D&D 5th Edition XII: Peasant Militias Can Defeat Smartphones?

[Stubbazubba]

I would argue that in general, skills are best modeled on a bell curve rather than a straight line. In general, people are average, and when they do things, they do them averagely.

But that's a model for a population's attempts. And since the skill system is not "See how far you jump" but "See whether or not you cross the gap," it makes no sense to put it on a bell curve, because you're not modeling someone's objective results of attempts at the same task again and again, or a population's attempt at the same task, you're modeling their subjective ability to overcome this certain obstacle. And that's just a percentage, whether it's 83.8% on 3d6 vs. DC 8, or 80% on d20 vs. DC 5.

You say "Training and practice (or lack thereof) move that curve to the left or right," but that's not actually true. It moves where you fall on that curve to the left or right, but the curve stays the same, because it's a measurement of a population. In order to really capture the bell curve, characters' abilities need to be on a bell curve (like rolling 3d6 for stats!).

Moreover on the trained professionals thing; so? How does that change with either d20 or a bell curve? In either system, trivial tasks fall off the bottom end as skill increases, and those who are experts but not masters still have a small chance to fail, just as you say should be (unless you're playing dicepools, in which case it doesn't, but that's modeling slightly different fluff than D&D does, though even in that case, as skill increases, the odds of failing to get 1 Hit are infinitesimally small).

And after all that, you haven't given me a skill that is so much better represented by a bell curve than a d20 that it makes up for the lack of transparency. Saying "lots of stuff, like bosses being incompetent," doesn't quite count.

-SNIP-

1) Have you played a Cortex game?

2) Sorry, I was looking at Icewraith's proposal, though it appears several other posters were talking about at least rolling 1d20 + Savage Worlds dice. That's an even worse idea, because each die step is only a +1 on average, and only goes up to +6.5 average, which means all of your training is still overshadowed entirely by randomness. Not OK.

3) I actually think ImperiousLeader's system sounds pretty good, if you must use Bounded Accuracy. It's basically just giving you +5/level to those checks but hiding it by subtracting it from the DC instead. So it's similar to what I've proposed, just with the math on the other side of the equation.

[1337 b4k4]

And since the skill system is not "See how far you jump" but "See whether or not you cross the gap," it makes no sense to put it on a bell curve, because you're not modeling someone's objective results of attempts at the same task again and again, or a population's attempt at the same task, you're modeling their subjective ability to overcome this certain obstacle. And that's just a percentage, whether it's 83.8% on 3d6 vs. DC 8, or 80% on d20 vs. DC 5.

And arguably this is why the D&D skill systems have always sucked. Because we're using binary results to determine binary outcomes rather than a continuum of outcomes, hence why I believe different systems should have different mechanics.

Combat uses a flat d20 and gets away with it because the results of combat are essentially a bell curve using Xd20 where X is the number of rounds.

Skills on the other hand in D&D are often single checks, obstensively because:

a) It doesn't make sense to model climbing a tree like a combat
and
b) Most skills aren't group affairs and people generally despise sitting around waiting for one character to do their thing (the Decker problem).

It therefore follows that skills in D&D should use a different mechanic so that we can model skills on a continuum of success and failure and skill checks are less "did you succeed" and more "how much (or little) did you accomplish".

[Raineh Daze]

Hence why I argue a bell curve is a much more reasonable modeling mechanic for "single shot" skills. Because real life skill improvement is all about raising the ceiling, leaving the floor where it is and shifting the curve closer to the ceiling.

Nope, can't say that I have ever seen or learnt anything in real life where improvement includes allowing you to do as poorly as the worst novice (or even one of their worse, rather than average, attempts) in conditions no different from normal.

It would be like my beating an Olympic sprinter in a sprint. That just won't happen, the floor has gone up at some point.

It therefore follows that skills in D&D should use a different mechanic so that we can model skills on a continuum of success and failure and skill checks are less "did you succeed" and more "how much (or little) did you accomplish".

I liked Flickerdart's thing. DC is sort of analogous to AC, with some other thing representing the number of points in excess of that DC you need to accumulate to finish it. Or something like that.

[BayardSPSR]

And arguably this is why the D&D skill systems have always sucked. Because we're using binary results to determine binary outcomes rather than a continuum of outcomes, hence why I believe different systems should have different mechanics.

