Also, whats up with AC at CR 6, 11, 13, and 16?What do you mean? If you look at the median/mode/mean (lesser extent, it might be off as much as 1 point from the median mode if I'm not mistaken). If you look at the whole progression each CR increase basically increases AC by 1.If you take the standard deviation and +/- it to the mean you find the range of reasonable numbers to expect. While the means at each CR match up pretty well, the standard deviation for HP increases as the CR increases. Which means the range of expected HP at higher CRs is greater.
EDIT: What I mean is the variance increases as CR increases.
EDIT 2: For AC, that range of numbers is substantially larger at CRs 6, 11, 13, and 16 than it is at any other CR.
Okay, you were looking at standard deviation. Now I see what you mean, your original statement was very vague. All it means is the range of AC at each of those levels is large. Meaning when the creatures were created the didn't really follow the guidelines. Monster are varied is really the only conclusion you can make, unless you know exactly what you will be fighting and can use that as your specific target for hp and AC.
It probably has to do with DR, increased hp, or other special abilities being used on creatures that then necessitated reducing other features (such as AC) to compensate to make the creature a managable enemy for a given CR. Though, to be honest many creatures do not meet the CR paradigm well in the first place so....
This is why we just use the average/median because it's nearly impossible to know exactly what you're fighting all the time. And because if went from fighting high AC low hp creatures at one level to fighting low AC high hp creatures at the next you could be for one at be useless at the other. So you go with a middle of the road approach. You can't be good at all things in all situations, but you can be "generally" good.
Also, whats up with AC at CR 6, 11, 13, and 16?What do you mean? If you look at the median/mode/mean (lesser extent, it might be off as much as 1 point from the median mode if I'm not mistaken). If you look at the whole progression each CR increase basically increases AC by 1.
I think what he means is that DPR formulas overvalue damage and under value accuracy. By showing the extreme that fact becomes more apparent than it otherwise would be. So he isn't saying it never happens, but that it always happens to a lesser, but still extremely valid, degree.DPR doesn't undervalue accuracy, it is predicting the likelihood of hitting and how much damage you will deal. It includes the probability for hitting against a target AC and the chance of scoring a critical hit. It does not account for damage above and beyond that required to kill a creature aka "Overkill", but except at low levels you general don't expect to kill a creature in a single blow so it's not as important (though still something to be mindful of).
The extremes just aren't valid because of a false dichotomy. You're not going to end up with a character that can have 100 dpr and only a 5% chance to hit or the same character that deals 10 damage and a 95% chance to hit. You just don't end up in such a situation because no single option changes a character that drastically. DPR can be used to evaluate the difference between two options and how a characters expected damage output will change because of these. BY choosing crazy extreme values you get weird results.
But if were looking at a real character a the choice is between a +1 to hit, or a -1 to hit for +2 damage that is a real choice that can represent something for that character.
You could take an opposite example from the one initially proposed by yourself and say that against an enemy that has 10,000 and the choice between a character which has a 95% chance to deal 10 damage or a character with 5% chance to deal 10,000 damage...
The base formula (the one that I quoted that BigDTBone wrote) over samples damage. Because of that it puts emphasis on damage.
Like I mentioned if you adjust the formula to clamp damage done on MAX(damage done, Enemy HP + 1), the DPR suddenly makes a lot more sense.
My scenario of 95% Accuracy 10 Damage vs. 5% Accuracy 1000000 Damage, and enemy health of 10 suddenly makes more sense.
95% Accuracy's DPR is 9.5.
5% Accuray's DPR is 0.55.
The choice then becomes clear.
Like I mentioned if you adjust the formula to clamp damage done on MAX(damage done, Enemy HP + 1), the DPR suddenly makes a lot more sense.
Sure, but it adds more complexity to the equation and is only important when you have the potential for your 1 round damage to exceed or equal the enemy's hp. Which isn't a common case above level 2, exlcuding succesive rounds on already damaged enemies. Of course that also doesn't account for what happens based on your allies actions either, or if you have another enemy adjacent to continue attacking, etc.
The point is that one single equation cannot soley determine what is best, but both can have their uses. The problem is that some people act like DPR formula is the end all be all. It is not. But it is useful.
This thread basically (at least the way I took it) was saying "you're dumb for trying to use dpr to understand anything!". I object to such a statement. People need to understand where dpr is useful and where it's limitations exist. Could it be improved, sure. Do I feel like doing the extra math for it? Not really. I know where the rough edges show on DPR calculations, so I avoided putting too much stock in it.
But if I want to know if weapon focus or rapid shot is better for archer ranger, the DPR formular represents a relative simply way for me to find out.
