Current Known Issues - AUG 30

[Armando]

Re: Current Known Issues - AUG 30

I don't know how pDif would always end up as 2 for 1 handed weapons at "capped attack". That would imply 0 variance (in other words, no +/- 0.4). It could be that these players were actually getting up to 2.4 pDif (very likely against TW mobs), the game then gave the +/- 0.4 variance (giving a range of 2.0 to 2.8), and then the 1 handed cap got applied, pushing them down to 2.0.

First off, PDIF variance isn't always -/+ 0.4. And where are you getting that 2.0 to 2.8 number? The original variance was 1.6 to 2.4. That aside, yes, it does not vary at all now. Why do people see a constant number and think of their damage any differently? It's no more capped than when it could vary.

I took RNG out to Dangruf Wadi last night, and solo'ed from level 11 to level 13. I've got the parse from that time, and there's definitely damage variance for my Axe hits.

You're not going to cap your PDIF on EP mobs.

[LyonheartLakshmi]

Re: Current Known Issues - AUG 30

First off, PDIF variance isn't always -/+ 0.4.

Did the variance ever go down to -/+ 0.0 against anything pre-patch? (And I mean other than when a player attacked a mob 40 levels above them.)

And where are you getting that 2.0 to 2.8 number? The original variance was 1.6 to 2.4.

I'm getting 2.0 to 2.8 as a hypothesis as to how SE might have changed, even mistakenly, pDif calculations. Given the current list of known issues and how 2 handers seem to be hitting for 3.0 pDif on non-crits and crits alike, it's not a stretch to guess at how SE might have mangled the code.

Old pseudo code:
1. calculate player attack/mob defense ratio
2. cap at 2.0.
3. add in variance
4. add 1.0 for critical hits
5. cap at 3.0 if necessary

Botched new pseudo code:
1. calculate player attack/mob defense ratio
2. add in variance
3. if 1 handed, cap at 2.0
4. add 1.0 for critical hits
5. cap at 3.0 if necessary (for 2 handed weapons)

That aside, yes, it does not vary at all now... You're not going to cap your PDIF on EP mobs.

If variance in damage exists against EP mobs, why doesn't it exist against TW mobs? At what point does that variance become 0? Is it just for H2H that variance has been reduced to 0? Is it just against TW mobs? If so, what does the possible pDif range against EP+ mobs look like?

There are too many questions that are unanswered to simply say "MNKs will still deal the same average damage." To me, it sounds like you've already reached the conclusion that MNK damage has not been "nerfed", and are looking for ways to support that conclusion with how pDif might currently be working.

I'm trying to understand how pDif is working first. Once I've gotten my head around that, then I'll conclude whether MNK/1-handed damage has been nerfed or MNK/1-handed damage is still the same on average as it was before.

[Amele]

Re: Current Known Issues - AUG 30

wow a lot of stuff here;

actually they could leave the str and damage the same and just reduce the dex to 1.5=1 which would make crits worth something. the problem with the ws is we are almost always hitting the max.

this wouldn't affect crits. the issue currently is that normal swing pdif caps at 3.0pDIF (something like 3.4 cRatio) which is a ridiculously high number people! we're talking something like 1,300 attack to cap on merit mobs. and crit pdif also caps at 3.0pDIF. changing dex/accuracy relation wouldn't affect this because the issue is attack. (solution: make crits cap higher than regular hits; do this by either lowering one or raising the other, or both. etc.)