Combat uses a flat d20 and gets away with it because the results of combat are essentially a bell curve using Xd20 where X is the number of rounds.

Skills on the other hand in D&D are often single checks, obstensively because:

a) It doesn't make sense to model climbing a tree like a combat
and
b) Most skills aren't group affairs and people generally despise sitting around waiting for one character to do their thing (the Decker problem).

It therefore follows that skills in D&D should use a different mechanic so that we can model skills on a continuum of success and failure and skill checks are less "did you succeed" and more "how much (or little) did you accomplish".

This I heartily agree to. A partial-success system also makes helping viable (rather than just hoping the party-member succeeded too), and could even provide an incentive to participate even when unskilled (a little help could add up).

Or it might not; it depends on the exact implementation.

[1337 b4k4]

Nope, can't say that I have ever seen or learnt anything in real life where improvement includes allowing you to do as poorly as the worst novice (or even one of their worse, rather than average, attempts) in conditions no different from normal.

It is always possible to do as bad or worse than an absolute novice. When you train in a skill, the floor never rises, the ceiling just gets higher and the probability of you hitting that floor gets lower, but the floor remains the same. I write software for a living. The senior developer on my team wrote the core software himself, wrote the software to control self driving cars for DARPA and has probably forgotten more things about software than I know. In short, he's an expert and could code circles around myself or anyone else on the team. Yet sometimes, whether it's a bad day or just a brain dead moment, he will make the simplest of mistakes that a CSC 101 student would could catch. Does he make them as often as a CSC 101 student? No of course not, but there is no magical floor in his skill set that means he never makes a rookie error, just that his skills and training make the possibility of that smaller than a rookie.

[Raineh Daze]

It is always possible to do as bad or worse than an absolute novice. When you train in a skill, the floor never rises, the ceiling just gets higher and the probability of you hitting that floor gets lower, but the floor remains the same. I write software for a living. The senior developer on my team wrote the core software himself, wrote the software to control self driving cars for DARPA and has probably forgotten more things about software than I know. In short, he's an expert and could code circles around myself or anyone else on the team. Yet sometimes, whether it's a bad day or just a brain dead moment, he will make the simplest of mistakes that a CSC 101 student would could catch. Does he make them as often as a CSC 101 student? No of course not, but there is no magical floor in his skill set that means he never makes a rookie error, just that his skills and training make the possibility of that smaller than a rookie.

You're thinking of mental skills. I'm thinking of physical ones, where sheer bodily ability and muscle memory are going to make it practically impossible for a master of it in peak condition to equal a novice's most disastrous mistake.

[Knaight]

Automatically succeeding is boring.

That gets back to the issue of the "Cloaking Device" Rogue that Mike Mearls and Rodney Thompson were discussing in one of the Podcasts.

It depends on what is automatically succeeded at. Opening an unlocked door shouldn't need a roll in D&D, but it would be a difficult task in Bunnies and Burrows. Walking across normal terrain really shouldn't take a roll either, up to and including most slopes. There is a class of things that isn't rolled for, and where exactly the line dividing that from what should be rolled for is is up in the air.

"but then you're subject to DM fiat and the game breaks down if you disagree on whether you're great or whatever".
I like the idea, the Internet think tank seems not to care for it.

That's a terrible, terrible argument. There is no DM fiat regarding disagreeing on whether one is great, because great is a game term that is every bit as defined as, say, +3. If that argument is actually standard for the internet think tank it isn't that they don't care for it, it's that they don't understand it at all.

[Grod_The_Giant]

It depends on what is automatically succeeded at. Opening an unlocked door shouldn't need a roll in D&D, but it would be a difficult task in Bunnies and Burrows. Walking across normal terrain really shouldn't take a roll either, up to and including most slopes. There is a class of things that isn't rolled for, and where exactly the line dividing that from what should be rolled for is is up in the air.

Agreed. In a properly done, non-rules-light system, the books will tell you that this is an Easy task and this is a Great task, just like they tell you that this is DC 15 and this is DC 25.



If we're sticking with the narrow bonus spreads of bounded accuracy, I'd support some kind of scale system, where both your skill and DCs are rated on a scale-- say, Apprentice -> Journeyman -> Master -> Legend. You roll for tasks on the same scale, auto-succeed on DCs for a lower scale, and can't attempt tasks on a higher scale. (Circumstantial bonuses/penalties aside).

