A search for an optimal resolution mechanic

[PoliteNewb]

These are NOT TRIVIAL OPERATIONS.

Yes, they fucking are. If you are not a chimpanzee.

You know what, fuck it, this is a waste of time and brain cells. PL is right, SINGLE DICE MECHANICS ARE THE ONE TRUE WAY.

[Username17]

That xd6+y mechanic definitely sucks. I can already tell that it is incredibly easy to have tasks that a party member can not succeed at, namely anything with a TN of 7+. Heck, if you have anyone with, say, 4d6+4 at the table he is very unlikely to ever roll below 15 or so. At the same time someone with 3d6 is unlikely to ever roll that high. I can't imagine this being intended.

Bonus suckage points for having a RNG that curves more and more the more skilled you get. This may actually be intended, but also makes it very hard to figure out what a +1 does to your chance of success.

It's the highest off your d6s plus your bonus plus 1 for every literal 6 you roll past the first one. So yeah, your chances are fucking bonkers. Having higher skills makes your theoretical range larger (1 die goes from 1-6, 3 dice goes from 1 to 8), but biases the rolls more and more towards one number (1 die has a 17% chance of rolling a 6, 3 dice has 35% chance of rolling a 6). That's just mental.

-Username17

[PhoneLobster]

Yes, they fucking are.

Then WHY do they MULTIPLE the TIME COST of your basic mechanic which WILL become wasted minutes and HOURS rapidly as it accumulates around the table ON EVERY ROLL EVER.

Here is a trick. TIME a guy executing single dice roll base mechanic.

Time a guy doing the Silhouette base mechanic.

Now time them each time they do that 40 or 60 times (edit: or after some quick estimates of rolls in my last gaming night... well over 150+ times, well over) during an RPG session for some accounting for gaming fatigue, which WILL hit Silhouette A LOT harder because of its ADDITIONAL COMPUTATIONAL COST.

You are not a chimpanzee? I don't give a shit, and in your case with your bald faced denial of mathematics in preference of a tribal attack vendetta it's kinda hard to tell little hairy buddy. This is measurable, you claim I'm wrong, GO MEASURE IT. Then come back with your chimp tail between your legs, or come back and lie some more, whatever.

Because I am objectively correct on this. You are defending a massively more complex mechanic. You are defending it without even knowing WHY it is complex and WHAT that complexity is even supposed to bring to the table, which is amazing enough, but this "THERE IS NO SUCH THING AS MULTIPLICATIVE COMPLEXITY!!!!! (or Multiplicative Time Costs)" is fucking bat shit innumerate non-reality based community bullshit.

[Username17]

Silhouette is not a time consuming mechanic. You roll a couple dice, you take the largest one, you add a small number to it, you check for leftover sixes and add one for each you find (if any). Then you compare it to a single digit DC. This is less time than it takes to generate a result on percentile dice, and depending on how you do with carrying to the tens place, may actually be faster than d20.

The issue is not speed. the issue is that if you have a 3 die skill you have a 35% chance of getting one literal number, which means that in a very real way the game is even less tolerant of modifiers and difficulty changes than a hypothetical D4 System. While the game is capable of putting out nominal values between about zero and 10, the numbers are so weighted that the game is less tolerant of a +1 modifier than Star Wars D6.

It's a bad system. But not because it takes a really long time to output numbers, because it doesn't.

-Username17

[Kaelik]

So Frank, I'm confused. What happens if you roll 3 dice and get 2/4/5? Does that mean you get a zero? Or does that mean you do all that shit starting from 5, but you were just saying six because as you increase your dice pool six will become the most common highest die?

[Whatever]

So Frank, I'm confused. What happens if you roll 3 dice and get 2/4/5? Does that mean you get a zero? Or does that mean you do all that shit starting from 5, but you were just saying six because as you increase your dice pool six will become the most common highest die?

If you roll no sixes, your dice pool output is just the highest number that you rolled (then you add whatever static modifier you get).

