A search for an optimal resolution mechanic

General questions, debates, and rants about RPGs

Moderator: Moderators

Post Reply
User avatar
Dean
Duke
Posts: 2059
Joined: Mon May 12, 2008 3:14 am

A search for an optimal resolution mechanic

Post by Dean »

I thought it might be interesting for us to all throw our collective conscious into a talk about the various resolution mechanics that are out there. Ideally to, through the sheer power of our mind-meld, come up with the best single resolution mechanic ever. More realistically I would be interested to air out what the various strengths and weaknesses are of the primarily used mechanics in RPG's today.

So lets see:


Static d20: You all know it and sort of love it. The basic d20 roll + modifiers to meet a set DC. It's nice in that it has a wide range and thusly can do cool things like say "You add a lasersight to your pistol, get a +1 to hit" without totally snapping but that same range makes it incredibly swingy. If a 6th grade class had a long jumping contest in the D20 world then 1 out of 20 kids would almost beat world records and others wouldn't make it the full length of a pizza box with a running start. And it's not that some kids would be better than others because it would change every time. D20 also has the plus of having it's resolution mechanic work on a single die so you can resolve multiple effects at once.

Dice Pool: As seen in every WoD game as well as Shadowrun. Some use d6's some use D10's. Dice pool systems have a strength of being very intuitive and simple to learn. It's one of the reason that WoD is largely still the home of girl gamers (no offense intended women. Your great). Die pool systems can do a lot of things d20 can't but it is not very granular. I've never seen a die pool system deal with things like the "lasersight" problem well simply because a die up or down even in specific situations is huge in these systems. Also suffer from a problem of making me pick up too many goddamn dice often (though not in every system). For some reason I have also noticed Die Pool systems have a hard time abjugating damage. Maybe that's just me, but I've noticed a trend.

Ummmm....Feng Shui: As seen in Feng Shui. Seriously does anything else use the Feng Shui positive/negative die mechanic? If it does I've never seen it. Anyway the up down dice thing is intuitive and crazy simple. Its math is easy to calculate and quick. A +1 to a check is still pretty big but not game-breakingly so so the system can still handle things like "A master made sword gives you +1", which makes people happy. I haven't had a ton of experience with Feng Shui so I'll leave the particulars of this for others

3d6: Basically like D20 but with a nicer statistical curve and made of three dice so there's some pluses and minuses there.

Finally: Bullshit like Amber or Quest for the Grail where you just have to have a high enough number and things work. This isn't really a "mechanic" so I'll leave it out of this discussion.


So what can be found here. Whats "The best". What can be done or designed to make a single incredibly solid simple resolution mechanic that I can put in every product I design hereafter. Let us find this thing.

Maybe there's some way to combine some of their strengths or to shore up their weaknesses. I will think on Dice pool systems now...
User avatar
shadzar
Prince
Posts: 4922
Joined: Fri Jun 26, 2009 6:08 pm

Post by shadzar »

Flip a coin. Heads you win, Tails you lose.
Play the game, not the rules.
Swordslinger wrote:Or fuck it... I'm just going to get weapon specialization in my cock and whip people to death with it. Given all the enemies are total pussies, it seems like the appropriate thing to do.
Lewis Black wrote:If the people of New Zealand want to be part of our world, I believe they should hop off their islands, and push 'em closer.
good read (Note to self Maxus sucks a barrel of cocks.)
User avatar
Midnight_v
Knight-Baron
Posts: 629
Joined: Thu May 15, 2008 10:27 pm
Location: Texas

Post by Midnight_v »

shadzar wrote:Flip a coin. Heads you win, Tails you lose.
I was wondering was I being trolled in that other thread. Now I'm sure. very funny shad.

Has anyone ever just used... I don't know a flat percentile mechanic? Or is that what the WoD does basically. It seems like a scale of 1-100 is easy enough to grasp and understand for all. I mean not that its strictly inherently better than D20 but I'd think it be used a bit more.
Don't hate the world you see, create the world you want....
Dear Midnight, you have actually made me sad. I took a day off of posting yesterday because of actual sadness you made me feel in my heart for you.
...If only you'd have stopped forever...
User avatar
shadzar
Prince
Posts: 4922
Joined: Fri Jun 26, 2009 6:08 pm

Post by shadzar »

Which resolution system is used for most major sporting events to decide who goes first?