So for, say, Climb, a DC 15 check might give you:
{table=head]Apprentice|Climb a tree
Journeyman|Climb a sheer stone wall
Master|Climb a glass wall
Legend|Climb a sunbeam[/table]

For an opposed check, you'd compare the scales of the two participants. If one is higher, that person just wins. You only roll if they're evenly matched.

[Raineh Daze]

If we're sticking with the narrow bonus spreads of bounded accuracy, I'd support some kind of scale system, where both your skill and DCs are rated on a scale-- say, Apprentice -> Journeyman -> Master -> Legend. You roll for tasks on the same scale, auto-succeed on DCs for a lower scale, and can't attempt tasks on a higher scale. (Circumstantial bonuses/penalties aside).

Does rather circumvent bounded accuracy. I'm all for that.

[ImperiousLeader]

I cannot untangle this to work out how you arrived at a DC of 10.

There's one level of difference between Bob's skill and the required check, so an Average check of 5, but I added one shift on the ladder. That gives you DC 10, a Good check of DC 10.

Basically, I want the DM to have the option of making you roll your skills against difficulties at your level. So if you need to make a Good DC with a Good skill, it's DC 5.

[Knaight]

Regarding floor rises, randomness, and such. A point that hasn't been made is that performance gets more consistent the better one is - a complete novice demonstrates the ends of their ability far more often than someone who is really good. To simulate that, one could just add multiple skill dice together. Unskilled would be 1d20, then it is just +1d6 per skill point, perhaps to a maximum of +5d6. 1d20+5d6 can theoretically go as low as 6, meaning that all but the very easiest DCs can still be failed, but that this will only rarely happen. At the same time, DC 50 is theoretically possible as well, with this being exceptionally rare. The average is 28, which is pretty solid, but there is still a reasonable chance of failing various tasks. Specifically:
DC 5: 0% Failure Chance
DC 10: 0.3% Failure Chance
DC 15: 2.95% Failure Chance
DC 20: 15.45% Failure Chance
DC 25: 37.64% Failure Chance
DC 30: 62.36% Failure Chance
DC 35: 84.55% Failure Chance
DC 40: 97.05% Failure Chance

This leaves all DCs over 20 beyond the untrained, which seems pretty reasonable. It also leaves obvious room for a skill training feat that doesn't suck: Bumping all of those d6 up to d8 is entirely reasonable. For that matter, if all DCs are kept in the 5-25 range, even the best have a reasonable chance of failing the hardest, without it being guaranteed. This seems pretty reasonable to me. Expanding the math further, this time using a success rate.


That looks pretty reasonable to me, though I'd be tempted to cut off the +5d6 and instead cap things at +4d6, particularly as attribute bonuses do play into this to some extent. Plus, that nicely works out to the lowest level having all but the highest possible and the highest level having all but the lowest possible, with all of them up in the air for everything else. It also has 5 difficulties and 5 skill levels, which is nice.

[1337 b4k4]

You're thinking of mental skills. I'm thinking of physical ones, where sheer bodily ability and muscle memory are going to make it practically impossible for a master of it in peak condition to equal a novice's most disastrous mistake.

Making a free throw, catching fly balls, kicking a soccer ball into an empty net, hit a layup, sinking a short put, these are all actions which professional athletes and novices alike should all be able to accomplish with regularity and are in fact considered basic skills of the profession, and are skills which time and time again the professionals fail at. Failure is normal, people fail. The difference between the expert and the novice is not that the expert never fails, but that the expert can be relied on to fail less often.

Also, it's good to see that we're on the same page of there being at least 2 different types of skills.

[Stubbazubba]

And arguably this is why the D&D skill systems have always sucked. Because we're using binary results to determine binary outcomes rather than a continuum of outcomes, hence why I believe different systems should have different mechanics.

Combat uses a flat d20 and gets away with it because the results of combat are essentially a bell curve using Xd20 where X is the number of rounds turns.

Fixed that for you. And frankly no, I don't think combat can be simplified to that. Because you're ignoring the interpretation of each of those rolls, the mechanics; most of the enemies need to roll higher than average to hit the good guys, and choice of tactics and powers and maneuvers and spells probably makes just as much, if not more difference in how the combat turns out. It also has to do with varying degrees of HP, healing, and other defensive/ablative concerns. So no, the results of combat are not essentially a bell curve, not at all.