There was a change in the pDif formula that made them cap earlier.

this isn't quite true. before, the cap on pdif was at cRatio 2.0 where func_pdif(cRatio) varied from 1.6 to 2.4 (this is something like 815 attack, which is 'only' going to take a net of 219 attack from gear with double minuet and red curry on merit mobs)



(this is speculation) now, func_pdif(cRatio) doesn't appear to cap,
(this is fact) the returned pDIF is clamped at 2.0 for 1h users
(this is accepted convention but was never proven) the old func_pdif(cRatio) return is normal, with mean and median both at approximately (pdif_min+pdif_max)/2.

this makes:

Isn't it just "2" on TW and 1.6 to 2 on anything just a bit higher than T?

correct, more or less. (at 2.2 cRatio you'd have a minimum around ~1.8 etc)

if they made no other changes, this would result in a slight reduction (starting at 0%) at 1.667 cRatio (about +131 attack from gear with double minuet and red curry on merit mobs, which is possible.) increasing to about 12% at 2.0cRatio, then diminishing to a %0 reduction at 2.4 cRatio.

if all of that is true, then:

They're still just as good. They saw lower numbers and the first thing they did was scream their heads off that they got nerfed, but in actuality it isn't so. Pre-patch your capped PDIF could vary between 1.6 and 2.4 (average: 2) and now it's always 2. Same thing, except you can't waste crit damage anymore. That's an improvement.

this not true, by exactly 12%.

how crits figure into this would vary heavily by player. assuming a 12% crit rate, you are losing (on average) .2pdif per hit and gaining (on average) .2pdif on a crit.

-.2 * .88 + .2 * .12 = average loss of .15pdif per hit at exactly 2.0 cRatio.

however! We haven't yet proven if the weighting of the 'new' func_pdif(cRatio) is normal and mean/median = 50% or the manner that the clamp adjusts the value is linear however; so we don't know if this is true or not.

Don't know about that. In fact, it doesn't even make any sense since the damage formula never took level into account (it modifies your Attack/Defense ratio from which PDIF is calculated, yeah, but you could obtain the same results by adding more Attack.)

I would argue that this means that damage formula *does* take level into account (and in particular, what BRP meant when he said it).

We'd need to test with a low level character (say, Lv.20 or so WAR) getting as many Attack buffs as possible (Minuets, Chaos Roll, Meat, Berserk) to overkill Attack on a low T mob to confirm what you're saying. There's a possibility that the 1.6 to 2.0 thing is just their Attack being uncapped.

right, but what used to be capped (2.0 cRatio) with a variance of 1.6 -> 2.4 and a mean and median of approximately 2.0, is now 'uncapped' (by 0.4) with a pdif of 1.6 to 2.0 (unknown mean and median) and it now takes a 2.4 cRatio (so 2.8->3.0 raw ratio against exp level mobs) to cap pdif at the 2.0 clamp.

[Amele]

Re: Current Known Issues - AUG 30

Did the variance ever go down to -/+ 0.0 against anything pre-patch? (And I mean other than when a player attacked a mob 40 levels above them.)

no, but the spread was different depending on which part of the function you were in.

in the particular range you are hypothesizing about, the old func_pdif(cRatio) does indeed vary by +/-0.4.

Old pseudo code:
1. calculate player attack/mob defense ratio
2. cap at 2.0.
3. add in variance
4. add 1.0 for critical hits
5. cap at 3.0 if necessary
Botched new pseudo code:
1. calculate player attack/mob defense ratio
2. add in variance
3. if 1 handed, cap at 2.0
4. add 1.0 for critical hits
5. cap at 3.0 if necessary (for 2 handed weapons)

that isn't terribly close to how it really works but that's not bad.
this is closer:
1. calculate cRatio = att/def + 0.05 * (player_level - mob_level)
2. if cRatio > 2.0 then cRatio = 2.0
3. call func_pdif(cRatio) with return pdif.
4. if crit then pdif = pdif + 1.0
5. if pdif > 3.0 then pdif = 3.0

today it -appears- works like:
1 calculate cRatio = att/def + 0.05 * (player_level - mob_level) <- maybe changed?
2 call func_pdif(cRatio) with return pdif.
3. if 2handed and pdif > 3.0 then pdif = 3.0
4. if 1 handed and pdif > 2.0 then pdif = 2.0
5. if crit then pdif = pdif + 1.0
6. if pdif > 3.0 then pdif = 3.0

obviously, there's something SE doesn't like about how it works now (the damage cap is not correcting properly) that could be any of: step 1. (mob level differences may be calculating wrong), step 3. step 4. (which may be the same step in the code) or step 6.