If you roll one six, your dice pool output is six. Again, add static modifier after.

If you roll more than one six, your dice pool output is 5+n, where n is the number of sixes rolled (it may be easier to conceptualize this as 6+y, where y is your extra sixes beyond the first).

As you add more dice, the odds of rolling multiple sixes go up, but the odds of rolling exactly one six go way up.

[Username17]

So Frank, I'm confused. What happens if you roll 3 dice and get 2/4/5? Does that mean you get a zero? Or does that mean you do all that shit starting from 5, but you were just saying six because as you increase your dice pool six will become the most common highest die?

If you roll a 2, a 4, and 5, your result is 5 plus your bonus (which is probably like +1 or +2). If, after taking the highest die out, there are sixes left over, you get +1 to your result for each remaining six. You don't even have to do that step if you don't roll any sixes.

It's not a slow mechanic, so I don't know what PL is on about. It is a terrible mechanic, because it is less tolerant of bonuses and penalties than "roll a d6". Which is impressive and terribad. So I don't know what Souran is on about either.

L5R, where you roll a pile of dice, find a set of the highest ones, add them together, and then add a bonus, is a slow mechanic. Framewerk, where you roll a pile of dice, try to arrange a set of them into short quasi-poker hands, take your biggest available arrangement of dice, add all those dice together, and then add your bonus, is a fucking table paralyzing mechanic. But Silhouette is quite fast. Most of the time you're just rolling a pile of dice and reporting which one is biggest. It is fucking awful, however, because it has such a small range of expected outputs. But that is totally different from being slow.

"Flip a coin" is not a slow resolution mechanic, it's just really bad.

-Username17

[souran]

@ Phonelobster I get that you don't like anything that does not play exactly like d20 because your math skills are poor. However, that really isn't a reason to let people at least discuss other systems.

Having played with silhouette it is not slow. The games resolution mechanics are taken from the mecha wargame that was the first product that they created. Except for warhammer grognards most wargamers won't put up with mechanics that don't work.

Really the system is quick because it can be played by visual inspection. The games range/math is bounded such that 90% of it can be done on your fingers if you need that kind of assistance like phonelobster.


@Frank
The rulebooks tell the players that Flat +X modifiers are crazy powerful in that system Further the game tries to keep the number of such modifiers both constrained and limited. Rerolls are used generously and lots of outlier cases are discussed in the rules along with MULTIPLE ways of treating them.

Seriously, DP9 is the fucking game company we wish Wizards of the Coast or White Wolf or Catalyst was. Want to see where they show that they understand the difficulty curve of their own system (something Wizards/White Wolf/Catalyst clearly don't) here it is

http://dp9.com//download/SilCOREdiceprob.pdf

Read there system FAQ and compare to it to the garbage that mike mearls puts out. These guys understand game design in a way that should make Frank and K proud.

The silhouette damage thresholding mechanic is easily the best alternative to hit points/damage boxes I have ever played. Its fast, its book keeping is really easy (it can be done with glass beads or die cut zelda hearts if you are really crafty), its just very playable. If D&D went to a damage thresholding mechanic like silhouette it would be the best change the game ever made.


Seriously, the Silhouette system is a system that I pull out my rulebooks to read and wonder why other companies can't produce games that are that functional. However, I will admit that I also don't play Silhouette games anymore because while they can produce a robust system they can't produce content that anybody would give 2 shits about.

Heavy Gear A.K.A. "Canadian Mechwarrior" is a good wargame but a terribly boring rpg. Jovian Chronicles is just "canadian mechwarrior in space" and is equally boring. Tribe 8 and Core Command both play like gurps without any subsystems thrown in to make it interesting. So they are boring to.

Anyway, I have hard all of Frank's complaints before and even if the game has a tightly constrained rng isn't that ok as long as you have a way to read your out enough variation in result to make people care about rolling the dice?