Skills on the other hand in D&D are often single checks, ostensibly because:

a) It doesn't make sense to model climbing a tree like a combat
and
b) Most skills aren't group affairs and people generally despise sitting around waiting for one character to do their thing (the Decker problem).

This is true. But let me link you to an article by The Angry DM which makes a
good case for the one roll skill system, namely, that skills should be used to resolve narrative questions, not just accomplish "stuff."

It therefore follows that skills in D&D should use a different mechanic so that we can model skills on a continuum of success and failure and skill checks are less "did you succeed" and more "how much (or little) did you accomplish".

And now you're talking about mechanics. You can do all of that without changing the RNG, just the interpretation of the results. You can have degrees of success, and Skill Challenges and whatever else you want, all with just d20 + modifiers vs. DC. And you still have that blessed transparency and simplicity of calculation that the d20 + mods RNG gives you. I am yet to see an argument against the d20 itself here.

But let's take this head on, too: Why do skills need more than one check to resolve? The vast majority of skill checks could just use one check to get on with the game for everyone. For everything else there's extended skill checks/Skill Challenges, which still need work, but the idea is almost there. Why would modeling it using a bell curve be an improvement?

When you train in a skill, the floor never rises, the ceiling just gets higher and the probability of you hitting that floor gets lower, but the floor remains the same.

This is a specific fluff we're talking about, then, called real life. And yes, in real life, this is true, and I prefer using d6 dicepools to model anything like realistic skill growth (though again, an average software designer has as much a chance of hacking the NSA as he does of making a novice mistake, which is kind of weird).

But we don't really want to see experts make novice mistakes in D&D, it's not really part of the fluff. Nor do we need untrained guys making experts look the fool even 5% of the time. We're honestly not concerned with "how much you accomplish," we want to know "does it work?" Because D&D is about context and lateral thinking, and not just accumulating enough effort until a problem gives in. That's the biggest problem with the HP system.

EDIT: @Knaight: You've pretty much just recreated The One Ring's system, which is very good, though it does take me a few seconds to add up more than 3 d6s and a d12 (they use a single d12 instead of a d20, and have an Adv/Dis mechanic, as well). Also, it's difficult to grok the probabilities without a chart in front of you, but the system works very well, nevertheless.

[Raineh Daze]

Making a free throw, catching fly balls, kicking a soccer ball into an empty net, hit a layup, sinking a short put, these are all actions which professional athletes and novices alike should all be able to accomplish with regularity and are in fact considered basic skills of the profession, and are skills which time and time again the professionals fail at. Failure is normal, people fail. The difference between the expert and the novice is not that the expert never fails, but that the expert can be relied on to fail less often.

Also, it's good to see that we're on the same page of there being at least 2 different types of skills.

The thing is, all those failures are not normal circumstances. If you put the untrained novice in the exact same situations, they'd almost constantly fail. Pressure is a terrible, terrible thing.

[Knaight]

EDIT: @Knaight: You've pretty much just recreated The One Ring's system, which is very good, though it does take me a few seconds to add up more than 3 d6s and a d12 (they use a single d12 instead of a d20, and have an Adv/Dis mechanic, as well). Also, it's difficult to grok the probabilities without a chart in front of you, but the system works very well, nevertheless.

I'm not surprised. After I checked the math and it came together beautifully, I assumed that somebody had already used it, particularly as there are similarities to the d6 system.

[1337 b4k4]

The thing is, all those failures are not normal circumstances. If you put the untrained novice in the exact same situations, they'd almost constantly fail. Pressure is a terrible, terrible thing.

So what do you define as "normal" circumstances for an athlete if not in the middle of a game? If you're talking about complete no pressure circumstances, then we're back to "why the heck are you rolling" and that will complete our devolution that occurs in every skill discussion to a debate over how much the rules should dictate when you roll rather than dictating how to roll when you've decided a roll is necessary. You'll have to forgive me if I'm not interested in heading down that path again.

[Raineh Daze]

So what do you define as "normal" circumstances for an athlete if not in the middle of a game? If you're talking about complete no pressure circumstances, then we're back to "why the heck are you rolling" and that will complete our devolution that occurs in every skill discussion to a debate over how much the rules should dictate when you roll rather than dictating how to roll when you've decided a roll is necessary. You'll have to forgive me if I'm not interested in heading down that path again.