Depending on the level of complexity you are looking for, some things require the simplest of mechanics.

There is no need to have a unified mechanic within a game that tries to be applied to an do everything.
Play the game, not the rules.
Swordslinger wrote:Or fuck it... I'm just going to get weapon specialization in my cock and whip people to death with it. Given all the enemies are total pussies, it seems like the appropriate thing to do.
Lewis Black wrote:If the people of New Zealand want to be part of our world, I believe they should hop off their islands, and push 'em closer.
good read (Note to self Maxus sucks a barrel of cocks.)
Username17
Serious Badass
Posts: 29894
Joined: Fri Mar 07, 2008 7:54 pm

Post by Username17 »

The positive/negative die is statistically identical to using 2d6 added together and then having a target number that is 7 higher. FATE uses the positive/negative dice too, and again it maps perfectly to simply rolling positive dice and looking for higher target numbers.

Percentile dice systems are used fairly often, and were a lot more common back in the 80s. Palladium, Call of Cthulhu, and so on were all based on percentile dice. The most common mistake is to have them be a "roll under" setup where you roll dice and try to roll less than your skill. That is stupid, because it makes degree of success pointlessly difficult to calculate. What you should do is add your skill to your roll and have the target number always be 100. Either way you suffer from the problems of fixed target numbers - which is that you can't have secret modifiers because the modifier, the target number, and the die roll are all on the players' side of the MC Screen. But if you're cool with that, it's fine.

There are also systems where you search for features in your dice. Cthulhutech's Framewerk system works like this, so does Deadlands. This taks forever and the odds are so hard to calculate that the authors never bother. Do not want.

But basically you have three good choices:
  • Flat Curves. Roll a number between X and Y, where all numbers are equally likely. Adding bonuses increases your chance of success linearly. This has the advantage of being easy to calculate odds for, and the disadvantage that adding multiple modifiers pushes you off the RNG really fast. Having a bigger spread of numbers makes the there be more different chances available but also makes the numbers you have to keep track of bigger. Common choices are d6, d20, and d100, but theoretically you can use any numeric base.

    Bell Curves Roll NdMs. Numbers in the "middle" are more likely, numbers at the "end" are less likely. Bells are loved by armchair statisticians because the results approximate the normal curve, and modifiers therefore act similarly to how things work in statistical sampling. What this means practically is that the modifiers that provide a large enough bonus in the middle of the range are a smaller portion of the total range, which allows you to add more modifiers before leaving the RNG completely. Unfortunately, this comes at the price of requiring combinatorials to calculate the odds of individual die rolls.

    Dice Pools Roll NsMs, counting how many dice rolled X or better. Has the advantage of infinite extensibility in one direction (adding more dice) and incredibly easy calculations of averages (dice/chance pr die = average hits). Unfortunately, system breaks down completely at the low end (as you have no more dice to take away) and the calculation of the chances to get a specific number of hits with a single roll is a ghastly N-polynomial that is not normally doable at the table.
-Username17
User avatar
Shazbot79
Journeyman
Posts: 106
Joined: Thu Jun 17, 2010 11:53 am

Post by Shazbot79 »

Midnight_v wrote: Has anyone ever just used... I don't know a flat percentile mechanic? Or is that what the WoD does basically. It seems like a scale of 1-100 is easy enough to grasp and understand for all. I mean not that its strictly inherently better than D20 but I'd think it be used a bit more.
Warhammer-based roleplaying games use a role under percentage system.

BRP/Runequest uses a role 100% or greater mechanic if I'm not mistaken.

I've been toying with the idea of using a die-step system, similar to Savage Worlds, but without all of the silly exploding d4 nonsense.

The idea is base stats rated from d4 to d12, and a number of skills rated from d4 to d12, with very little in the way of static modifiers. Player rolls stat die + skill die vs. target number. Meet or beat for success. Not entirely sure how this would work out as I haven't sat down to crunch the numbers on it.
PhoneLobster
King
Posts: 6403
Joined: Fri Mar 07, 2008 7:54 pm

Post by PhoneLobster »

FrankTrollman wrote:What this means practically is that the modifiers that provide a large enough bonus in the middle of the range are a smaller portion of the total range, which allows you to add more modifiers before leaving the RNG completely.
... sigh... no, it really doesn't Frank.