Thirdly, the issue at hand (DPR calculations) is in general highly abstract and not representative of the diversity of actual play, I hence I do not consider that point all too relevant.
Truer words have never been uttered, and will never be as universally ignored on these boards.
Thirdly, the issue at hand (DPR calculations) is in general highly abstract and not representative of the diversity of actual play, I hence I do not consider that point all too relevant.Truer words have never been uttered, and will never be as universally ignored on these boards.
As has been stated many many many many times, it's not intended to.
Like I mentioned if you adjust the formula to clamp damage done on MAX(damage done, Enemy HP + 1), the DPR suddenly makes a lot more sense.Sure, but it adds more complexity to the equation and is only important when you have the potential for your 1 round damage to exceed or equal the enemy's hp. Which isn't a common case above level 2, exlcuding succesive rounds on already damaged enemies. Of course that also doesn't account for what happens based on your allies actions either, or if you have another enemy adjacent to continue attacking, etc.
The point is that one single equation cannot soley determine what is best, but both can have their uses. The problem is that some people act like DPR formula is the end all be all. It is not. But it is useful.
This thread basically (at least the way I took it) was saying "you're dumb for trying to use dpr to understand anything!". I object to such a statement. People need to understand where dpr is useful and where it's limitations exist. Could it be improved, sure. Do I feel like doing the extra math for it? Not really. I know where the rough edges show on DPR calculations, so I avoided putting too much stock in it.
But if I want to know if weapon focus or rapid shot is better for archer ranger, the DPR formular represents a relative simply way for me to find out.
I never said anyone was dumb for using it, and in fact in my first post I even said that it could be used to compare options.
Just because something isn't accurate doesn't mean that it's entirely useless. At the very least it represents the game as we understand it best at this point. You build upon knowledge, you don't just throw it away.
I'm sorry I came off that way to you Claxon, and I hope this clarifies my stance.
RPG Superstar 2012 Top 32
|
| 1 person marked this as a favorite. |
Just because something isn't accurate doesn't mean that it's entirely useless. At the very least it represents the game as we understand it best at this point. You build upon knowledge, you don't just throw it away.
Yet just like a page back, you were talking about how people shouldn't continue using it.
So it's not useless, it's just not worth using? You don't want to throw it away, you just think people should stop using it?
Just because something isn't accurate doesn't mean that it's entirely useless. At the very least it represents the game as we understand it best at this point. You build upon knowledge, you don't just throw it away.Yet just like a page back, you were talking about how people shouldn't continue using it.
So it's not useless, it's just not worth using? You don't want to throw it away, you just think people should stop using it?
You misquoted me, they shouldn't use it to make character decisions. For comparing things like weapon focus or rapid shot is another matter.
Edit: I've gone back and scanned my posts, can you quote anywhere that I said that?
|
RPG Superstar 2012 Top 32
|
You misquoted me, they shouldn't use it to make character decisions. For comparing things like weapon focus or rapid shot is another matter.
Comparing Weapon Focus and Rapid Shot is different than making character decisions? Say what now?
Edit: I've gone back and scanned my posts, can you quote anywhere that I said that?
In this post, you said:
I'm just saying the method is wrong, and we should probably try to figure out why, rather than continue to use it.
And that was in the context of your demonstration of how DPR treats the decision to use Weapon Focus or Power Attack. I'm not sure how that's different than your other comment's context of using DPR to decide between Weapon Focus and Rapid Shot.
You misquoted me, they shouldn't use it to make character decisions. For comparing things like weapon focus or rapid shot is another matter.Comparing Weapon Focus and Rapid Shot is different than making character decisions? Say what now?
Edit: I've gone back and scanned my posts, can you quote anywhere that I said that?In this post, you said:
I'm just saying the method is wrong, and we should probably try to figure out why, rather than continue to use it.And that was in the context of your demonstration of how DPR treats the decision to use Weapon Focus or Power Attack. I'm not sure how that's different than your other comment's context of using DPR to decide between Weapon Focus and Rapid Shot.
To expand on that "rather than continue to use it", Note the condition. I'm not saying hold the presses and stop using the existing DPR formula now. There isn't anything better to use. I don't have any method that's better. All I'm saying that rather than ignore the issues with it, we should figure out why there are issues with it.
I don't see how any of that treads into "you're dumb for trying to use dpr to understand anything!" territory.
EDIT: I didn't say "people shouldn't continue using it." I said "we should figure out why, rather than continue to use it".
We know why there are issues, because you choose to examine corner cases and an apply the tool in scenarios where it isn't intended to be used.
