Exp STILL NERFED after September 6 patch

[Ratt]

I don't see how we have enough data one way or the other?

I've updated the exp chart (http://seq.sf.net/exp.php) and will keep it updated as data comes in...

Data should be collected with a 10 Race, 10 Class character if at all possible.

[SwedishChef]

Background:

They said they added bonuses and penalties. Due to the nature of how previous exp changes have been made, my first idea is that they added a multiplier to the old exp... simple, aye?

Well, when I look at the data in the posts by fee, it seems more like a combination of several multipliers, which change differently based on mob level.

It gets easier to understand the current data if we start with analyzing the 4/9 test, since the last version of exp return is similar for high levels.

Mob  Pre-4/9 value    4/9 test value   Mul1  Mul2  Mul3
47   198810           69583            0.5   0.7   1
48   207360           82944            0.5   0.8   1
49   216090           97240            0.5   0.9   1
50   225000           112500           0.5   1.0   1
51   234090           128749           0.5   1.1   1
52   243360           146016           0.5   1.2   1
53   252810           164326           0.5   1.3   1
54   262440           238820           0.5   1.4   1.3
55   272250           265443           0.5   1.5   1.3
56   282240           293529           0.5   1.6   1.3
57   292410           ?                0.5   1.7   1.3
58   302760           354228           0.5   1.8   1.3

You get from old exp value to fee's test value by multiplying the old exp value with all the three different multipliers in turn.

Theory:
The first multiplier is based on you being level 60, and was set at 0.5 to make level 60 seem twice as long as before (All non-hell levels would probably have a multiplier less than 1, which would be compensated by a higher than 1 multiplier for previous hell levels. How far from 1 would be dependent on how abnormal the level used to be. 60 got a very low multiplier because it used to be very short.).

The second multiplier is based on mob level, either in relation to your own, or just mob level straight. Further testing with different levels of testing person would have been needed to see the difference. It goes from 0 for a level 40 mob (which didn't use to give exp before patch, right?) to 2.0 for a level 60 mob (outside the tested window is all speculation of course).

The third multiplier was 1.3 for all mobs level 54 and higher. This looks like it was added to compensate for mobs gaining in strength in a non-linear fashion. This lead to a small jump in exp return at level 54, but nothing really bad.


OK, onwards to the 6/9 patch.

Mob  Pre-4/9 value    6/9 test value   Mul1  Mul2  Mul3
47   198810           149107           0.75  1     1
48   207360           155520           0.75  1     1
49   216090           162067           0.75  1     1
50   225000           168750           0.75  1     1
51   234090           175567           0.75  1     1
52   243360           182520           0.75  1     1
53   252810           189607           0.75  1     1
54   262440           196830           0.75  1     1
55   272250           398165           0.75  1.5   1.3
56   282240           440293           0.75  1.6   1.3

Theory:
Now, multiplier 1 was set to 0.75, this to make level 60 exp more like it used to be. This was mentioned in the patch message.

Multiplier 2 was set back to 1 (no effect) for all mobs under level 55, which caused a de-nerf of exp return for mobs under 50, and a removal of the recently added bonus for mobs level 51-54. The de-nerf was mentioned in the patch message. The bonus for mob levels 51-54 had hardly been there long enough to mention removal of I guess... and it had to go or leveling would become too fast.

Multiplier 3 was also set back to 1 for all mobs under level 55... this was an odd change, since it only impacted level 54 mobs. It just moved the start of the 1.3 multiplier bonus up a bit in levels.

The 1.3 modifier might be what is mentioned in the patch message as group bonus for fighting mobs close to your level. Normally, a solo character would not see this very much. It is also possible that some additional multiplier was set based on your group status. Any conclusions about this fall due to lack of info.



Summary of my theories... we now get pre-4/9 exp for mobs up to level 54, and substantially higher exp for 55 and up mobs. It is unknown whether any, all or some of these multipliers apply to AA exp gained.



I don't plan on testing this myself, I prefer theory over practice . If somebody feels like checking my theories, add the multipliers to an Excel or similar chart, extend them to lower or higher mob level ranges, and test if they fit experimental data. It should be noted it is not unlikely the multipliers might change pattern, especially multiplier 3 (the 1.3 one) might change for higher mobs, we have very little info on those.