[CatharzGodfoot]

I've been trying to figure out one or two resolution mechanics for a fantasy RPG. I was initially going to go with Frank's TNE system:
[*]1d20+attribute Vs opponent TN+attribute followed by
[*]TN+attribute+level - opponent 3d6+attribute+level compared to a graduated scale.

Unfortunately, with only four attributes the attributes, every ability was using half of the attributes, and it often seemed very forced (why does a dragon's earthquake power run off of willpower or dexterity as well as strength?). With five or six attributes it was barely better (why does a dragon's earthquake power run off of charisma or speed as well as strength?).

So I thought, 'hey, I'll add another possible resolution mechanic':
[*]1d20+attribute Vs TN+attribute
A dragon's breath still does 1d20+Perception Vs 7+dexterity followed by 8+Strength - 3d6+Strength compared to burning/wounded/disintegrated. Now, however, the earthquake roar is 1d20+strength Vs 7+dexterity for prone.

But is an extra resolution mechanic really needed? And is there even a reason to add attributes on the attacker's side when you aren't doing SAME or a skill-based system?
The system I'm contemplating right now is
[*]3d6+level+X - (level+attribute) compared to a graduated scale, with the first effect on the scale always being applied.
So a dragon's fire breath is Vs Strength for buffeted (always occurs)/burning/wounded/disintegrated, a dragon's earthquake roar is Vs Dexterity for buffeted/shaken/prone/buried.
Now I run into the problem where a dragon's bite is either Vs Dexterity or Vs Strength, but can't be both.

Anyone got a good solution for this?

[PoliteNewb]

Then come back with your chimp tail between your legs, or come back and lie some more, whatever.

Just by the way, chimpanzees don't have tails. But they do prefer single-dice mechanics.

[DSMatticus]

Wow. PL is a terrible person to talk to about dice. He hates everything but 1dX+Y so hard he forgets how to read.

[PhoneLobster]

Fine tell me it's not slow But I'm telling you all TIME IT and do the math.

You only need to be (less than) 10 seconds longer on average to lose 15 minutes per 100 rolls. You think 10 seconds is too harsh? Too generous? THEN GO TIME THE DAMN THING. But at around that rate, or even less, over a campaign you will lose hours of extra game time to sheer complexity of resolution.

I know none of you want to admit time costs exist and on a basic mechanic they don't need to be a huge difference to add up fast but this IS real and your denial of it is frankly shameful.

[Username17]

Fine tell me it's not slow But I'm telling you all TIME IT and do the math.

You only need to be (less than) 10 seconds longer on average to lose 15 minutes per 100 rolls. You think 10 seconds is too harsh? Too generous? THEN GO TIME THE DAMN THING. But at around that rate, or even less, over a campaign you will lose hours of extra game time to sheer complexity of resolution.

I know none of you want to admit time costs exist and on a basic mechanic they don't need to be a huge difference to add up fast but this IS real and your denial of it is frankly shameful.

Uh... reporting the highest die on one of the d6s is faster than reporting the roll on a d20. D6s are easier to read than d20s are. When we get into counting seconds, anything that uses dice with a tens place on them automatically loses. Hell, anything that doesn't use one or more six sided dice automatically loses. D6s physically spend less actual time rolling before coming to a stop than do d20s or any other die type.

Taking extra time is simply not a problem that Silhouette has. The system is garbage, but resolution time is not one of the reasons.

-Username17

[PhoneLobster]

Frank

I say you are flat out wrong. Go time it.

[Username17]

Frank

I say you are flat out wrong. Go time it.

3 die Silhouette: 3 seconds.
d20: 4 seconds.

That's five attempts each, disregarding the highest and lowest. Which is a damn good thing for the d20, because the highest was actually like 25 seconds because the d20 rolled off the table and I had to roll one of the other d20s.

Now go fuck yourself.

-Username17

[PhoneLobster]

3 die? really This is a three die system?

Also I'm going to heap doubt on your methodology. I mean especially with this off the table bullshit.