I suspect it's step 1, 3+4 - since I would think SE would've worded it differently if it were specifically about critical hits.

If variance in damage exists against EP mobs, why doesn't it exist against TW mobs? At what point does that variance become 0? Is it just for H2H that variance has been reduced to 0? Is it just against TW mobs? If so, what does the possible pDif range against EP+ mobs look like?

variance should go to zero as cRatio approaches 2.4 (which I've exceeded on EP mobs before.) if you're asking in general? the pDif range at 1.667<=cRatio<=2.0 appears to look like:

pdif_min : cRatio - 0.4
pdif_max : 2.0

this is all 1handed jobs. (h2h works like dual wield).

There are too many questions that are unanswered to simply say "MNKs will still deal the same average damage." To me, it sounds like you've already reached the conclusion that MNK damage has not been "nerfed", and are looking for ways to support that conclusion with how pDif might currently be working.
I'm trying to understand how pDif is working first. Once I've gotten my head around that, then I'll conclude whether MNK/1-handed damage has been nerfed or MNK/1-handed damage is still the same on average as it was before.

it's normal until you exceed 1.667 cRatio, at which point it is (probably) slightly reduced until you reach 2.4 cRatio.

[Armando]

Re: Current Known Issues - AUG 30

Did the variance ever go down to -/+ 0.0 against anything pre-patch? (And I mean other than when a player attacked a mob 40 levels above them.)

No, but it does for Ranged Attacks at certain points. And the variance could be anything S-E feels like making it (there were tiers in the old PDIF system in which max PDIF would stay at 1.0 for a while until you got to a certain point, and only your min PDIF would keep going up, for example.)

I'm getting 2.0 to 2.8 as a hypothesis as to how SE might have changed, even mistakenly, pDif calculations. Given the current list of known issues and how 2 handers seem to be hitting for 3.0 pDif on non-crits and crits alike, it's not a stretch to guess at how SE might have mangled the code.

Old pseudo code:
1. calculate player attack/mob defense ratio
2. cap at 2.0.
3. add in variance
4. add 1.0 for critical hits
5. cap at 3.0 if necessary

Botched new pseudo code:
1. calculate player attack/mob defense ratio
2. add in variance
3. if 1 handed, cap at 2.0
4. add 1.0 for critical hits
5. cap at 3.0 if necessary (for 2 handed weapons)

I think you're looking at it wrong. PDIF is independent from cRatio; it doesn't revolve around it. The game just runs a bunch of IF statements:

IF a < cRatio < b, min PDIF = this1 and max PDIF = that1
IF b < cRatio < c, min PDIF = this2 and max PDIF = that2

And so on. Those "this" and "that" can be anything, from a constant to a function. And those functions don't even have to be functions of cRatio. For example, it could've been made so over a certain cRatio range, your max PDIF would be a function of your INT and your min PDIF would be a function of your MND. Then you'd have two players with the same Attack having different PDIF ranges on the same mob because of their different INT and MND scores.

But that's neither here nor there. I guess what I want to get at is this: Assuming the one-hander change was deliberate, why would S-E give us a PDIF range that's entirely beyond 2.0, and then cap every randomly generated number at 2.0, when they could've just added fixed values for min and max PDIF when your cRatio is 2 in the first place?

If variance in damage exists against EP mobs, why doesn't it exist against TW mobs? At what point does that variance become 0? Is it just for H2H that variance has been reduced to 0? Is it just against TW mobs? If so, what does the possible pDif range against EP+ mobs look like?

There are too many questions that are unanswered to simply say "MNKs will still deal the same average damage." To me, it sounds like you've already reached the conclusion that MNK damage has not been "nerfed", and are looking for ways to support that conclusion with how pDif might currently be working.

I'm trying to understand how pDif is working first. Once I've gotten my head around that, then I'll conclude whether MNK/1-handed damage has been nerfed or MNK/1-handed damage is still the same on average as it was before.