They're normal circumstances for the athlete--that is, it would be a roll that they would be able to pass on a 10. Normal circumstances for most anybody, however... yeah, the athlete shouldn't be failing to kick a ball into an open net with no pressure whatsoever and nothing to make them hurry. :|

[Friv]

So what do you define as "normal" circumstances for an athlete if not in the middle of a game? If you're talking about complete no pressure circumstances, then we're back to "why the heck are you rolling" and that will complete our devolution that occurs in every skill discussion to a debate over how much the rules should dictate when you roll rather than dictating how to roll when you've decided a roll is necessary. You'll have to forgive me if I'm not interested in heading down that path again.

I think the point is, an expert athlete (d20+12) against a barely talented amateur (d20+2) in a game should be a foregone conclusion, right?

But it's not. In fact, the amateur is going to win one game out of five.

And if you can't use the system for a straight competition... you can't really use the system.

[Raineh Daze]

I think the point is, an expert athlete (d20+12) against a barely talented amateur (d20+2) in a game should be a foregone conclusion, right?

Eh, not at all. In this case, it's not a competition (because d20-based opposed rolls suck. Hugely.) but rather a simple task. Kick a ball into a large target, no pressure. Not right up close, but hey.

Now, that would be DC 10, because apparently that's easy. So, slightly more than half the time, an average, unathletic adult with no skill will manage to succeed. Sounds about right.

But adding on dice, and with modifiers capped, you get a scenario where this athlete will fail to do the same simple task more than is reasonably possible... because their floor really hasn't changed at all.

[archaeo]

The thing is, all those failures are not normal circumstances. If you put the untrained novice in the exact same situations, they'd almost constantly fail. Pressure is a terrible, terrible thing.

Indeed, studies have shown that in high-pressure situations, experts tend to perform better than novices. It's a tenant of behavioral psychology; a highly accomplished athlete will thrive off pressure, while a novice will often crumple in the face of it. You've probably noticed this yourself, in whatever sphere of expertise you enjoy in real life.

All that said, I fail to see how D&D fails to model this effectively. If you're the Tiger Woods of killing goblins, you're going to sometimes fail to hit them. If you're the Michael Jordan of sneaking through a crowded city, you will occasionally be seen. Hell, you don't even have the model the failure as an actual "failure," if you don't want; a DM can cheerfully tell you that the goblin just moved quicker, or a townsman happened to look up from his cabbage stand at the wrong moment.

Auto-success hardly seems like something that should be required for a good RPG; it fails the test of being a fun game, since it creates unpleasant DM/player dynamics, while failing the test of being a good simulation thanks to the fact that there's no such thing as a flawless individual who does it right literally every time.

[1337 b4k4]

Ok we are so completely off track from where we started which was my assertion that:

1) Unified Mechanics across a whole system is not automatically a good thing
2) Players are able to understand different actions / activities are governed by different mechanics
3) As long as within clearly defined groups of actions the mechanic is uniform, the use of differing mechanics for different systems / actions is not an undue burden on the players
4) That certain mechanics better lend themselves to modeling certain activities

None of what I have been arguing for the past few pages has said anything at all about whether I believe the current system in next is either good or bad. At best, I've offered my opinion that the current system is bad, in large part because WotC is sticking with a foolish consistency rather than choosing the most elegant mechanics to model the things they want to model.

Auto-success hardly seems like something that should be required for a good RPG; it fails the test of being a fun game, since it creates unpleasant DM/player dynamics, while failing the test of being a good simulation thanks to the fact that there's no such thing as a flawless individual who does it right literally every time.

Agreed mostly. There are plenty of things that it is appropriate to have auto successes for in a game. They're simply not things we should waste pages and rules on modeling.

As a follow on to that, I don't think we should be that worried about having a mechanic that can model differing skill levels between people based on "no pressure" situations. If whatever mechanic we choose models that (a la say Traveller's "time" factor for skills) fine, that's great. But it's no sweat off my nose if it doesn't because frankly, when a dragon is hunting you, the orcs are pissed and the thief is getting away, that's not a no pressure situation.

[Raineh Daze]

Indeed, studies have shown that in high-pressure situations, experts tend to perform better than novices. It's a tenant of behavioral psychology; a highly accomplished athlete will thrive off pressure, while a novice will often crumple in the face of it. You've probably noticed this yourself, in whatever sphere of expertise you enjoy in real life.

