Let's Talk About Core Game Mechanics

[Gorgon Zee]

I'd be careful categorizing things as immersive or not. Some people find it immersion-breaking to understand the math because, for them, it pulls them out of the story and into the mechanics.

Not disagreeing with you — that’s why I like linear systems — there is no math to understand. Your skill is 80; you are 80% likely to succeed. No math (at least until you have to decide on extreme results!).

Anyway, even if I'm an expert swordsman I couldn't tell you the odds of my attack landing against a new, unknown adversary, except in the most general sense. So at first glance I disagree with that premise.

I’m going to ignore your “unknown” adjective, because by definition, you cannot know your odds against an unknown opposition! Instead I’ll speak to the case where you are opposing against a known opposition.

For a while I was an expert TKD competitor, and I pretty much could tell you the odds of landing a blow against a newbie, an average club member, an average state champion and an Olympic contender. I’d regard myself as expert software developer, and I do regularly estimate the success odds of tasks in terms of completion within X days and generally they are pretty accurate.

Of course, not to the exact percentage, or probably even 10%, but enough to get a general feel. And definitely enough for usual cases. So I do think that a system should be able to do that without, as you say, me needing to do math.

When someone plays BRP, they rarely feel a disconnect between their skill numbers and the results they see in play. They can tell if they are unlucky and they don’t spend time out-of-character doing math to work out if something is a good idea. Contrast with the One Ring — is it better to roll with favored (advantage ) and 2 skill dice or make a regular roll with three skill dice if I need a special success? I’ve been running 20 sessions now, and I can’t even guess at the answer! And I really, really feel that the very competent characters would know that

 

[Staffan]

An issue with skills in many RPGs is that they rely too much on random chance. Linear systems feel particularly prone to this, but other types of systems can also have that problem.

Basically, in real life most people have a number of things they can do reliably. There are also a number of things I know I can't do, and some things I might be able to do. Someone who's more experienced in a particular field has more things they can do reliably, and they can probably do the basics faster and better than a newbie. But there are very few tasks where one person would have, say, a 50% chance of success and another more skilled person would have a 70% chance.

Ironically, class/level systems often deal better with this, at least for those things that are part of the class/level part of the system (which tend to be combat-oriented). For example, a skill-based system might have a Berserking (or Rage, or Frenzy, or whatever) skill you can roll against to enter a berserker rage with certain bonuses and consequences. So whether or not you manage to rage is a matter of chance: sometimes you do, sometimes you don't. On the other hand, a class/level system would likely have Rage as a class ability that you can activate when you feel it's appropriate. Improving your ability to rage would either give you bigger bonuses, or be able to do it more often, or something like that, but it wouldn't be a stochastic improvement the way a skill-based system would do it. And of course, there are also systems that combine skills with some kind of special abilities.

 

[mamba]

Eclipse Phase can. Because of the way its special and critical results are calculated, you can very much end up with an average roll that produces a non-average output. Crits in particular are scattered across the whole of the die roll range (essentially in its percentile resolution, its when you roll doubles

ah yes, doubles as crits, if you go by the total it still is all over the place, did not consider that case / their total for this.

The point remains that the way I understood the post, the complaint about the bell curve was that it makes ‘average’ results more likely compared to the linear chance of a single die.

This does assume that the special things happen on min and max values, and depending on your other rules you can compensate for the uneven chance of specific totals (crit fail on 2 and 3, crit success on 11 and 12, or crits as doubles…) so yes, ultimately it is about the rules in the aggregate, not just about whether you roll on die vs 2 or 3

 

[billd91]

An issue with skills in many RPGs is that they rely too much on random chance. Linear systems feel particularly prone to this, but other types of systems can also have that problem.

I don't think linear systems are any more prone to over-relying on random chance than non-linear ones - if you're using a random generator, you're using a random generator. Whether they produce a lot of failed results or not (or crits, fumbles, or successes) depends on the target numbers being set, not the fact that a random generator is involved.

The stochastic element provided by the random generator (whether dice, cards, counting birds flying by the window in a 2 minute period, etc) is there to bring in the element of unknown and possibly complex factors that we can't model that make the difference between succeeding and failing at some task. And maybe a lot of games DO rely on such rolls too much and should back off and only go to them when the situation actually needs them to resolve uncertainty when there are notable consequences. 5e is written to promote that sort of GMing and I think it's a good suggestion. Over-reliance on rolling for everything does tend to frustrate by dragging out the game and pacing or saddling PCs with failures when skills are mid and target results set too difficult.

Basically, in real life most people have a number of things they can do reliably. There are also a number of things I know I can't do, and some things I might be able to do. Someone who's more experienced in a particular field has more things they can do reliably, and they can probably do the basics faster and better than a newbie. But there are very few tasks where one person would have, say, a 50% chance of success and another more skilled person would have a 70% chance.

Part of the problem here may be one of context. Yes, a more skilled person may be a lot faster at reliably completing a task than someone with less skill. So what does failure mean? Perhaps it just means that the lower skilled PC didn't have enough time to accomplish the task given the time constraint and the failure result (or increased chance of failure) means a higher chance of a result that isn't helpful or is of unacceptable quality to accomplish what needed to be accomplished. For something like picking a lock, 70% vs 50% means the first PC is likely to succeed at it in fewer attempts (or minutes of picking at the lock mechanism) than the latter. For an artisan, a 70% vs 50% to forge some daggers means they are more likely to have the order ready on time than the latter smith.
But in both cases, if we're limited to situations where success is uncertain because there's a particular time constraint, the setting is distracting (like combat is going on in the same room), or there's a consequence (like being flogged by the master artisan for bungling the order, you slovenly apprentice, you), that character with the 70% chance is likely to get by more often than the one with 50% and avoid negative consequences because they're better, faster, or more efficient.

 

[Bill Zebub]

Yeah, but I bet after a few rounds you'd have some sense of it, and if not conveyed with numbers in a game, then what?

I'd argue that to the extent that is true, it's not at the resolution of the delta between linear and normal.

 

[Bill Zebub]

Not disagreeing with you — that’s why I like linear systems — there is no math to understand. Your skill is 80; you are 80% likely to succeed. No math (at least until you have to decide on extreme results!).

Skill 80 = 80% doesn't take into account the difficulty (or armor class) of the challenge, nor situational modifiers. And personally I don't like RPGs that ignore those things.

I’m going to ignore your “unknown” adjective, because by definition, you cannot know your odds against an unknown opposition! Instead I’ll speak to the case where you are opposing against a known opposition.

For a while I was an expert TKD competitor, and I pretty much could tell you the odds of landing a blow against a newbie, an average club member, an average state champion and an Olympic contender. I’d regard myself as expert software developer, and I do regularly estimate the success odds of tasks in terms of completion within X days and generally they are pretty accurate.

Of course, not to the exact percentage, or probably even 10%, but enough to get a general feel. And definitely enough for usual cases. So I do think that a system should be able to do that without, as you say, me needing to do math.

As I said above, I doubt even very educated guesses are at finer resolution than the difference between linear and normal distributions.

 

[Bill Zebub]

Basically, in real life most people have a number of things they can do reliably. There are also a number of things I know I can't do, and some things I might be able to do. Someone who's more experienced in a particular field has more things they can do reliably, and they can probably do the basics faster and better than a newbie. But there are very few tasks where one person would have, say, a 50% chance of success and another more skilled person would have a 70% chance.

This is why I grant auto-success on tasks at which PCs have been trained, unless there are complicating factors.