It's not quite so much that I've pre-convinced myself that they weren't nerfed. It's just that I can't quite wrap my head around some of the hypotheses floating around - they just don't make much sense. And yes, the PDIF change is 2.0 for all weapons, not H2H. Torzak was doing some testing on Asuran Fists and was getting the same numbers over and over again because all 8 hits had a fixed PDIF, and Asuran can't crit. I also saw screenshots of a Maneater doing the same damage over and over again.

Why do you have variance on EPs and not on TWs? Easy, your cRatio is capped on TWs and not on EPs, and S-E made it so the PDIF range is fixed at 2 only when your Attack is capped. Capping Attack when soloing anything that gives EXP is near-impossible without buffs. Even if you solo with meat I guarantee you're not going to cap your Attack on EP mobs at Lv.13.

this not true, by exactly 12%.

how crits figure into this would vary heavily by player. assuming a 12% crit rate, you are losing (on average) .2pdif per hit and gaining (on average) .2pdif on a crit.

-.2 * .88 + .2 * .12 = average loss of .15pdif per hit at exactly 2.0 cRatio.

Where are you getting those numbers? I'm not following the reasoning behind them. My reasoning is:

Pre-patch, pDIF varied between 1.6 and 2.4. And although the distribution of all the possible outcomes wasn't perfectly even, your average damage was right in the middle of the average between the min and max possible outcomes. I tested this much myself by recording 168 non-crit hits on TW mobs with capped Attack and STR. My damage varied between 22 and 33. (22 + 33)/2 = 27.5. The average of those 168 hits was 27.494.

So, basically, under the old PDIF ranges, your average non-crit damage would basically correspond to the middle point between your min/max PDIF or your min/max damage. So the difference between the old PDIF ranges and the new ones are moot. Now, crits are a different story. Under the old PDIF ranges, about half of your hits would roll a PDIF higher than 2.0. Because the max was 3, anything over a 2 would lose part of that +1 PDIF (e.g. rolling a 2.3 SHOULD give a PDIF of 3.3, but it's capped at 3.0, so you only gained +0.7.) Now, you're always getting the full +1 PDIF.

Where's the damage lost?

I would argue that this means that damage formula *does* take level into account (and in particular, what BRP meant when he said it).

You argue it based on what?

right, but what used to be capped (2.0 cRatio) with a variance of 1.6 -> 2.4 and a mean and median of approximately 2.0, is now 'uncapped' (by 0.4) with a pdif of 1.6 to 2.0 (unknown mean and median) and it now takes a 2.4 cRatio (so 2.8->3.0 raw ratio against exp level mobs) to cap pdif at the 2.0 clamp.

Until it's proven that there's different formulas being applied for mobs that are TW to EM and mobs that are T to IT, you can't assume those numbers.

[hongman]

Re: Current Known Issues - AUG 30

Reading all of this reminds me of being back at school.

[Mhurron]

Re: Current Known Issues - AUG 30

Reading all of this reminds me of being back at school.

Except people care about this math.

[hongman]

Re: Current Known Issues - AUG 30

Who said I was complaining? It was just a freaking comment.

[Feba]

Re: Current Known Issues - AUG 30

Really, my math teacher tried to get people into math by explaining how they could use it in real life. Their examples? Something about calculating the area of a triangle made it easier to take care of their lawn or something. Yeah, that's math that makes me give a shit.

[Callisto]

Re: Current Known Issues - AUG 30

Except people care about this math.

Lock the thread, Mhurron wins.

You sir, are awarded one(1) intarwebz.

[Lmnop]

Re: Current Known Issues - AUG 30

Yeah, that's math that makes me give a shit.

You=Hero.

Lock the thread

ARE YOU KIDDING?!? Amele and Armando disagreeing and both heavy-handed with the dirty math? No way I'm missing this.