It REALLY doesn't.
Phonelobster's Self Proclaimed Greatest Hits Collection : (no really, they are awesome)
Roog
Master
Posts: 204
Joined: Mon Sep 15, 2008 9:26 am
Location: NZ

Post by Roog »

Why not? It's definitely true for a normal distribution.
PhoneLobster
King
Posts: 6403
Joined: Fri Mar 07, 2008 7:54 pm

Post by PhoneLobster »

Roog wrote:Why not? It's definitely true for a normal distribution.
Because Frank's statement isn't about the behaviour of normal distributions, its a reference to an old slieght of hand argument of his where he claims that curved mechanics "go to 11".

Both the basic linear and curved RNGs only go to 100%, despite his pretense to the contrary Frank's curved RNGs don't actually go to 110.

If you are adding bonuses from the mid point to the end on a 1d20 RNG and a 3d6 RNG (which is the kind of comparison Frank makes when he talks about the implications for the far end of the RNG)... you have 10 discrete points of bonus you can add to the 1d20 and 8 on the 3d6. And they average to pretty close to the same value for each point of bonus.

Frank refuses to use an average value for calculating the value of a bonus on a curved RNG since it makes his claims about going to 11 look bad.

But this leads to fail because if you do what he does, and measure your bonus values during rules design from the median only, then your resulting RPG system is going to be broken to all hell expressly BECAUSE of the behavior of normal distributions making all your numbers not work the way Frank pretends they work.
Last edited by PhoneLobster on Sat Jan 22, 2011 11:44 am, edited 1 time in total.
Phonelobster's Self Proclaimed Greatest Hits Collection : (no really, they are awesome)
CCarter
Knight
Posts: 454
Joined: Fri Jun 11, 2010 10:41 pm

Post by CCarter »

Thoughts:
D20 additive (what the OP called "static" above): pretty solid. Great for opposed rolls since you just compare totals. Requires subtraction when calculating level of achievement, unless some other secondary system determines this instead (e.g. a separate damage roll).

d20, roll under: similar to d20 additive but if level of achievement isn't important, this is potentially faster than d20 additive: you can roll 10 dice against the TN and count 'hits' when the PC is shot at very quickly...if you want levels of achievement that are proportionate rather than absolute (and so increase slowly as ratings increase) this is more easily calculated than with d20-additive. For example, roll under Attack to hit, or [Attack/4] for a special success. Occasionally seen with a 'Blackjack' success system where higher rolls are better unless it fails. Like any roll-under system, this may may seem counterintuitive to players as lower rolls are better. Difficulty modifiers can be applied to the d20, but there's also a psychological bias toward unmodified difficulty unless specific rules apply.

3d6 roll under: bell curve distribution; slow in play since dice have to be rolled in sets. Odd shifts in probability curve from bonuses (see PL rant...)

Dice pool: great for determining how well you succeed (# successes) but IMHO lack of transparency means its not as great for use in assigning chances of success - that is, its better for rolls that give an "amount" result (damage) rather than those that determine whether a character will pass/fail. Result level is vaguely proportional to dice pool, so a zero dice pool gives an intuitive zero success level. Opposed rolls may give more cut and dried results than would normally be the case in d20 systems. (In my opinion large pools are a bit unwieldy, but YMMV: people play Exalted, after all...).
Die pool systems often have a 'safety value' mechanic (spending Willpower or Edge or equivalent) to compensate for invariability in opposed rolls. These systems frequently use a soak mechanic, since soaking requires effective result level generation to work.

Dice pool, matches (One Roll Engine system). Fast and gives two separate success outputs on a roll - 'width' and 'height' - if you need that. Not much transparency. In standard ORE, its hard to apply difficulty factors to unless dice pool is modified (chance of success largely determined by character ability rather than any difficulty factor the GM applies).

Dice pool, use highest in pool: rarely used. Gives 'diminishing returns' (each additional die is worth less) so pools usually have to be fairly small.
Probability wise, if pool >1, the highest roll is actually the most common roll.

D100: fairly transparent since people are used to working in percentages. Roll under is similar to d20 roll under, except slower since each d100 is actually a roll of two d10s together, unless you actually have a 100-sided dice (and they take forever to stop rolling, anyway); some some systems do use the 1s place( e.g. Warhammer hit location or criticals based on the ones place, e.g. TSR's godawful Amazing Engine system or HarnMaster)

Positive/Negative die...as seen in Feng Shui..., you might get some mechanical mileage out of having an outcome roll on the same scale as the initial attribute, but the positive/negative die assignment is clumsy. Its roll up/roll down system means that while a +1/-1 bonus or penalty is substantial, characters also explode with huge successes/failures quite often.

