What are the practical limits of d20+mod vs DC?

[Stalker0]

Mind you, there are other ways to play with d20+mod vs DC instead of simply increasing the modifier and the DC.

What 4E has called "brutal" dice (with respect to rolling for weapon damage), i.e. allowing for a re-roll if the result is below a specified number, is another possible way to tinker with the probabilities. Let's say that skill allows you to re-roll results that are below your "skill level"; if you have a skill level of 5, say, than a roll of 1-4 on a d20 is always disregarded. Effectively, you will always get a result of between 5 and 20, and are rolling 1d16+4.

Assuming no other modifiers (and a DC cap of 20), this means that success is never impossible even for an unskilled character, but highly skilled characters will succeed more often, possibly even automatically.

An unskilled character would succeed on a DC 10 check about 55% of the time. A skill 5 character, would succeed about 75% of the time (12/16) and success would be automatic for a skill 10 character.

EDIT: Past a certain point, it becomes more efficient to simply change the die type instead of re-rolling the d20. At skill 9, you might as well be rolling 1d12+8, for example.

I really like this idea, I think we should discuss it more. But lets do it outside the thread, its a little OT to the original OP

 

[Crazy Jerome]

I really like this idea, I think we should discuss it more. But lets do it outside the thread, its a little OT to the original OP

It could be on topic here, as long as you talked about why you wanted to do things a certain way, and what effect that might have on the "feel" of the system. The center of this topic is rather nebulous, and I don't see how people can talk about it at all without dragging in other things, at least part of the time.

 

[Crazy Jerome]

...I find that fictional positioning of this sort can also be more fun than fiddly modifiers for dealing with things like slippers and pitons. Instead of encouraging scrounging for bonuses, which can quickly become tedious, use the fiction to help shape the unfolding consequences of failure and success. This means that climbing barehanded rather than with rope and line doesn't necessarily become more likely to fail, but has different (but not necessarily worse, in some absolute sense) consequences for failure.

Does any of that make sense?

Makes sense to me. This is one of the things I was wondering about with the effect of the math on the feel. I can know that, pace the earlier posts, that with a d20, I want most of the rolls to be withing the 40% to 80% success range, and most of those close to 65%. I know mathematically that this leads to results that our group likes. But absent scaling, that doesn't even allow the full +1 to +10 modifiers that feels more correct.

And then I also know that a range of +1 to +8 doesn't tend to give us much happiness, at least not with stacking. +1 is too fiddly, +2 almost is, and if you get any bigger than that, you can't have more than 2 mods. It creates the impression in my mind that if I could state all the things that people want at our table, then do the math for each thing--the Venn diagram showing where hte math intersected would be awfully smal--perhaps even non-existantl.

There is a bit of the Legend and Lore stuff about tiered ranks of skills (e.g. "expert") still buzzing around in the back of my mind, though not exactly as they presented it. Though I'm not sure that centers the feel of the system perfectly, either. It is definitely a way to avoid the fiddly bits. You've got the mods that you have. You need major stuff to shift "ranks", and the only way to get it is to engage heavily with the fiction. Asssuming the system was set up that way, and I can't see any reason for having such a system unless it is set up that way.

 

[Kaodi]

The main problem with FireLance's proposal, as I see it, is that with changing die sizes it that it makes it difficult to impossible to have automatic failure rules. With a d20, you have a 5% chance of rolling a 1. With a d4, you have a 25% chance.

 

[Crazy Jerome]

Another funny thing about fiddly bits in game models is that they so seldom work the way they do in real life, when you consider the result of the fiddly mod on the chance of success.

It seems to me a lot of minor edges in skills are binary in effect. If you know the edge, or have developed the knack, or spent a bit of time on the technique--then the rest of your skill in the field can be applied in certain circumstances fully. And if you don't, it can't, or is much hampered. Such things act, in real life, more like "proficiency" than a minor mod. There is a sense in which being an expert or master or guru is having a somewhat complete set of such things.

For example, late in her life, we finally figured out with my grandmother that the way to learn her techniques was not to ask, but go do the thing while she watched, and have her correct our techniques. Because she assumed so many little things that she didn't even think to say. I was rolling out dough for shortcake, and she offhandedly mentioned that I should flip it upside down and rotate it 90 degrees halfway through. This would cause it to roll more evenly and not stick as much. This took her all of 15 seconds to teach. No one else in my family knew it, and everyone in it cooks. There was a lot of little things like that, and precious little that was major all by itself.

Now in a game, that isn't even worth a +1 to "cooking", in any system. Yet, if you know it when it applies, it is a major situational improvement on your success rate. I suppose traditionally in games, this is where a lot of the variance in the d20 roll is coming from--you know some such things and not others. But of course that is the wrong place for it, if you do the same thing over and over.

"Let it Ride" would make the d20 range make more sense here, assuming you don't mind retroactively including such explanations. If I sneak into the goblin camp and roll a 5, I didn't know something critical, even though I'm a decent sneak. If I roll a 15, I did know it. So that is why I can now sneak through the whole camp with that one roll.