[LyonheartLakshmi]

Re: Current Known Issues - AUG 30

that isn't terribly close to how it really works but that's not bad.
this is closer:
1. calculate cRatio = att/def + 0.05 * (player_level - mob_level)
2. if cRatio > 2.0 then cRatio = 2.0
3. call func_pdif(cRatio) with return pdif.
4. if crit then pdif = pdif + 1.0
5. if pdif > 3.0 then pdif = 3.0

today it -appears- works like:
1 calculate cRatio = att/def + 0.05 * (player_level - mob_level) <- maybe changed?
2 call func_pdif(cRatio) with return pdif.
3. if 2handed and pdif > 3.0 then pdif = 3.0
4. if 1 handed and pdif > 2.0 then pdif = 2.0
5. if crit then pdif = pdif + 1.0
6. if pdif > 3.0 then pdif = 3.0

Thanks for the clarification. I basically just trying to point out how steps #2 and #3 in the old calculation seem to have gotten reversed.

Before, if your cRatio was exactly 2.0, you could expect the variance introduced by func_pdif to average out, so that overall you were doing the same damage as you would with no variance. If the new damage calculations are indeed acting the way you described, values which rise above 2.0 due to func_pdif in that situation would get capped down, while values which went below 2.0 due to func_pdif would be unchanged. This would result in lower average damage (since there would be no 2.2s and 2.3s to balance out the 1.8s and 1.9s, etc).

However, if a player has enough attack to be at a cRatio of 2.4 (I've also ready 2.333 as the target cRatio elsewhere), then you'll have high enough values returned by func_pdif that you'll always get 2.0. Under this condition, you would actually see a benefit as Armando has mentioned, since you would always get a full +1.0 from crits without losing anything to variance.

[Amele]

Re: Current Known Issues - AUG 30

aside, I hate how I can't quote train here, it makes things harder to do in context.

No, but it does for Ranged Attacks at certain points. And the variance could be anything S-E feels like making it (there were tiers in the old PDIF system in which max PDIF would stay at 1.0 for a while until you got to a certain point, and only your min PDIF would keep going up, for example.)

all indications so far show that the actual formulas didn't change, just how far it "slides" for various jobs.

I think you're looking at it wrong. PDIF is independent from cRatio; it doesn't revolve around it. The game just runs a bunch of IF statements:
IF a < cRatio < b, min PDIF = this1 and max PDIF = that1
IF b < cRatio < c, min PDIF = this2 and max PDIF = that2
And so on. Those "this" and "that" can be anything, from a constant to a function. And those functions don't even have to be functions of cRatio.

actually, except in cases when the pdif_min or max was constant the pdif range was entirely dependent on cRatio (being a function of cRatio) although you're right, SE didn't have to code it that way.

Assuming the one-hander change was deliberate, why would S-E give us a PDIF range that's entirely beyond 2.0, and then cap every randomly generated number at 2.0, when they could've just added fixed values for min and max PDIF when your cRatio is 2 in the first place?It's not quite so much that I've pre-convinced myself that they weren't nerfed. It's just that I can't quite wrap my head around some of the hypotheses floating around - they just don't make much sense.

well, apparently mob pdifs are messed up too, so it's possible that all of this is just some large convoluted bug. that aside; the reason people are hypothesizing 2.4(well, 2.333) cRatio being the new cap (besides 1.2*cRatio - 0.8 = 2.0) is that the formula range does not appear to have been changed.

that is, all testing so far (not that's it's been conclusive) has shown that the pdif max and min are the same for a given cRatio as they were pre-patch, except for the apparent 'anomaly' above 1.667 pdif on 1h weapons. - the easiest answer is a simple clamp of func_pdif(cRatio) to 2.0 which is what has people crying foul.

Where are you getting those numbers? I'm not following the reasoning behind them. My reasoning is:
Pre-patch, pDIF varied between 1.6 and 2.4. And although the distribution of all the possible outcomes wasn't perfectly even, your average damage was right in the middle of the average between the min and max possible outcomes. I tested this much myself by recording 168 non-crit hits on TW mobs with capped Attack and STR. My damage varied between 22 and 33. (22 + 33)/2 = 27.5. The average of those 168 hits was 27.494.

ok, so we agree that the old system had a mean of ~(max+min)/2. that makes the rest of this alot simpler. lol.