Edit: basically, DPR is perfectly fine. If there is some other metric you want to measure then you should feel free to do so, but just because you want to measure something else doesn't mean there is an issue with DPR.
Ie, just because you like nails doesn't mean that screwdrivers have issues that we need to examine.
Do you know why I'm doing that?
We know why there are issues, because you choose to examine corner cases and an apply the tool in scenarios where it isn't intended to be used.
Edit: basically, DPR is perfectly fine. If there is some other metric you want to measure then you should feel free to do so, but just because you want to measure something else doesn't mean there is an issue with DPR.
Ie, just because you like nails doesn't mean that screwdrivers have issues that we need to examine.
Then why are you commenting on a thread that started with this?
Since there is a current discussion going on about Power Attack being a trap, I would like to offer some considerations on the common practice of DPR calculations and how the miss out on some important aspects. I shall assume that the reader is familiar with the common DPR formula and will discuss it in depth.
Do you know why I'm doing that?
Calling out screwdrivers because you have a pocketful of nails?
Presumably because you have a pocket full of nails. IE, you are in search of a tool to measure something else and DPR doesn't do it.
So figure out what it is you want to measure and measure it. That is not DPR's problem.
We know why there are issues, because you choose to examine corner cases and an apply the tool in scenarios where it isn't intended to be used.
Edit: basically, DPR is perfectly fine. If there is some other metric you want to measure then you should feel free to do so, but just because you want to measure something else doesn't mean there is an issue with DPR.
Ie, just because you like nails doesn't mean that screwdrivers have issues that we need to examine.
Then why are you commenting on a thread that started with this?
Since there is a current discussion going on about Power Attack being a trap, I would like to offer some considerations on the common practice of DPR calculations and how the miss out on some important aspects. I shall assume that the reader is familiar with the common DPR formula and will discuss it in depth.
Because you aren't offering your new metrics to measure your new ideas, you are just criticizing DPR. DPR doesn't have the problems you suggest.
A screwdriver isn't broken just because you can't use it to hammer nails.
Do you know why I'm doing that?Calling out screwdrivers because you have a pocketful of nails?
Presumably because you have a pocket full of nails. IE, you are in search of a tool to measure something else and DPR doesn't do it.
So figure out what it is you want to measure and measure it. That is DPR's problem.
No, I'm attempting to have a discussion on the OP's topic.
|
DPR is fine - as long as you understand its limitations when using it. It's far from the be-all end-all for character creation.
My own DPR calculations generally include higher ACs as I find that in practice ACs are higher than Bestiary standrds. Both because NPCs have higher ACs in general (at least until super high levels)- and often Bestiary creatures are given some basic armor in addition to their profiles.
We know why there are issues, because you choose to examine corner cases and an apply the tool in scenarios where it isn't intended to be used.
Edit: basically, DPR is perfectly fine. If there is some other metric you want to measure then you should feel free to do so, but just because you want to measure something else doesn't mean there is an issue with DPR.
Ie, just because you like nails doesn't mean that screwdrivers have issues that we need to examine.
Then why are you commenting on a thread that started with this?
Since there is a current discussion going on about Power Attack being a trap, I would like to offer some considerations on the common practice of DPR calculations and how the miss out on some important aspects. I shall assume that the reader is familiar with the common DPR formula and will discuss it in depth.Because you aren't offering your new metrics to measure your new ideas, you are just criticizing DPR. DPR doesn't have the problems you suggest.
A screwdriver isn't broken just because you can't use it to hammer nails.
Can you tell me where anything in the OP's topic amounted to offering my new metrics? And if I were to make a new metric? Then I'd first have to understand what the problem with DPR is. To provide a new metric without first "criticizing" DPR, I'd have to go read tea leaves.
As for:
DPR doesn't have the problems you suggest.
Then I ask you to take what I've suggested, and prove that it is wrong.
EDIT: And as I've said before, I don't have a new formula, metric, what have you.
DPR is fine - as long as you understand its limitations when using it. It's far from the be-all end-all for character creation.
My own DPR calculations generally include higher ACs as I find that in practice ACs are higher than Bestiary standrds. Both because NPCs have higher ACs in general (at least until super high levels)- and often Bestiary creatures are given some basic armor in addition to their profiles.
One thing you can do instead is look at DPR against a standard AC and the also look at the variance of a +/- to hit or damage. You can take a look at how much your DPR can vary with buffs that way, and it becomes useful for someone buffing the party to know where they will do the most good. For instance, a +1 to hit means a hell of a lot more on a rogue (very high damage with low base to hit) than it does on a fighter (above average of both hit and damage). Every +1/-1 you get will be .05% of your damage, assuming no auto-hits (which in my experience is impractical).
|
RPG Superstar 2012 Top 32
|
| 2 people marked this as a favorite. |
If you want to try and actually evaluate the effect of +1 to hit, you'll need to integrate an equation for damage on a successful hit with respect to the chance of hitting across the domain of potential ACs. That will still not be accurate, but it will be way more predictive than DPR...