Sphere of expertise? I have no sphere of expertise.

As a follow on to that, I don't think we should be that worried about having a mechanic that can model differing skill levels between people based on "no pressure" situations. If whatever mechanic we choose models that (a la say Traveller's "time" factor for skills) fine, that's great. But it's no sweat off my nose if it doesn't because frankly, when a dragon is hunting you, the orcs are pissed and the thief is getting away, that's not a no pressure situation.

This would be the situation where I'd expect the expert to have a 50/50 (or 55/45) chance and the untrained neonate no chance. This... is not really something that easily happens without raising both the ceiling and the floor, or messing around with complicated dicerolls that have no clear probabilities.

[Seerow]

Auto-success hardly seems like something that should be required for a good RPG; it fails the test of being a fun game, since it creates unpleasant DM/player dynamics, while failing the test of being a good simulation thanks to the fact that there's no such thing as a flawless individual who does it right literally every time.

It all really depends on what that failure rate is placed at.

5% is simply far too high for most tasks, especially as the maximum difference between the top and low end. You put me in a race against Husain Bolt, I guarantee you I will lose every time. I'd be willing to put money on him getting a slow start and tripping in the same race, and still being able to get up and beat me.

Similarly, I might be able to nail a free throw occasionally, but put me in a 1 on 1 with Michael Jordan or any pro, and I'm willing to bet if I get a shot off, it's 1 in a thousand, not 1 in 20. And if he misses a shot, it's certainly not 1 in 20. And at the end of the game, the chances of me having come out the winner is effectively nil.

We can keep on going through different examples. But the main point really is that there are situations where comparing an average or below average person vs a pro, that person has no chance of winning. Ever. It simply won't happen. You can say this is due to an average out over a much greater number of rolls; but do you want to turn every skill check ever into effectively combat, where you have opposed rolls every round until some threshold is met? Sounds like it would needlessly drag out things that should be able to be handled very quickly.

Even if you look exclusively at an individual's performance against a set DC, success rates get much better than 95%. Yes you can find videos of pros flubbing in a professional sport, but when a team in football gets the touchdown and goes for the kick, it's more or less a formality. The vast majority of those are going to succeed. When the pro golfer fails the 3ft put, it gets put on Youtube because it is so rare that something like that happens it is shocking to the audience. We're talking about people with 99.9% averages here, that will effectively succeed every time.



And that's just looking at regular humans. In D&D past a certain level, you are beyond human. I know there's a group out there who refuses to accept this, but it is part of what D&D is. High level D&D characters are effectively super human, which makes them far beyond the professionals of real life in their field of interest/skill. Tying them to the shackles of what the average everyman can do is so counter productive to the game that I honestly cannot conceive people still arguing that it is a good thing.

[SiuiS]

Hence why I argue a bell curve is a much more reasonable modeling mechanic for "single shot" skills. Because real life skill improvement is all about raising the ceiling, leaving the floor where it is and shifting the curve closer to the ceiling.

I think you need to draw diagrams. "Shifting the curve" seems to be completely lost on some people.

But that's a model for a population's attempts. And since the skill system is not "See how far you jump" but "See whether or not you cross the gap," it makes no sense to put it on a bell curve, because you're not modeling someone's objective results of attempts at the same task again and again, or a population's attempt at the same task, you're modeling their subjective ability to overcome this certain obstacle. And that's just a percentage, whether it's 83.8% on 3d6 vs. DC 8, or 80% on d20 vs. DC 5.

What? No. It's a curve because the curve measures THAT PERSON'S odds. If the novice, the regular and the expert all have a bell curve for success, the novice's curve will peak at a failure number and Peter off before the middle transition point. A regular will have a centrally peaked curve, and a master will have a curve weighted after the success point; with about the same numbers and the same floor (but higher ceiling) the three have the same numbers but different success rates. Assuming success at the mid point of the scale, the amateur is failing 80% of the time, the regular is evenly split with half of his curves on either side of the benchmark and the master is only failing 18% of the time (as opposed to 20% to account for more room due to higher ceiling).

1) Have you played a Cortex game?

No, but I have played D&D, and an assortment of other games, which are much closer to the actual implementation we are talking about. There are games that don't work this way, but D&D does work like D&D and needs to be judged in that light.