So, basically, under the old PDIF ranges, your average non-crit damage would basically correspond to the middle point between your min/max PDIF or your min/max damage. So the difference between the old PDIF ranges and the new ones are moot. Now, crits are a different story. Under the old PDIF ranges, about half of your hits would roll a PDIF higher than 2.0. Because the max was 3, anything over a 2 would lose part of that +1 PDIF (e.g. rolling a 2.3 SHOULD give a PDIF of 3.3, but it's capped at 3.0, so you only gained +0.7.) Now, you're always getting the full +1 PDIF.

right. you always get the full +1 pdif from a crit.

you're losing damage in the range where the old function would be >2.0 max and <2.0 min (cRatio > 1.667 and less than 2.333) because at 1.8 cRatio for instance, what used to be: 2.16 max and 1.36 min, is now 2.00 max and 1.36 min.

assuming that the distribution is still averaged at 2.16+1.36/2 = 1.76, you are losing a small amount of damage because the bell curve still tails at the low end, but is forcible cut at the upper end. (sorry I can't make a diagram).

in this particular case, it drops the mean to ~1.746 (assuming normal distribution) if you assume linear distribution you lose even more. (mean drops to: 1.743)

at the point where cRatio = 2.0 (pdif 1.6~2.4) and you cap, this would drop the mean from 2.0 to 1.89, which is (my mistake) ~5.5% lost dot.(assuming linear distribution) not 12%.

you also go from .89 to 1.0 average pdif gain from a crit (which you mentioned) which is ~12% gain. but you do not generally speaking crit 33% of the time. (which is the number of times you'd need to crit to make up for the dot lost from regular hits) in the event that you do, then this would be an equivalent change.

Where's the damage lost?

lost damage because you cap at 2.0 pdif, whereas before you cap at 2.4. gained damage from crits always being 1.0 is insufficient to make up for this loss in the area of ~2.0 cRatio.

above a cRatio of 2.333 there is gained damage. in range 1.667 to slightly less than 2.33 (the point where your crit rate makes up for the lost pdif) there is lost damage.

You argue it based on what?

moblevel is directly involved in cRatio calculation (being: attack/defense - 0.05*(mLVL-pLVL) ) and since pdif_max and pdif_min (and by extension pdif_average) are directly dependent on cRatio - the mob's level therefore has an effect on pdif. because pdif has a direct bearing on the damage formula; by extension so does cRatio, and therefore mLVL. Q.E.D.

Until it's proven that there's different formulas being applied for mobs that are TW to EM and mobs that are T to IT, you can't assume those numbers.

all of my numbers specifically assume that there *is not* a fundamental difference between TW and EM and T and etc. mobs. although they do change the formula in understood ways. (mobs with equal stats but higher level will take less damage than mobs of those same stats but lower level).

edit:

However, if a player has enough attack to be at a cRatio of 2.4 (I've also ready 2.333 as the target cRatio elsewhere), then you'll have high enough values returned by func_pdif that you'll always get 2.0.

yeah. 2.333 is right assuming the cRatio to pdif min and max calculation has not changed (there is no indication it has) 2.4 cRatio is easier to type / remember and is 'close enough'.

[Callisto]

Re: Current Known Issues - AUG 30

Gawd I'm like right at the cusp of understanding what's happening, and then barely not understanding it, and not sure where I'm missing a gap lol. I have to ask you guys for help w/ another topic later.

I need to read this topic over like two or three times more, I think I might print the thread and take it with for the weekend road trip lol.

[Yellow Mage]

Re: Current Known Issues - AUG 30

Part of me wishes I knew what the heck pDIF was, but the other part is living by the maxim "ignorance is bliss."

What depresses me is math is supposed to be my forte, and I can hardly understand what's going on here.