Actually, that is DPR. What did you think DPR meant?
tl;dr
DPR leaves out the value of accuracy by itself. My considerations are intended to show that accuracy has important effects by itself. They are are not intended to be used in character building to judge the power of a build (though they can be).
Consider what I said not an engineers tool to build characters but a philosopher's argument to establish that focussing on accuracy tends to be a good idea, even if DPR might not show this.
The real problem with your case is that it doesn't actually comment on DPR.
The 30/.3 and 10/.9 have equivalent DPR. Thus you tell us... functionally nothing about the actual formula, simply that if you have the choice between the two, you'll generally want accuracy.
And, to be more accurate about it-- you'll require more damage in order to be willing to sacrifice that accuracy. To which the general response is, and frankly should be, "of course. Why are we reading this?"
The game already tells us this. Look at Weapon Specialization and Weapon Focus and it already illustrates this point.
So... honestly, your core argument is pointless, because anyone with a modicum of system mastery can already see it.
Now! If you want to illustrate to us that more accuracy > more damage even in opposition of the DPR formula, and can provide rigorous evidence to prove it, you have something interesting.
Which is, to shift gears slightly, sort of my issue with Ssyvan's repetition of this idea that he's trying to follow what the OP says. The OP says literally nothing about DPR. If we're trying to discuss DPR in this thread, based on what the first post actually says, we are wasting time. All the OP does is offer a way to decide between two builds with equivalent DPR. Is that what we want to discuss, or are we okay with veering away from what the OP actually says?
As he said he assumes the reader is already familiar with the DPR formula. He chose those two options 30/.3 and 10/.9 because they have the same DPR. This is important because DPR tells you "there is no mechanical benefit to either choice." The OP then goes on to prove that there is a mechanical benefit to one of the options.
So the OP has already met your challenge quoted below:
Now! If you want to illustrate to us that more accuracy > more damage even in opposition of the DPR formula, and can provide rigorous evidence to prove it, you have something interesting.
When he shows that the option with higher accuracy is better than the option with higher damage.
On DPR:
DPR is perhaps the WORST indicator of character combat potential possible. It is a completely worthless statistic. Not just slightly useful, not flawed, but totally useless in every way. Not a single character decision should EVER be made by reference to DPR. In fact, it's DPR Olympics that I think has caused some of the questionable choices by the design staff. But that's another discussion.
DPR is wholly mathematically incapable of telling you anything useful about a character build. You'd be just as well off counting the number of vowels in your character's name and multiplying it by 4. That's how irrelevant it is.
Why? Because it fails as any sort of mathematical predictor. Look at the formula posted earlier in the thread, the one that is generally accepted as a valid DPR calculation. What do you notice? It's a linear equation. A linear equation cannot mathematically describe the combination of damage and accuracy with respect to lethality! Ever.
Accuracy is not a linear function in PF (neither is damage, because of monster hit point variations, but ignore that for the moment). The fact that it varies between static endpoints (5% and 95%) based on AC and will not have a consistent impact on damage outcomes means that you cannot simulate its effect with a linear equation.
Any attempt to consistently evaluate a particular combination of To Hit, Damage, AC, and HP would need to calculate the area under the curve (using calculus) formed by the domain of AC values and the relationship between accuracy and damage, extended into a three-dimensional volume if intending to take into account monster HP. A linear equation cannot even come close to simulating the effect of changes in this relationship. A linear DPR formula is less than useless, because it creates a false sense of providing information when it doesn't.
I don't know where the focus on DPR came from... personally, I blame MMOs and DPS. There might be a valid relationship there, because the amount of damage is...
DPR does give you a yardstick with which to measure certain options, but there are some issues with it, which is what I'm (we're?) trying to discuss.
In fact it is clear where DPR falls short is by the variables missing from its equation. As some have pointed out, Ratio of Full Attacks vs. Standard Attacks (useful for gauging the value of Vital strike and Cleave), Enemy HP (to prevent oversampling of damage), or as of yet discussed your own AC and HP (To gauge the risk of attempting certain actions. Like Fighting Defensively, or using Charge).