Step Die (Earthdawn, Cortex, Savage Worlds): very straightford and probably fast. Likelihood of 1s doesn't go down much as you go up in die sizes, so opposed rolls often go in favoured of the underdog.

A couple of good links for this subject would be:
http://www.darkshire.net/jhkim/rpg/systemdesign/
(Dice Rolling section, parts 1 to 4)

And general article on dice rolling mechanisms here:
http://www.diku.dk/hjemmesider/ansatte/torbenm/Troll/
Last edited by CCarter on Sat Jan 22, 2011 8:42 pm, edited 1 time in total.
Manxome
Knight-Baron
Posts: 977
Joined: Fri Mar 07, 2008 7:54 pm

Re: A search for an optimal resolution mechanic

Post by Manxome »

deanruel87 wrote:Static d20: ... but that same range makes it incredibly swingy. If a 6th grade class had a long jumping contest in the D20 world then 1 out of 20 kids would almost beat world records and others wouldn't make it the full length of a pizza box with a running start.
That's crap. You may have played a game at some point that used d20s and where that happened to be true, but those things are completely unrelated. What you do, if you have any sense at all, is decide how much you want your narrative results to vary, and then scale your numerical range so that it covers that narrative range.

You could have a 1d6 system where rolling a 1 means you jump and inch and rolling a 6 means you leap a mile, or you could have a 1d100 system where rolling a 1 means you jump 3 feet and rolling a 100 means you jump 6 feet. That's totally arbitrary.

The difference is that larger ranges give you more granularity (you can represent smaller bonuses/penalties), but make the math harder because the players are dealing with larger numbers (and probably more of them, if more possible bonuses exist due to that granularity).

PhoneLobster wrote:If you are adding bonuses from the mid point to the end on a 1d20 RNG and a 3d6 RNG (which is the kind of comparison Frank makes when he talks about the implications for the far end of the RNG)... you have 10 discrete points of bonus you can add to the 1d20 and 8 on the 3d6. And they average to pretty close to the same value for each point of bonus.

Frank refuses to use an average value for calculating the value of a bonus on a curved RNG since it makes his claims about going to 11 look bad.
Looking only at averages is no less bullshit. The entire thing about curves is that a +1 has different values at different points in the curve. If you pretend that it has a single, consistent value, then you are effectively pretending that your probability distribution is uniform instead of curved, and so of course it will look like the curve is making no difference.

Because a +1 is worth different amounts at different points on the curve, there are some places where it is worth more than average and other places where it is worth less. For a given threshold of give-a-damn (that is larger than a +1 on the uniform RNG), the curve has smaller bonuses that you still might care about (depending where you are on the curve) than a uniform distribution with the same average. So if you hand out bonuses at the smallest level that the player might care about, each bonus is a smaller fraction of the RNG in a curved system than a uniform one (ignoring rounding), and therefore you can hand out more of them before going completely off the RNG.

For example, you could rationally claim not to care about any bonus smaller than +2 on a d20 system, but to still potentially care about a +1 on a 3d6 system. And then there would be 5 steps you individually cared about before going off the RNG on the d20 system, but still 8 on the 3d6.

You can debate whether handing out bonuses at the smallest level the player situationally cares about is the best design strategy. And of course, if you don't care about anything smaller than +2 on a d20 system, then it would be better to consider d10 as your uniform system, in which case 3d6 is forcing you to deal with bigger numbers in order to have more steps on the RNG. So it's not obvious that's a good choice, but it is an actual choice. That is a mathematical property that curved distributions provably have, whether you think it is worth it or not.
Anguirus
Journeyman
Posts: 168
Joined: Mon Mar 16, 2009 1:16 am
Location: Manhattan

Post by Anguirus »

You also have dicless systems (like Marvel Universe RPG) and card based systems. I don't have anything to contribute other than to point out that these thing also exist.
Sighs and leers and crocodile tears.
quanta
Journeyman
Posts: 134
Joined: Mon Jun 14, 2010 12:17 am

Post by quanta »