[DSMatticus]

Also I'm going to heap doubt on your methodology. I mean especially with this off the table bullshit.

If you were going to doubt any response that didn't give you the answers you wanted (and let's be honest, you were, and there would be justifiable reasons to because it's certainly not a controlled experiment with a large, unbiased sample size, so potential problems you can bitch about are endless), why the fuck did you ask him to time it anyway? It sounds like every kind of pointless.

[PhoneLobster]

why the fuck did you ask him to time it anyway? It sounds like every kind of pointless.

Frankly?

Because I knew he would pull something totally dishonest.

Like using a non-variable die pool at nearly minimum levels.

And not varying the variable multiplier from the extra success for any of the rolls.

And pretending that one dice is harder to physically marshal and keep on the table than a handful of a variable number of dice.

And deciding to compare a flat die roll of significantly greater RNG size purely to exploit the difficulty of higher numbers in his favor when the actual HONEST comparably sized RNG for a flat die roll would be a 1d6 or maybe at a stretch a 1d10. I mean fuck, why didn't he just go and use a 1d100 at that rate?

You know, things like that.

I sure as heck didn't expect him to even TRY to come up with an honest and in depth consideration of almost ANY of the actual factors involved.

Good thing I turned out to be right about his intellectual honesty.

Now if you WANT a methodology I would NOT frown on I can describe one for you. But I'm going to be honest, I don't think Frank or any of the "there is no complexity cost!" crowd would be even WILLING to even accept the most basic of practical experimental methodologies on this.

[echoVanguard]

After a thorough and rigorous experiment, I have determined that you are both idiots.

I performed 10 rolls of a d20 and 10 rolls of 3d6. I started the stopwatch the instant the die or dice was released, and stopped it as soon as I could mentally determine the desired value (numerical value for the d20, highest numerical value for the 3d6). I rolled the 3d6 inside a box to keep the dice from scattering around, since the first few attempts resulted in dice flying all over the place. I did not count any abnormal results, such as someone interrupting my concentration or my cat stealing one of the dice.

Here are my results (all values are in Seconds)
d20: 1.28, 1.43, 1.40, 1.63, 0.96, 1.20, 1.60, 1.47, 1.42, 1.29. Mean 1.364, Median 1.41.
3d6h: 0.98, 1.11, 1.03, 1.11, 0.95, 1.03, 1.30, 1.32, 1.59, 1.81. Mean 1.223, Median 1.11.

Comparing the differences between these datasets, the mean of all differences was 0.141 seconds and the median of these differences was 0.215 seconds.

According to Human Benchmark, the average human reaction time is 215 milliseconds (or 0.215 seconds to keep our units consistent). Since both the mean and the median of the differences between the two methods fall within this threshold, the real outcome is that the difference is so small as to be unnoticeable.

Neither of these two methods are inherently superior to the other, and quite frankly, I'm appalled at the methodology you two have demonstrated in attempting to prove your positions. PL, asserting something and then challenging your opponent to prove differently is intellectually dishonest, especially when you discredit their results afterwards without formally defining your requirements beforehand - the burden of proof is always on the asserter. and Frank, attempting to claim that your failure to experiment rigorously is evidence for or against any position is quite ridiculous, and seriously undermines the credibility of any assertion you make.

echo

[PhoneLobster]

you are both idiots....highest numerical value for the 3d6

Um. Why are you testing only a single step of one of the discussed mechanics and only doing it with 3d6 when until someone tells me otherwise this is supposed to be a variable dice pool and presumably is usually larger than that?

One of the reasons I feel comfortable challenging others to "prove it" when they claim complexity and the related multiplicative time and complexity costs don't exist is because this is basically an actual field of math and the actual way it works is that an inefficient algorithm with multiple variable inputs will have an exponential growth in time and complexity costs as the variables grow in size and range.

As an experiment just try SAYING what you would say to tell someone to make a roll with silhoute or a single dice plus a modifier. One goes...
"Roll with +Z"
The other goes...
"Roll Xd6 with +Y per extra success and +Z modifier"
Note the THREE variables on one and the ONE variable on the other.