If you want to try and actually evaluate the effect of +1 to hit, you'll need to integrate an equation for damage on a successful hit with respect to the chance of hitting across the domain of potential ACs. That will still not be accurate, but it will be way more predictive than DPR...Actually, that is DPR. What did you think DPR meant?Nope, it is not DPR. It's the integral of the relationship between accuracy and damage. The number would not necessarily relate to an actual damage amount. Instead it would be a metric comparing the relative outcomes of damage within the parameters. The whole reason not to use DPR in the first place is that an average in one set of conditions doesn't give the whole picture. You would compare values to gauge effectiveness, not work the equation to say "I'm going to hit for this much."
It's definitely not the DPR calculation given earlier in the thread (which is my point).
Right, I see your point. But I think the general assumption is the user of the DPR formula would gauge the results across a range of conditions.
|
RPG Superstar 2012 Top 32
|
Nope, it is not DPR. It's the integral of the relationship between accuracy and damage.If you want to try and actually evaluate the effect of +1 to hit, you'll need to integrate an equation for damage on a successful hit with respect to the chance of hitting across the domain of potential ACs. That will still not be accurate, but it will be way more predictive than DPR...Actually, that is DPR. What did you think DPR meant?
DPR includes the relationship between accuracy and damage. Where did you get the idea that it didn't?
Instead it would be a metric comparing the relative outcomes of damage within the parameters. The whole reason not to use DPR in the first place is that an average in one set of conditions doesn't give the whole picture. You would compare values to gauge effectiveness,
So, run the DPR calculation more than once, against a range of AC values? That seems to be what you're describing, but that's a very long way from "DPR has absolutely no value at all".
not work the equation to say "I'm going to hit for this much."
This is not what DPR says.
I don't know why you think it is.
If you want to try and actually evaluate the effect of +1 to hit, you'll need to integrate an equation for damage on a successful hit with respect to the chance of hitting across the domain of potential ACs. That will still not be accurate, but it will be way more predictive than DPR...Actually, that is DPR. What did you think DPR meant?
The formula generally used for DPR is:
((%hit*average damage) + (%crit chance*crit multiplier-1*%hit*average damage)) repeated for each attack.
Average damage is easy to determine based off of the build, so the only real question is %hit, for which you need to know an enemy AC. It is generally calculated against an AC that is relevant for the level you are discussing so that you have a %hit and can compare numbers. Note: unless you are near a corner case where %hit is only a 1 or 20, a +1 to hit results in adding 5% of your base damage. That corner case happens frequently, unfortunately.
I haven't participated in newer DPR Olympics, but the original one on here asked people to also include how much variance there was if you received a buff or debuff (+1/-1), how much you got with an extra haste attack, your standard action damage, and a few other considerations. It also forced builds to meet minimum defensive standards and to acknowledge when or why you might deviate from them (low AC but above average HP on a barbarian for instance, or bad AC but miss chance).
Nope, it is not DPR. It's the integral of the relationship between accuracy and damage.If you want to try and actually evaluate the effect of +1 to hit, you'll need to integrate an equation for damage on a successful hit with respect to the chance of hitting across the domain of potential ACs. That will still not be accurate, but it will be way more predictive than DPR...Actually, that is DPR. What did you think DPR meant?DPR includes the relationship between accuracy and damage. Where did you get the idea that it didn't?
Instead it would be a metric comparing the relative outcomes of damage within the parameters. The whole reason not to use DPR in the first place is that an average in one set of conditions doesn't give the whole picture. You would compare values to gauge effectiveness,So, run the DPR calculation more than once, against a range of AC values? That seems to be what you're describing, but that's a very long way from "DPR has absolutely no value at all".
not work the equation to say "I'm going to hit for this much."This is not what DPR says.
I don't know why you think it is.
Rather the opposite is likely, the reason he thinks it is worthless is because he believes that is what it is. There is a bunch of misunderstanding regarding precisely what DPR is. Which is odd because it is a straight forward topic.
| 3 people marked this as a favorite. |
Rather the opposite is likely, the reason he thinks it is worthless is because he believes that is what it is. There is a bunch of misunderstanding regarding precisely what DPR is. Which is odd because it is a straight forward topic.
I suspect a lot of it stems from the crowd whose thought process can be summed up as "Optimizers mention DPR, therefore it is EVIL and WRONG!"
| 2 people marked this as a favorite. |
@Eirikrautha:
Firstly, there is NO calculus involved, nor should there be. All values used in Pathfinder are discrete, not continuous. You can't roll a 26.89 on an attack roll, you can't have an AC of 19.6385, and you can't roll a damage of 67.538. Everything is discrete. Calculus is not needed, areas under the curve are not needed. Sums and probability-weighted averages, not integrals, are the mathematically correct tool.