I don't think there is any one optimal resolution mechanic, but it would be interesting to determine what's optimal for a particular game's design. For example, although it could be impractical, you could in theory do a bunch of design work specifying what sort of range of successes/failures you want the game to have for various scenarios as well as shit like what range of tactics you want to cover and how players/npcs of various levels should compare to each other and input little more about your dice roll + modifier mechanics other than shit like "chance of success is a monotonic function of resulting number with an approximately negative second derivative (more like second difference, but eh)" and "possible range of modifiers is bounded above and below".
Last edited by quanta on Sat Jan 22, 2011 8:40 pm, edited 1 time in total.
PhoneLobster
King
Posts: 6403
Joined: Fri Mar 07, 2008 7:54 pm

Re: A search for an optimal resolution mechanic

Post by PhoneLobster »

Manxome wrote:Looking only at averages is no less bullshit.
You use averages when looking at bonuses at an unknown point in the curve or over a known range of points in the curve. Because it's an approximate guess that is right in a general sense. It is significantly LESS bullshit than Frank's method of picking the largest extreme of bonus value on the graph which is only ever right at that one point.

But ideally yes you would take into direct account that the bonus changes, that's a great deal HARDER because that means you need to try and account for all potential DCs and Modifier combinations in advance, and that's pretty hard core.

It also means however that this...
not to care about any bonus smaller than +2 on a d20 system, but to still potentially care about a +1 on a 3d6 system. And then there would be 5 steps you individually cared about before going off the RNG on the d20 system, but still 8 on the 3d6.
Isn't true. Because if you only care about bonuses of +2 and larger on a d20 then there are significantly LESS steps of +1 on a 3d6 that you care about before the +1 changes value to something you don't care about.

Which is largely the point of my criticism of Frank's "it goes to 11!" argument. Because he declared that he cared about bonuses worth about +15% then pretended that he could keep adding iterative "bonuses he cared about" to a 3d6 for longer, even when those bonuses stopped being worth 15% from the second step onwards.
Phonelobster's Self Proclaimed Greatest Hits Collection : (no really, they are awesome)
CCarter
Knight
Posts: 454
Joined: Fri Jun 11, 2010 10:41 pm

Post by CCarter »

Shazbot79 wrote:
Midnight_v wrote: Has anyone ever just used... I don't know a flat percentile mechanic? Or is that what the WoD does basically. It seems like a scale of 1-100 is easy enough to grasp and understand for all. I mean not that its strictly inherently better than D20 but I'd think it be used a bit more.
Warhammer-based roleplaying games use a role under percentage system.

BRP/Runequest uses a role 100% or greater mechanic if I'm not mistaken.

I've been toying with the idea of using a die-step system, similar to Savage Worlds, but without all of the silly exploding d4 nonsense.

The idea is base stats rated from d4 to d12, and a number of skills rated from d4 to d12, with very little in the way of static modifiers. Player rolls stat die + skill die vs. target number. Meet or beat for success. Not entirely sure how this would work out as I haven't sat down to crunch the numbers on it.
BRP traditionally just used your basic roll under % on d100. Very few systems use additive d100 - Rolemaster is one (though results are looked up on tables rather than using the % directly). Synnibarr uses additive d100 for shot rolls.
Your regular gamer probably doesn't enjoy double-digit addition, though you can do additive d100 by just rolling d10+ tens place and then if you fail by 1, roll again vs. TN 11 adding the units.

Separate dice for skill and stat (each ranging from d4 to d12) is exactly how the Cortex system works btw. The Supernatural and Firefly (or Serenity, whatever) RPGs both use the system, among others I'm probably not aware of.
Roog
Master
Posts: 204
Joined: Mon Sep 15, 2008 9:26 am
Location: NZ

Re: A search for an optimal resolution mechanic

Post by Roog »

PhoneLobster wrote:You use averages when looking at bonuses at an unknown point in the curve or over a known range of points in the curve. Because it's an approximate guess that is right in a general sense. It is significantly LESS bullshit than Frank's method of picking the largest extreme of bonus value on the graph which is only ever right at that one point.
Taking the average fails as a measure for dice curve comparisons, as it is too coarse, and removes any distinction between the curves.

People obviously care about different size bonuses at different point of the curve. I would seriously care about a +1 on d20 that took a save from 5% failure to 0%, and I would hardly give a shit about a bonus that takes hit chance from 95% to 100%.

