Does a GM need more dice than a d2?

[pemerton]

Until we know what the GM's job is, and what the purpose is of the GM rolling dice, how can we know what the probabilities should be?

 

[GMMichael]

Slightly more than one in seven rolls will have a different result with a 50% vs. 65% chance. . .

Not what I was getting at. Roll against 65% eight times. P,P,P,F,P,F,P,P. Hmm, it's a good night; I just got six Passes. Was that with a d2 or a d20? One can't say with statistical certainty until one makes many more attempts. I'd expect the distinction to become much easier when judging 50% versus 75% or higher.

At the point youre using d2 you may as well just wing it - part of the game element of RPGs is uncertainty, more randomness is more fun

For some reason, maybe it was the 3e game design, I thought that WotC aimed to maintain a certain Pass percentage as characters leveled up. Say 50% - more likely 65%. It was pretty evident in the standard bonuses to spellcasting DCs versus saving spell bonuses (or "modifiers").

Long story short: D&D, at some points, strives for odds close to 50/50 (or a d2). I wouldn't, however, suggest that all D&D should be "just wing it."

I don't personally think assuming something with a 25% change of failure as always treating as a success or the inverse as a failure is a good way to manage things, so I guess I'm disagreeing with the premise.

What if your answer gives us an indirect way to evaluate your GM style/preference? I'm hearing that you don't think a GM A) should be allowed to bend such rules and/or B) isn't capable of or willing to inject a decent reflection of 25% odds into GM fiat. I could be wrong, of course

Until we know what the GM's job is, and what the purpose is of the GM rolling dice, how can we know what the probabilities should be?

See original post.

 

[Blue]

Not what I was getting at. Roll against 65% eight times. P,P,P,F,P,F,P,P. Hmm, it's a good night; I just got six Passes. Was that with a d2 or a d20? One can't say with statistical certainty until one makes many more attempts. I'd expect the distinction to become much easier when judging 50% versus 75% or higher.

It may not be what you were getting at, but it was exactly what I was getting at and I wish you'd address it.

Multiple meaningful results per character per session would be different if a system that could handle a 65% (as a sample) chance of success vs. one where it's collapsed into a 50% only granularity.

Across all of the characters, that's a huge impact on each and every session if we were to use a d2 vs. a d3 (66.67%). There is a diminishing return of how often per character per session it matters when you are looking at large dice sizes/small granularity. d20 vs. d% might not even average one per session across all characters. But it is abundantly clear mathematically that not being granular enough, expecially with the extreme end of a d2, will have a heavy affect on the results of every session of play.

 

[pemerton]

See original post.

The OP says this:

In the majority of RPGs that I've seen, the Game Master narrates what happens in the world, and what happens when PCs take actions. Sometimes, the GM doesn't know what happens, so the GM asks for a roll, or makes his own roll

<snip>

When it comes to a simple does-this-happen-or-not, the GM either knows or isn't sure. If the GM isn't sure, it's probably because the odds of one result or the other are very close to 50%. Why not just flip a coin to resolve it? Is more precision really necessary?

So you're really only discussing GM-as-storyteller RPGing?

 

[Thomas Shey]

What if your answer gives us an indirect way to evaluate your GM style/preference? I'm hearing that you don't think a GM A) should be allowed to bend such rules and/or B) isn't capable of or willing to inject a decent reflection of 25% odds into GM fiat. I could be wrong, of course

I think the word "allowed" is a bit heavy in your first clause, but I think its a bad idea generally. As to your second, I think questioning whether that will be done reliably across GMs is perfectly legitimate.

 

[clearstream]

Not what I was getting at. Roll against 65% eight times. P,P,P,F,P,F,P,P. Hmm, it's a good night; I just got six Passes. Was that with a d2 or a d20? One can't say with statistical certainty until one makes many more attempts. I'd expect the distinction to become much easier when judging 50% versus 75% or higher.


For some reason, maybe it was the 3e game design, I thought that WotC aimed to maintain a certain Pass percentage as characters leveled up. Say 50% - more likely 65%. It was pretty evident in the standard bonuses to spellcasting DCs versus saving spell bonuses (or "modifiers").

Long story short: D&D, at some points, strives for odds close to 50/50 (or a d2). I wouldn't, however, suggest that all D&D should be "just wing it."


What if your answer gives us an indirect way to evaluate your GM style/preference? I'm hearing that you don't think a GM A) should be allowed to bend such rules and/or B) isn't capable of or willing to inject a decent reflection of 25% odds into GM fiat. I could be wrong, of course


See original post.

Taking into account that principles in play will lead us to different answers, I think we can look at methodologies for selecting among outcomes as tools and discuss their merits on that basis. That isn't to say one is better than the other, but we can say it is featured in certain ways.

My impression of 5e is that it aims for odds of 65/35 for anything a character has proficiency with (which often covers much of what they will choose to do.) Some of the features of the d20 method are linearity and moderate granularity. The impact of modifiers can be more than expected. Consider

• If I gain +4 to a 50/50 (say through Bardic Inspiration) then where I had 10 ways out of 20 of failing I now have 6.
• If I gain +4 to a 65/35, where I had 7 ways of failing I now have 3. This is a more palpable delta than the 50/50 case.