Second, it is not useless as a predictor. Granted, it's not perfect as a predictor, but it's certainly meaningful. If you took every PF character, plotted their calculated DPR on one axis and their total real-game damage divided by total real-game rounds of combat on a second axis, there would be a correlation between the two. R squared of a trendline would not be zero, as you imply by saying it is "a completely worthless statistic" that is "totally useless in every way."
Personally I would wager that a DPR value calculated using "standard + full attack" instead of just "full-attack" (except for archers) would have a higher R squared, but either way, there is undoubtedly a positive correlation between calculated DPR and real-game DPR. Not 100% correlation, but correlation nonetheless.
I suspect a lot of it stems from the crowd whose thought process can be summed up as "Optimizers mention DPR, therefore it is EVIL and WRONG!"YAY CAMPS, NO ONE EXISTS OUTSIDE THEM!! \(^o^)/
I am so so sorry Jiggy, so sorry.
Edit: Just to be clear I'm not being sarcastic. I'm apologizing for independently constructing the idea of a camp, and lumping you into one side, when all you were doing was making observations.
I suspect a lot of it stems from the crowd whose thought process can be summed up as "Optimizers mention DPR, therefore it is EVIL and WRONG!"YAY CAMPS, NO ONE EXISTS OUTSIDE THEM!! \(^o^)/
Umm... I think the number of alternative positions in this are very slim. I'm not going to say they don't exist, but I can't imagine there being very many.
|
RPG Superstar 2012 Top 32
|
Umm... I think the number of alternative positions in this are very slim. I'm not going to say they don't exist, but I can't imagine there being very many.I suspect a lot of it stems from the crowd whose thought process can be summed up as "Optimizers mention DPR, therefore it is EVIL and WRONG!"YAY CAMPS, NO ONE EXISTS OUTSIDE THEM!! \(^o^)/
Reasons to think DPR is unhelpful, other than "optimizers mention it so it's evil and wrong":
• Thinking it's like the "damage" column in the Monster Creation chart, where the number is how much damage is dealt if EVERY attack hits.• Thinking it's being used as the only tool for character creation, rather than being applied only to specific elements of it.
• Simply not understanding that it includes accuracy and other factors.
Okay, I thought there were more, but it turns out most of them are just different ways to misunderstand exactly what DPR is (and is for).
But still, plenty of room outside the finger-pointing box.
Umm... I think the number of alternative positions in this are very slim. I'm not going to say they don't exist, but I can't imagine there being very many.I suspect a lot of it stems from the crowd whose thought process can be summed up as "Optimizers mention DPR, therefore it is EVIL and WRONG!"YAY CAMPS, NO ONE EXISTS OUTSIDE THEM!! \(^o^)/Reasons to think DPR is unhelpful, other than "optimizers mention it so it's evil and wrong":
• Thinking it's like the "damage" column in the Monster Creation chart, where the number is how much damage is dealt if EVERY attack hits.
• Thinking it's being used as the only tool for character creation, rather than being applied only to specific elements of it.
• Simply not understanding that it includes accuracy and other factors.Okay, I thought there were more, but it turns out most of them are just different ways to misunderstand exactly what DPR is (and is for).
But still, plenty of room outside the finger-pointing box.
Huh. Turns out that whole Socratic method thing works. Neat. Good summary by the way.
Umm... I think the number of alternative positions in this are very slim. I'm not going to say they don't exist, but I can't imagine there being very many.I suspect a lot of it stems from the crowd whose thought process can be summed up as "Optimizers mention DPR, therefore it is EVIL and WRONG!"YAY CAMPS, NO ONE EXISTS OUTSIDE THEM!! \(^o^)/Reasons to think DPR is unhelpful, other than "optimizers mention it so it's evil and wrong":
• Thinking it's like the "damage" column in the Monster Creation chart, where the number is how much damage is dealt if EVERY attack hits.
• Thinking it's being used as the only tool for character creation, rather than being applied only to specific elements of it.
• Simply not understanding that it includes accuracy and other factors.Okay, I thought there were more, but it turns out most of them are just different ways to misunderstand exactly what DPR is (and is for).
But still, plenty of room outside the finger-pointing box.
True, there are plenty of ways that someone to misunderstand what DPR is.
Of course, I was commenting on where a lot of those misunderstandings originate from, not claiming that only anti-optimizers dislike DPR. Helps when you read things in context. I think it is fair to say that the rabid anti-optimizers spread a lot of misinformation about things like optimizing, DPR, etc.
I'm an optimizer that disdains dpr simply because too many people embrace it as the single metric that defines a damage dealing character.
Does that count?