Any metric for comparison of dice curves would need to take the changing size of bonus required to make the player care at different points in the same curve into account.
DragonChild
Knight-Baron
Posts: 583
Joined: Sun Mar 09, 2008 7:39 am

Post by DragonChild »

This thread needs more pictures. I am here to provide. These pictures are created by me, for a talk I did at an anime con about designing RPG system.

First: http://s457.photobucket.com/albums/qq29 ... graph1.png

This graph shows the exact odds of rolling the number you need to on a 3d6 or a 1d20. It's pretty clearcut, I think. if you need to roll a low number, you want 3d6. If you need to roll a high number, you want 1d20.

But this: http://s457.photobucket.com/albums/qq29 ... graph2.png

Is the derivative. And it's more interesting. This graph shows you what odds off success bonus you get from having an ADDITIONAL +1 onto that 1d20 or 3d6. Getting a +1 on a 1d20 is always going to increase your chance of success by 5%. But getting it on a 3d6 could be a bonus smaller than 2%, or as high as 13%.

Onto dicepools. These graph use "shadowrun dice". You roll a fistful of d6s, count the 5s and 6s for "hits", and do not count 1s, botches, or any of that shit.

This is the "plain" image. Pretty self explanatory: http://s457.photobucket.com/albums/qq29 ... graph3.png

But this is the derivative image. Pretty complicated shit: http://s457.photobucket.com/albums/qq29 ... graph4.png
PhoneLobster
King
Posts: 6403
Joined: Fri Mar 07, 2008 7:54 pm

Re: A search for an optimal resolution mechanic

Post by PhoneLobster »

Roog wrote:Taking the average fails as a measure for dice curve comparisons, as it is too coarse, and removes any distinction between the curves.
1) The average is pretty much definitively incapable of being too "coarse". It is after all an average of the "coarseness" of the range of values it is averaging. Meaning some portion of the raw data (in this case about half) is actually MORE "coarse".

2) It "removes" the distinction because when you talk about the full span of the range of the RNG there ISN'T much distinction.

3) An average is still more useful for giving an informative answer to the question "What does +2 give me?" when the DC and other modifiers are unknown or known to be spread over a potential range. Which will typically be the case during rules design.

And despite all that really I had a whole rest of a post talking about yeah, you would also ideally measure the specific values of bonuses at different points in the curve. You know, if and when you can, instead of just running with the bonus value at median for the entire curve... because that is incredibly dumb.

Edit: Oh and while I'm here...
Roog wrote: I would seriously care about a +1 on d20 that took a save from 5% failure to 0%, and I would hardly give a shit about a bonus that takes hit chance from 95% to 100%.
That is ALMOST a useful question to ask, but your sample is somewhat poorly framed and highly subjective.

Now a question like how much do I like a +2 on a 1d20 when I already have a 90% chance to hit, and how much I like it when I already have a 95% chance to hit is probably a better one.

Mixing hits and saves and talking about identical odds and modifiers basically only makes us wonder about the value of hits vs the value of saves rather than the value of the modifiers and the nature of the RNG.
Last edited by PhoneLobster on Sun Jan 23, 2011 12:40 am, edited 2 times in total.
Phonelobster's Self Proclaimed Greatest Hits Collection : (no really, they are awesome)
Parthenon
Knight-Baron
Posts: 912
Joined: Sat Jan 24, 2009 6:07 pm

Re: A search for an optimal resolution mechanic

Post by Parthenon »

deanruel87 wrote:Ummmm....Feng Shui: As seen in Feng Shui. Seriously does anything else use the Feng Shui positive/negative die mechanic? If it does I've never seen it. Anyway the up down dice thing is intuitive and crazy simple. Its math is easy to calculate and quick. A +1 to a check is still pretty big but not game-breakingly so so the system can still handle things like "A master made sword gives you +1", which makes people happy. I haven't had a ton of experience with Feng Shui so I'll leave the particulars of this for others
This is useless. It gives me no idea how the Feng Shui resolution system works. Can someone explain it for me properly please?
Roog
Master
Posts: 204
Joined: Mon Sep 15, 2008 9:26 am
Location: NZ

Re: A search for an optimal resolution mechanic

Post by Roog »

PhoneLobster wrote:1) The average is pretty much definitively incapable of being too "coarse". It is after all an average of the "coarseness" of the range of values it is averaging. Meaning some portion of the raw data (in this case about half) is actually MORE "coarse".
Of course the average can be too coarse - if it fails to capture meaningful distinctions, then it is to coarse to be used in any situation where those distinctions are important.

