out impossible. Criticals tend to lead to experts failing simple tasks 5% of the time, and amateurs succeeding at the toughest task 5% of the time. Obviously you can try to treat these as "complications" and other amelioratory measures, but I think it's simpler without them. In contrast, variable skills and difficulty modifiers seem to allow for fairly smooth transition between "impossible", "difficult", "normal", "easy" and "basically impossible to mess up". It just codifies the sort of DM judgement that normally happens: you have Skill 15, a +10 modifier for doing an Easy task that's easily repeatable with no risks, and you're rolling a D20 - just succeed already. But it might not work well in practice, of course.

What else is there?

Percentile Skill

One obvious change would be to just swap this for the good old percentile. In many ways this makes sense. People can intuitively handle percentages: your chances of succeeding at a task equal your skill.

One downside is that it seems percentile skill actually makes a lot less sense if you think about it a bit. Or rather, it still makes sense, but only if you bear firmly in mind that the only thing it represents is your chance of succeeding on a task of 'average difficulty', however that may be defined. However, it's very tempting to start using the character's skill to determine the difficulty of a task. Only got 10% in Mechanic? Well, then fixing this car will be Difficult for you (-5). Got 80% in Mechanic? You'll have no problem fixing this car, it's Dead Easy (+15).

This subjective evaluation isn't wrong, but it distorts the very intuitiveness that's one of the boons of percentile dice. Low skill and high skill are both inflated out of proportion.

Another slight difficulty is that, reasonably enough, the intuitiveness of the system also highlights its flaws. People want to be Good At Shooting, and (as any Call of Cthulhu forum can attest) find it frustrating if their skills are capped. "But I want to play a crack shot!", they think, "why can I only hit a target 80% of the time?" This is a justifiable concern, though the extent will vary between games; I think a lot of it comes down to the difficulty in pinpointing exactly what it is that you're 80% good at. Hitting a human-sized target? Hitting a stationary human in good lighting conditions? Landing an effective hit on a target during open combat, with smoke and noise and movement distracting you? How will the DM interpret it? What kind of bonuses and penalties are you likely to get? How does it work for other skills, like Interpretive Dance or Nuclear Physics? In this regard, weirdly, I think the abstractness of non-percentile systems actually helps, because they avoid this kind of deceptive intuitiveness.

A final point is that this is a fairly heroic game, where performing great feats should be fairly common. This suggests that very skilled people should succeed reliably even at difficult tasks (i.e. those with penalties) - which suggests skills high enough to overcome penalties. There's a natural resistance to allowing skills of 125%, which simply doesn't apply to having Skill 25.

Because of this complexity, I'm not sure it would end up an improvement over the d20 version I've suggested.

Difficulty Pool

Another variant, venturing closer into Dan's territory, is a version of the same thing using a pool of cumulative d10s. Yes, sounds weird, let me explain. Normally, you might roll 2d10 and aim to roll under your skill. For a very simple task, this might drop to 1d10. More difficult tasks would add dice instead. It would still be possible to succeed, but low rolls would become more extreme outliers with every die.

This system adds a bell-curve; you're more likely to roll a mean result (11) or close to it on 2d10 than on 1d20. The more dice added, the stronger the tendency becomes. However, on easy tasks with a single die, a lot of variation is possible. Is that a problem? Without doing some serious mathematical analysis I can't say for sure, and this isn't really the time. It's a possibility, though.

Dice Pool

Option three: give up and adopt a dice pool. As long as I pick and stick to one means of measuring success (that is, either numbers rolled or number of X-or-higher) and avoid the issues of Hellcats & Hockeysticks, it's probably fine. You get, say, 1-4 dice in some area of expertise, plus a trait may allow you a bonus die or a reroll. The main advantages here are that dice pools handle degrees of success quite neatly, it's easy to handle bonuses and penalties (just add or remove dice), and it's decent for a high-skilled setting because rolling a single 4+ on even 1d6 is pretty good odds.

On the downside, I think it actually makes it quite difficult to model the distinction between high-skilled PCs and low-skilled NPCs. You need a significant gap in their capabilities, which dice pools don't necessarily lend themselves to. If the lowest possible hand is one die, the dimmest bandit has a 50% chance of solving Fermat's last theorem (4+ on 1d6). If the lowest is zero, that chance drops from 50% to 0%, which is a pretty steep distinction and doesn't sit well with me. Also, we'd want to ensure that even a fresh Monitor recruit has a better shot at things than an untrained civilian.

You can shift the odds easily enough - maybe rolling a 10 on 1d10 is a success? - but the more you do that, the more dice you need when PCs come into question. In this regard, non-pool systems have an advantage because you can simply create a gap between the base PC and NPC numbers, and it's fairly simple to make this fine (percentile) or coarse (1d6) as you please.

Of course, there may be ways around this that I haven't spotted. I'm really not up on dicepool games. I could, of course, have NPCs simply work in a different way, but then I have to design two systems.

The TL;DR here is basically: I need to settle on a skill system. I'm dithering between "what I've got now, plus injury system" and "short stat-based system with stat damage, plus traits". The dice mechanics are important but secondary. Still honestly not that sure where to go next - perhaps drafting a version of each skill system to see how it'd work?

No comments:

Post a Comment