However, that also makes me wonder about a really simple, modless mechanic. All skills are handled with a d6. Roll the right number, you succeed. Your needed number is:

1+ auto success, anyone can do it, or your opposition makes it that way
2+ the problem might sneak up on you, but probably not
3+ you've got a definite edge but not definitive
4+ evenly matched
5+ the problem/opposition has a definite edge
6+ you might get lucky

The mods are all major things that can shift the ground. I need a 6+ to hit the halberd master, but make him fight on ice while I have special shoes, and we can move it. Or you might extend the contest to multiple tests, preferably some that aren't so tough.

What makes this mechanic interesting to me from a feel perspective is that the odds of those things are mostly pretty close to how I would expect, given the descriptions. If two opponents are evenly matched, then 50% is about right, not the 2/3 or 65% that feels better. But if one side has a real edge, then the 2/3 success makes sense. It breaks down on the other numbers for single checks, but not for a series.

 

[LurkAway]

However, that also makes me wonder about a really simple, modless mechanic. All skills are handled with a d6. Roll the right number, you succeed. Your needed number is:

1+ auto success, anyone can do it, or your opposition makes it that way
2+ the problem might sneak up on you, but probably not
3+ you've got a definite edge but not definitive
4+ evenly matched
5+ the problem/opposition has a definite edge
6+ you might get lucky

I like the directness and simplicity of a single die roll, but it does require the DM to do all the work: 1) to know the skill of the PC, 2) to know the skill of the NPC or thing, 3) to evaluate who has the edge or not. The players cannot do this, and aren't really involved at all because they don't know the stats of the NPC so they can't contribute to making an evaluation.

However, if you have an opposed roll, then the player can at least feel like he's contributing with a die roll. Also, the DM doesn't need to constantly eyeball the relative difference between PC and opposition with every roll. He just needs to be fair-minded when he initially designed the NPC or thing, or to trust the system to have done so fairly, and then the die rolls will take care of the probabilities.

Instead, I would love to have something like the above as a guideline when determining DCs. With no guideline, it can feel arbitrary when determining a DC for an obstacle.

If that makes sense?

 

[Viktyr Gehrig]

Bell curve systems have an unfortunate side effect of making it really difficult to estimate probabilities in play and making the value of bonuses unpredictable. The only bell curve system I've seen that avoids this issue is Fudge, since the average value of the dice is always zero-- you have a 1-in-2 chance of succeeding (or failing) when your bonus equals the difficulty, with the probability of success (or failure) scaling exponentially per point of difference.

Then, if you're using smaller numbers, you run into far fewer cases where difficulties fall outside of the range of possible results.

 

[LurkAway]

With attack rolls, it's binary success or failure, but the damage rolls and conditional effects then moderate that with how much success or failure. Skill rolls don't have that nuance. Probably they don't need to thanks to roleplaying (whereas roll-playing is required to fairly adjucate for hit and damage).

Can the merits of d20+mod vs DC and swinginess of die types and probabilities be discussed without knowing what it's all trying to model?

For cooking rice, possibilities are:
0 - you're not able to make cooked rice at all (no roll allowed)
1 - you burn the rice (roll again but pay for more rice)
2 - you undercook the rice (but you can roll again)
3 - you make clumpy rice
4 - you make great rice
5 - you make perfect rice
6 - you make godly rice

(I know this is abstracted in the sense that not all steps are equally probable)

Your rank in cooking rice could be:
0 - you don't know how to cook rice at all (but just ask someone for instructions, and you'll get a rank!)
1 - you know about rice and hot water vaguely (but a little bit of training and instructions will get you to the next rank)
2 - you know to boil water, add rice, cover, and simmer on low heat
3 - you know the ratio of rice to water should be 1:2
4 - you are an excellent rice cooker, you know to fluff the rice to separate the grains, that rice pilaf should sauted in oil in advance, etc.
5 - you are a master rice cooker, ie. a famous chef
6 - you are an epic rice cooker, ie. a God of Food

(this is not a linear curve, but a levelling system could make it easy to reach level 1 and hard to reach 5 and hardest to reach 6)

A healthy level 6 rice cookier has less swinginess and should make level 5 rice at the worst case scenario, whereas the level 1 rice cookier should have lots of swinginess and achieve level 1 to 3 rice, with a tiny chance of level 4 rice, and no chance of level 5 or 6 rice.

So with skill rolls, what kind of DCs are we setting (to cook rice or not, or to make cooked rice of a certain quality) and can the mechanic include a fictional fudge factor that helps us to roleplay with the roll-play and does d20+mod help or hinder with that?

 

[Ringlerun]

I simple solution is to make a natural d20 roll a success no matter what the DC is or make the d20 open ended. If you roll a natural 20 then roll again and add so its possible for a 1st level character with some luck to get succeed on a DC of 40.

 

[Kaodi]

In any case, I do think, for the sake of simplicity, that it would be at least superficially cleaner if the higher possible bonus was +20 on a d20 roll.

Maybe that could be something like a combination of +10 skill, +4 ability, +2 circumstance, +2 morale or insight, +2 luck or sacred.