In this case the average % change due to some fixed bonus is too coarse to use in discussion over the difference between a uniform distribution and some other distribution. This is because for an arbitrary integer RNG over range [a+1,b], the average % increase of a +1 bonus over the relevant range will be 100/[b-a].

As the measure, average % change due to some fixed bonus, is independent of the shape of the curve, it is too coarse to capture any information related to the shape of the curve.

PhoneLobster wrote:3) An average is still more useful for giving an informative answer to the question "What does +2 give me?" when the DC and other modifiers are unknown or known to be spread over a potential range. Which will typically be the case during rules design.
If you want the answer to the question "What does +2 give me?" to be a simple %, then this requires the RNG to be flat. If you are willing to accept other answers, the you could use other RNGs.
E.g. The answer "When I'm likely to fail, +2 doubles my chance of success; when I'm likely to succeed, +2 halves my chance of failure." could be generated from a double-tailed exponential distribution.

PhoneLobster wrote:That is ALMOST a useful question to ask, but your sample is somewhat poorly framed and highly subjective.

Now a question like how much do I like a +2 on a 1d20 when I already have a 90% chance to hit, and how much I like it when I already have a 95% chance to hit is probably a better one.
The point I was attempting to make, is that a fixed % change to the odds has a variable utility, depending on the circumstances. If that utility correlates with initial chance of success, then the RNG curve could be matched to that correlation, to achieve a variety of effects.
PhoneLobster
King
Posts: 6403
Joined: Fri Mar 07, 2008 7:54 pm

Re: A search for an optimal resolution mechanic

Post by PhoneLobster »

Roog wrote:Of course the average can be too coarse - if it fails to capture meaningful distinctions, then it is to coarse to be used in any situation where those distinctions are important.
And then you suggest that basically in all cases we need to actually calculate the exact value at every point in the range.

Averages in cases like this are used as a shortcut. Because we both know that the OP and most folks around here aren't going to answer the question "what does the bonus on my +2 magic sword do?" with a graph, look up table, or formula when asked by a player. And we would be lucky as hell if they even sit down and use one of those things (and manage to USEFULLY use one) while considering the design implications of every bonus in the game.

Now you could argue that you actually NEED to do that and the "coarseness" caused by using an average as a shortcut when describing or designing bonuses for a curved RNG is just too big to deal with.

But at that point you have effectively presented an argument stating that curved RNGs just plain aren't suitable for the vast majority of game designers, or even players.

Which is fine by me because that would be just one more of about a bajillion reasons why they should never ever be used ever.

Also dice pools are purely horrendous.
Phonelobster's Self Proclaimed Greatest Hits Collection : (no really, they are awesome)
Username17
Serious Badass
Posts: 29894
Joined: Fri Mar 07, 2008 7:54 pm

Post by Username17 »

Roog, you are wasting your time. PhoneLobster does not debate rationally or honestly when people start talking about using different random number generators. Seriously: the reason it sounds like he isn't understanding very simple things about statistics is that he is deliberately misrepresenting your arguments so he can spout off more one-true-wayism about flat RNGs. Debate or discussion with him on this topic is a waste of everyone's time and will cause him to spam this thread into illegibility. He has done it before, he'll do it again.

-Username17
PhoneLobster
King
Posts: 6403
Joined: Fri Mar 07, 2008 7:54 pm

Post by PhoneLobster »

That's Frank for you. Never using reason when he could use ad hominem.
Phonelobster's Self Proclaimed Greatest Hits Collection : (no really, they are awesome)
User avatar
angelfromanotherpin
Overlord
Posts: 9745
Joined: Fri Mar 07, 2008 7:54 pm

Post by angelfromanotherpin »

PL, I was saving this for Tzor, but he quieted down the last couple of days. So congratulations, PL, you win medal.

Image
CCarter
Knight
Posts: 454
Joined: Fri Jun 11, 2010 10:41 pm

Re: A search for an optimal resolution mechanic

Post by CCarter »

Parthenon wrote: This is useless. It gives me no idea how the Feng Shui resolution system works. Can someone explain it for me properly please?
You have a positive d6 and a negative d6 (e.g. a light-colored and dark-colored die respectively). You roll them and add the positive and negative result to get your total modifier. in the case of a 6, you roll up that number, whether its on the positive or negative die. IIRC, in the case of double 6's, you roll both again and also something weird happens.
Post Reply