3) I actually think ImperiousLeader's system sounds pretty good, if you must use Bounded Accuracy. It's basically just giving you +5/level to those checks but hiding it by subtracting it from the DC instead. So it's similar to what I've proposed, just with the math on the other side of the equation.

Yeah, that comes up but folks don't like internal mechanics, and they don't like non-objective, fixed DCs.

It is always possible to do as bad or worse than an absolute novice. When you train in a skill, the floor never rises, the ceiling just gets higher and the probability of you hitting that floor gets lower, but the floor remains the same. I write software for a living. The senior developer on my team wrote the core software himself, wrote the software to control self driving cars for DARPA and has probably forgotten more things about software than I know. In short, he's an expert and could code circles around myself or anyone else on the team. Yet sometimes, whether it's a bad day or just a brain dead moment, he will make the simplest of mistakes that a CSC 101 student would could catch. Does he make them as often as a CSC 101 student? No of course not, but there is no magical floor in his skill set that means he never makes a rookie error, just that his skills and training make the possibility of that smaller than a rookie.

Yeah. Part of the trouble is people are making base mistakes about what counts as a challenge. Kicking a ball in a net isn't a skill check. Getting the ball into the net past the goalie is an Athletics skill roll.

Physical skills also still have the same floor; you fail the jump. You fail to stick the landing on your tumble. You fail to escape the undertow. You get winded and don't outrun the horse. The worst possible result is "does not succeed" and that does happen to professionals.

That's a terrible, terrible argument. There is no DM fiat regarding disagreeing on whether one is great, because great is a game term that is every bit as defined as, say, +3. If that argument is actually standard for the internet think tank it isn't that they don't care for it, it's that they don't understand it at all.

That's actually a mistake on my part, because I misread the system proposed as basically one which had been previously dismissed. I caught the difference in a later quote.

If we're sticking with the narrow bonus spreads of bounded accuracy, I'd support some kind of scale system, where both your skill and DCs are rated on a scale-- say, Apprentice -> Journeyman -> Master -> Legend. You roll for tasks on the same scale, auto-succeed on DCs for a lower scale, and can't attempt tasks on a higher scale. (Circumstantial bonuses/penalties aside).

So for, say, Climb, a DC 15 check might give you:
{table=head]Apprentice|Climb a tree
Journeyman|Climb a sheer stone wall
Master|Climb a glass wall
Legend|Climb a sunbeam[/table]

For an opposed check, you'd compare the scales of the two participants. If one is higher, that person just wins. You only roll if they're evenly matched.

That's how ACKS does it. You not only get better at current tasks, you also gain new abilities. The heal skill for example, lets you stabilize and identify wounds, apply healing herbs and (I think) attempt to cure paralysis. At second rank, you get a lower DC on those abilities, get a 1:10 shot of a nonmagic cure spell and can treat more people. At third rank you improve basic functions down to 1:2 chance and improve healing to 3:10, and gain an additional possible use.

I don't remember the other skills off hand, but there are a slew of them which do this.

Regarding floor rises, randomness, and such. A point that hasn't been made is that performance gets more consistent the better one is - a complete novice demonstrates the ends of their ability far more often than someone who is really good.

No, 133t and I both address that. The skewed curve handles it well as you've noticed.

[QUOTE=Stubbazubba;15744421]Fixed that for you. And frankly no, I don't think combat can be simplified to that. Because you're ignoring the interpretation of each of those rolls, the mechanics; most of the enemies need to roll higher than average to hit the good guys, and choice of tactics and powers and maneuvers and spells probably makes just as much, if not more difference in how the combat turns out. It also has to do with varying degrees of HP, healing, and other defensive/ablative concerns. So no, the results of combat are not essentially a bell curve, not at all.[/wuote]

Each round of combat also involves actions specifically designed to alter the curve, so it still makes sense.

This is true. But let me link you to an article by The Angry DM which makes a
good case for the one roll skill system, namely, that skills should be used to resolve narrative questions, not just accomplish "stuff."

We've brought this up. D&D is a rules-heavy system and so people don't feel that narrative is sufficient to make decisions on rolling (see the "barbarians must always roll strength versus commoners and dragons because REALISM" argument) because it basically breaks down to DM fiat on whether its narratively appropriate.

The thing is, all those failures are not normal circumstances. If you put the untrained novice in the exact same situations, they'd almost constantly fail. Pressure is a terrible, terrible thing.