Now go back to your experimental methodology.

For a realistic study of the time costs for actual practical usage you actually need to give a RANGE of actual relatively standard inputs for ALL actual input variables the mechanics DO use in practice,. This means you probably want to write a list in advance of +Zs for your flat roll tests, and +X, +Y and +Zs for your Silhoutte test. And you almost certainly need a friend so you can read off the list and mark off times for THEM to resolve the roll (and check their accuracy as a side survey).

You also need to account for basic physical dice management. Your single dice (which probably shouldn't be a d20, unless you are running more than 2 sample mechanics, which isn't a bad idea) may or may not have several of the same dice in front of them. But the Silhoute and any variable dice pool mechanic you test NEEDS to have at least up to the maximum tested variable dice pool so they can draw appropriate VARIABLE dice pools for the tests.

Now there are other things that might be nice like trying to measure the impact of gaming fatigue in long sessions and other bullshit. But really at a BASIC starting point you need to measure all the fundamental things you ACTUALLY in actual practical need to do in order to USE the mechanic in an RPG. This means yes for a VARIABLE dice pool mechanic, collecting the correct amount of dice, clearing the area of possibly confusing extras and then rolling. And sure you can eliminate outliers like cat attacks for convenience but ignoring the whole thing where the dice pool isn't always 3d6 is pretty bad practice.

So sure. If you perform a test where all you do is perform a single read action off a d20 (neglecting your required variable modifier bonus required for practical purposes) and a read action for highest of 3d6 (which will commonly be averaging somewhere in the 1 point something read actions) and neglect well, the entire rest of the complex mechanic, you are going to have difficulty differentiating their value.

Because you DID just measure two trivially different complexities and they WERE both just short of the minimum individual time/complexity cost for a single operation that you are likely to encounter.

If you want a perceptible difference I suggest you try measuring a more realistic representation of the mechanics at the table.

[K]

I find that as the operations that I am expected to do increases, it makes the game less fun.

It doesn't matter how long those operations are or how easy they are to do.

[Lago PARANOIA]

A median time of 1-1.5 seconds?! Bitch, please. That's just how long it takes to reach over the notes and actually pick up the d20.

[Username17]

A median time of 1-1.5 seconds?! Bitch, please. That's just how long it takes to reach over the notes and actually pick up the d20.

Pretty much. My methodology was to put down the watch, reach over and grab the dice, roll them, read them, pick up the watch again. Anyone measuring from the point of already having dice in hand would get smaller numbers, but that is hardly relevant to any actual game.

Honestly, other than Phone Lobster being an insane person who won't admit when he is factually wrong about something, I don't know why we are still having this conversation. It takes less time to roll six sided dice than it does to roll other dice, but not so much time difference that you'd normally notice at an actual table. One person having "dice rolling rituals" will slow the game more than switching everyone from percentiles to d6s will speed it up by a huge margin.

Again and still: Silhouette is not a slow mechanic. It's slightly faster than d20, though not so much faster that you'd notice most of the time. That is not why it's bad. It's bad because the game system is less tolerant of circumstantial modifiers than Star Wars d6 or Feng Shui, and that's fucking insane.

-Username17

[PhoneLobster]

A median time of 1-1.5 seconds?! Bitch, please. That's just how long it takes to reach over the notes and actually pick up the d20.

Aaaand how long does it take you to reach over and pick up a variable number of d6s you've just now determined/been told from a larger pile of d6s and clear away similar d6s for the roll?

Simple thought exercise Lago? Same amount of time or more?

[...You Lost Me]

Right, so when someone says "roll 1d20", I'm picking through my dice pile for a twenty-sided die, which could take a while. When someone says "roll 4d6", I reach over to my pile of dice that are all d6s because no other dice are rolled in the game and pick up four of them.

Yup, d20 still took more time. You really are bad at this.