As you say, these impacts will be masked by randomness, but better than halving times one fails is likely noticeable over not-quite halving. Additionally, consider

• With a d2 I can index any two outcomes (typically, but not necessarily, succeed or fail.) Many contemporary game designers feel there is additional and important expressiveness in at least three (e.g. succeed, succeed with complication, fail) or more.
• Die pools have the additional feature of mapping quite well elements in fiction with elements in system, e.g. add a die for taking stress or accepting an increased cost of failure.

One can then consider progression in character leverage over the game world, which is commonly desired in RPG. It would certainly be possible to construct a d2 system that say slid the index along a progression curve so that one character's succeed/fail was another's succeed/succeed with complication. More design space is probably opened up by using a more nuanced method however: whether that be a die pool or some other approach. The dice methods in Blades in the Dark, Legend of Five Rings, Burning Wheel, Runequest, and 5e all offer features that are expressive beyond what I think would be very easily accomplished with a d2.

 

[aramis erak]

Long story short: D&D, at some points, strives for odds close to 50/50 (or a d2). I wouldn't, however, suggest that all D&D should be "just wing it."

It actually seems to aim for less than 50% in older editions. Standard footman in armor is Dex 10 (no AC mod), chain (AC5), at level 1 being THAC-0 19... so hitting on 14+, not 11+, so 35%. Most level appropriate foes keep the same 30-40% range. Likewise, dwarven detects (1-2 on d6), and a number of other such things.

And 3E onwards, has a 3 state outcome range, not 2. CS, S, F, with CS being 5%; the nat 1 is a mandatory failure, not a fumble, and so not a separate failure state.

Now, in 3E, the default for a fighter is likely St 15, for a +2, and +1 from BAB, for +3, shooting for 16+, thus needing a 13+ to hit.

Even in 5E, where the 15 St (+2) PB 2 combine to +4, and the average warrior is dex 12, for a net AC17 in chain, the standard war armor, an still a 13+.

 

[Gorgon Zee]

When it comes to a simple does-this-happen-or-not, the GM either knows or isn't sure. If the GM isn't sure, it's probably because the odds of one result or the other are very close to 50%. Why not just flip a coin to resolve it? Is more precision really necessary?

If one result has 75% odds of happening, I think most GMs can say, "this is easily more likely than not, so yes, it happens." Suppose those odds drop to 70%. Does the GM really need to roll a d20, or can she say that it's close enough to 50/50, and just flip a coin?

Just because about half the posts assume you are asking about a system where everyone flips coins, I want to be clear that I’m answering this question: does the GM need more than a coin flip?

Being more specific, if I was to modify the rules of a system I run so that the GM resolves everything by flipping a coin, would the players notice a strong difference? I think the system matters — but perhaps not in the way you expect.

To a large extent, the crunchier the system, the less noticeable will be the effect. In D&D4E I am confident that if I ran pretty much any adventure, and instead of rolling a d20, I flipped a coin for a 5 or 15 result, no-one would notice much. There is so much crunch on the player side, that it generates enough randomness and differentiation, that i honestly think the only remark I might get is a comment on my guys never rolling crits.

13th Age, however, has much of its monster design work off dice roll results — a power triggers special effects only on a natural even, or so on. That would make it much harder to run. I suspect I could just flip a coin and pick a power if the result was heads, but it would be much less fun for me, honestly — and I suspect harder to guard against biases.

It would be hardest in very narrative games. In many of those games, there are few dice rolls, and they are very important. AGON, for example, would be hard to make work as there are typically only a few contest rolls in an evening, and the GM dice roll sets the difficulty for them. Setting it by fiat, or choosing one of two options — both of them would seriously detract from the game.

So, for me, I could likely pull it off for simulationist high-crunch games, but not so much for most narrative games.

 

[clearstream]

So, for me, I could likely pull it off for simulationist high-crunch games, but not so much for most narrative games.

Two games that were cited often in a thread on simulationist games were Runequest and Rolemaster. I don't believe one could pull it off for those simulationist games. As an aside, I interpret 4e less as simulationist and more as gamist.

 

[Celebrim]

Yes.

I sometimes make reference to "Celebrim's World Simplest RPG" which has an apparently complete rule set that involves flipping a coin for everything. The mental toy is intended to explain why RPGs have rules in the first place, and why RPG rules tend to become heavy over time.

The problem with flipping a coin for everything is that there is no verisimilitude. Jumping over a puddle and jumping over the Atlantic Ocean are equally difficult and equally easy. So you immediately find that you have the need for more granular rules if you don't want your game to be absurd and potentially meaningless.

You can attempt to achieve granularity by flipping multiple coins, but this turns out not to work very well because of the limited sorts of probabilities you can generate. Yes, you can generate easily the 3 most essential probabilities in most RPG systems - 50/50, 75/25, and 25/75. And it turns out that having things like 95/5 and 5/95, which you can simulate with 5 coins, needing at least 1 or else all 5 successes, are really nice as well. But the problem gets to be with coin-based systems the shifts between difficulties when you modify chances based on character skill or circumstance. Not only are they not intuitive, but small shifts can great large impacts on the actual chances of success or failure that aren't intended. A good example of this in an actual published system were the actual odds of success on the original Skill Challenge framework published for 4e that was hidden beneath the intuitive layers of coin tosses that hid unintuitive math.

So no, you need more than 1 d2, and if you try to write a system that depends on nothing but d2's it turns out to be harder than it looks.