Which is covered by a skewed bell curve.

I think the point is, an expert athlete (d20+12) against a barely talented amateur (d20+2) in a game should be a foregone conclusion, right?

But it's not. In fact, the amateur is going to win one game out of five.

And if you can't use the system for a straight competition... you can't really use the system.

Not in next he's not, because the amateur gets d20+2 and the professional gets (best of d20 or 10)+12, meaning that the master will occasionally flub and roll a 22, but that's a tie.

[Kurald Galain]

Rolling engages the players and lends an air of suspense to the game because success is not certain. Even a master thief gets sloppy sometimes.

I completely disagree. You do not challenge players by having them randomly fail at times, and the whole point of being a master thief is that you don't fail at routine tasks.

Perhaps there needs to be a distinction that for certain tasks, a 1 is still an auto-fail, and for others it's not, just for playability reasons. I'm not convinced this is necessary though; there's not necessarily the need for Darkstalker to exist.

[Raineh Daze]

Which is covered by a skewed bell curve.

I'm not a fan of a bell curve just for skills. The normal distribution is astonishingly rare, though it's been months since I could remember what's actually more common (I know it has a tail). And 'roll Xd6 + Ability Modifier + Extraneous Dice' or 'roll Xd6 + ability dice + extraneous dice' seems... excessive. I cannot see a reason to go for something that takes several more rolls but produces the same result as static modifiers, only with failure as a one in a million chance.

I like the 3.5 bell curve variant, but that changes all the rolls.

[Moreb Benhk]

Rolling a 1 in combat vs. a Peasant is even more unempowering, doubly so for the skilled Fighter or deadly assassin. Some times you just mess up. It happens. The fun part of the game is justifying why it happened and dealing with the consequences. If you are always succeeding in stealth then there is no drama, no excitement, nothing. I activate my cloaking device and wander around town stealing everyone's coin purses.

Automatic successes are just as bad as "I win button" wizarding or taking 10/20.

Once again you've dichotomised the issue. It's not about never/always autosucceeding. It's about sometimes autosucceeding. At easy things that you used to have to roll on. Similar to taking 10/20 - its useful if implemented right - can do in certain situations against easier things with no pressure, cant be used in harder situations.

The fighting example isn't so bad, as others have mentioned, because you are rolling many more times (sometimes multiple times per round) in a combat. So a single botch isn't so huge. However, that said, I'd be perfectly happy to ditch the 1-auto-misses rule as I don't think it adds much.

Or you could just eliminate all the crazy extra modifiers and let a well tested system work as intended.

You mean use a variant on 3.5's system minus some of the craziness?

If Autowin is already baked into the system, why does everything else need to be converted to an autowin mechanic?

I don't know what you mean by that

[Moreb Benhk]

I'm not a fan of a bell curve just for skills. The normal distribution is astonishingly rare, though it's been months since I could remember what's actually more common (I know it has a tail). And 'roll Xd6 + Ability Modifier + Extraneous Dice' or 'roll Xd6 + ability dice + extraneous dice' seems... excessive. I cannot see a reason to go for something that takes several more rolls but produces the same result as static modifiers, only with failure as a one in a million chance.

I like the 3.5 bell curve variant, but that changes all the rolls.

Fireman's hat is pretty common. Although that's just a skewed normal curve which is what I think people are saying.

[Ashdate]

New playtest document is in! And skills... are out?

Good luck making those easy DC 10 checks now!

[Felhammer]

Once again you've dichotomised the issue. It's not about never/always autosucceeding. It's about sometimes autosucceeding. At easy things that you used to have to roll on. Similar to taking 10/20 - its useful if implemented right - can do in certain situations against easier things with no pressure, cant be used in harder situations.

I think there is a clear difference between walking up a step and running around town with a cloaking device.

Part of my issue is that Autosuccess makes no sense for any skill that involves opposed rolls (Stealth, Sleight of Hand, Perception, Sense Motive, bluff, intimidate, diplomacy, etc.). I can see it being used for tasks whose DCs are more static (climbing up a wall, balancing on a rope, swimming, etc.). However, that clearly divides skills in half between those with and those without auto-success buttons. It is simply easier to have everything run off the same skill mechanic and allow the DM to trivialize/gloss-over skill checks via auto-success when they make sense for the plot.

Which is the exact system Next already uses.