Let's just agree that Power Attack is a generally good feat with a "higher risk, higher reward" mechanic and leave it at that.
| 1 person marked this as a favorite. |
Let's just agree that Power Attack is a generally good feat with a "higher risk, higher reward" mechanic and leave it at that.
Power attack was the last thread, this thread is DPR.
| 2 people marked this as a favorite. |
Jiggy,
The other alternative is that while its ease of calculation makes it a very important metric its not the full story. Excess damage done by you or your party before the opponents turn is meaningless. This view gives accuracy an even higher importance than it has just from its inclusion in the DPR calculations.
Also, though shalt not tempt the dice gods, polyhedral be thy names.
| 1 person marked this as a favorite. |
Yes.
Maybe this is a way to put it that everyone will understand:
Accuracy has effects that are not captured by commonly used DPR calculations. This is precisely what I mean when I say that DPR undervalues accuracy.
This does not make those calculations useless, nor should it motivate a crusade against DPR. It also does not mean that DPR fails to consider accuracy.
Nor should anyone draw the conclusion that power attack is a bad feat or that it's never a good idea to push damage. In fact, I intended no hands on advice at all, but to point out some mathematical facts about the combat rule mechanism. My goal is understanding, rather than application. That is not to say that no practical application can be made, just that I am not terribly interested in those practical applications.
KutuluKultist makes a good point. Accuracy does have effects that are not captured by commonly used DPR calculations. (As he mentions, this does not mean DPR fails to consider accuracy.)
It should also be noted, damage also has effects that are not captured by commonly used DPR calculations. Of course, this also does not mean DPR fails to consider damage.
Example A) Against an opponent with 7 HP and AC 13, you are doing 1d6+6 damage (not Power Attacking) or 1d6+9 damage (Power Attacking.) In this case, the Power Attack is obviously the wrong option, even if it would generate a higher DPR. In this case, DPR undervalues accuracy.
Example B) Against an opponent with 13 HP and high AC, you are again attacking for 1d6+6 damage (without PA) or 1d6+9 damage (with PA.) In this case, while DPR might tell you not to use Power Attack, if you hit with a Power Attack you are 50% likely to KO them outright, whereas if you hit with a regular attack you need a crit to KO them. In this case, DPR overvalues accuracy.
Example C) You have a rider effect on a successful hit, such as a magus using Frostbite. In this case, DPR undervalues accuracy.
Example D) The opponent has a defensive effect that triggers when it is hit, such as Stunning Barrier or spell-storing armor. In this case, DPR overvalues accuracy.
Now, I'm not saying that anyone is advocating using DPR as a guideline in any of these cases. My only point here is that, sometimes DPR undervalues accuracy, and sometimes it overvalues accuracy.
KutuluKultist makes a good point. Accuracy does have effects that are not captured by commonly used DPR calculations. (As he mentions, this does not mean DPR fails to consider accuracy.)
It should also be noted, damage also has effects that are not captured by commonly used DPR calculations. Of course, this also does not mean DPR fails to consider damage.
Example A) Against an opponent with 7 HP and AC 13, you are doing 1d6+6 damage (not Power Attacking) or 1d6+9 damage (Power Attacking.) In this case, the Power Attack is obviously the wrong option, even if it would generate a higher DPR. In this case, DPR undervalues accuracy.
Example B) Against an opponent with 13 HP and high AC, you are again attacking for 1d6+6 damage (without PA) or 1d6+9 damage (with PA.) In this case, while DPR might tell you not to use Power Attack, if you hit with a Power Attack you are 50% likely to KO them outright, whereas if you hit with a regular attack you need a crit to KO them. In this case, DPR overvalues accuracy.
Example C) You have a rider effect on a successful hit, such as a magus using Frostbite. In this case, DPR undervalues accuracy.
Example D) The opponent has a defensive effect that triggers when it is hit, such as Stunning Barrier or spell-storing armor. In this case, DPR overvalues accuracy.
Now, I'm not saying that anyone is advocating using DPR as a guideline in any of these cases. My only point here is that, sometimes DPR undervalues accuracy, and sometimes it overvalues accuracy.
This is again asking a screwdriver to hammer a nail. DPR happens in a vacuum. DPR tells you average full round damage output against a given armor class. That is all. Nothing says the AC is even a single opponent, or an opponent at all. Because it is measuring a raw damage output.
The only value DPR places on accuracy is as its function of likeliness to hit on a given attack. You know THE EXACT SAME VALUE IT HAS WHEN YOU ROLL A D20 AT THE TABLE.
DPR isn't meant to be a tool that tells you is it better to power attack or not in this specific circumstance. And it certainly isn't a tool that tells you if you should take a particular feat based on how you perform in a specific circumstance. DPR happens in a vacuum. If you find yourself saying, "what about when XYZ happens or if the opponent has ABC hit points left, or they have DR, or whatever," you are trying to use a screwdriver to hammer a nail.
As I just said, "I'm not saying that anyone is advocating using DPR as a guideline in any of these cases."
The post was to show there's another side of the story, mainly directed at the people who have mentioned that DPR undervalues accuracy.
|
RPG Superstar 2012 Top 32
|
So far, the only examples I've seen of "DPR undervalues accuracy" have been either ridiculous made-up numbers that don't reflect the actual game, or already-one-shotting situations at 1st-2nd level where the math is weird for pretty much every mechanic in the game. Neither of these is a valid foundation for a general claim that DPR undervalues accuracy. If anyone has other, more relevant examples, I'd be curious to see them.
So far, the only examples I've seen of "DPR undervalues accuracy" have been either ridiculous made-up numbers that don't reflect the actual game, or already-one-shotting situations at 1st-2nd level where the math is weird for pretty much every mechanic in the game. Neither of these is a valid foundation for a general claim that DPR undervalues accuracy. If anyone has other, more relevant examples, I'd be curious to see them.
I'll see what I can do when I get home. I worked a bit more on that simulation I mentioned and wrote some unit tests which caught some bugs I'd missed. By the time I fell asleep I was getting the values that DPR and RTK were predicting (so I'm a bit more confident in the simulation's accuracy).
Can I ask if there is anything specific you'd like to see?
-I was working with the 6th level Half-Elf fighter in the NPC Codex against CR 6 monster creation values within the standard deviation provided in that linked google docs spreadsheet. But it'd be a simple matter to run the test at different given values.
Edit: The simulation runs a million "encounters". The attacker always does full-attacks. The defender never attacks. The simulation runs until the attacker has reduced the defender to -1 HP.
I capture the following data:
-Average damage done over these million encounters (Should match DPR)
-Average length of an encounter in rounds (Should match RTK predicitons)
-The number of encounters that lasted 1 round, 2 rounds, 3 rounds, ...
-The percentage of encounters that lasted 1, 2, 3... rounds
-Percentage of rounds that resulted in damage
If there is something else you'd like to know, just let me know.
Well, keep in mind that if an enemy is badly hurt, its HP could drop low enough that the "1st-2nd level math" situation (i.e., overkill is an issue and accuracy becomes undervalued) is relevant at all levels.
"Rider effects" are also real-game scenarios where accuracy is undervalued. For example, if you are trying to disrupt a caster's spell, generally making sure you hit is more important than hitting hard.
I'll reiterate, I'm not trying to put up a straw man and say anyone is arguing that you should use DPR as a deciding metric in these situations. It just shows that there is extra value to accuracy.
Well, keep in mind that if an enemy is badly hurt, its HP could drop low enough that the "1st-2nd level math" situation (i.e., overkill is an issue and accuracy becomes undervalued) is relevant at all levels.
"Rider effects" are also real-game scenarios where accuracy is undervalued. For example, if you are trying to disrupt a caster's spell, generally making sure you hit is more important than hitting hard.
I'll reiterate, I'm not trying to put up a straw man and say anyone is arguing that you should use DPR as a deciding metric in these situations. It just shows that there is extra value to accuracy.
In other news; the color purple is terrible at predicting the stock market.
I know that no one is saying you should be using the color purple for that, but since we were all talking about purple anyway and predicting the stock market is a thing that we want to do, I felt like I should bring it up.
Well, keep in mind that if an enemy is badly hurt, its HP could drop low enough that the "1st-2nd level math" situation (i.e., overkill is an issue and accuracy becomes undervalued) is relevant at all levels.
"Rider effects" are also real-game scenarios where accuracy is undervalued. For example, if you are trying to disrupt a caster's spell, generally making sure you hit is more important than hitting hard.
I'll reiterate, I'm not trying to put up a straw man and say anyone is arguing that you should use DPR as a deciding metric in these situations. It just shows that there is extra value to accuracy.
In other news; the color purple is terrible at predicting the stock market.
I know that no one is saying you should be using the color purple for that, but since we were all talking about purple anyway and predicting the stock market is a thing that we want to do, I felt like I should bring it up.
You're using DPR as a build calculator, right? (i.e. Using the number DPR generates for a given build and comparing that number to another build. The build that generates the highest number is the better build)
So for me to accept DPR, it needs to be shown that DPR properly predicts the better build in the general case.
For you to accept that there are flaws in DPR, it needs to be shown that DPR doesn't properly predict the better build in the general case.
Is that correct?
EDIT: I'm trying to set some guidelines here for what needs to be shown to who.