You could also test with a non-60 character, to see what their hell level removal modifier (multiplier number 1) is. Once you have that, you can test some different mobs and form a hypothesis about what the other two multipliers are for them.


PS: btw Ratt, on what ZEM did you base your chart for exp return based on mob level vs player level? It would be much more helpful knowing that.

EDIT: Fixed table alignment error.

[link129]

I was seeing the same amount of exp on mobs lvl 43-50. 115865 exp points per kill. Killing a lvl 54 mob it jumped to 231730 exp points per kill. Just some info, may be usefull.

[Cryonic]

Link:

That is because all the client is told while you are in a zone is when to update the exp bar. It is NOT told the exact amount of exp you gained killing that mob, so the amount SEQ was showing you was 1/330th the amount of exp needed to clear your current level.

to get the TRUE amount of exp killing a mob you now have to zone in, record your current TOTAL exp which is sent to you everytime you zone, kill something, then zone again and record the new amount.

Subtract A from B and you get what you ACTUALLY got for killing that mob.

P.S. Notice that 231730 is DOUBLE 115865.

[Ratt]

Originally posted by SwedishChef

PS: btw Ratt, on what ZEM did you base your chart for exp return based on mob level vs player level? It would be much more helpful knowing that.

Sorry... meant to mention those two tests were conducted in Maidens Eye... not sure what the ZEM is for that zone.

Has anyone found the new octet offset for AA experience on zoning? That would help a bit... everything I'm testing with is regular experience right now. I think AA has been munged with as well, so the modifiers are even more different yet.

[T.C. Jaguar]

I was playing around with some numbers just to see what the effects of L51+ smoothing might look like. Just to reiterate: these numbers are hypothetical and just for fun.

I made a linear progressive "curve", with the assumptions that L51 cannot be twice as hard (or harder) as L50, and that L60 does not have to adhere to the curve.

I am reasonably sure about the L60 multiplier based on Fee's empirical evidence.

	Previous XP needed	Multiplier	New XP needed	Difference
50	10,291,400*		n/a		10,291,400*	n/a
51	23,976,500		0.71		17,023,315	-6,953,185
52	25,996,300		0.91		23,656,633	-2,339,667
53	28,118,100		1.12		31,492,272	 3,374,172
54	46,090,700		0.83		38,255,281	-7,835,419
55	50,205,900		0.92		46,189,428	-4,016,472
56	54,529,300		1.00		54,529,300	         0
57	59,065,700		1.05		62,018,985	 2,953,285
58	63,819,900		1.11		70,840,089	 7,020,189
59	89,334,600		0.89		79,507,794	-9,826,806
60	53,463,000		1.33		71,105,790	17,642,790
	------------------------------------------------------------------
								    18,887

*Not sure if L50 has changed since the pre-51 smoothing several months ago.

I will be curious to see how close (or far) I am from the actual multipliers.

[DeltaHuey]

Quote:

1. Level 60 (10/10 race/class) Regular Exp for level 56 Mob - 415,808
2. Level 56 (10/10 race/class) Regular Exp for a level 56 Mob - 495,720

Interpreting this using the theory Baelin is saying (which I 100% beleive HAS to be the right way) shows the following.

assume level 60 is 75% nerf to make the level last longer and that your level in relation to the mob doesn't matter, it is just a straight bonus on mob level.

415,808/0.75 = 554,410 xp .... apply this to the 56 xp

495,720/X = 554,410 xp .... you get X = 89.5%

Now, level 56 would of had the experience requirements raised up to help smooth level 59, so an 89.5% penalty on mobs makes sense.

Why is this so hard to beleive? Verant awhile ago made levels 1-10 faster. They did it by making mobs give more xp based on your level between 1 and 10. Level 1 has always had a level bonus. It makes sense that leaving the actual xp numeral the same, and only extending what they do in 1 through 10 to levels 51 through 60 is by far the easiest thing to do.

Also, I beleive 55+ mobs give a straight bonus based on THEIR level, having absolutely nothing to do with YOUR level. How do you think Verant would add a level dependant experiene modifier for a mob when you are dealing with groups.

The easiest explanation is usually the right one.

I'm going to present an easy way to prove if this is right or not, and not actually do it because I don't have SEQ =).

Have a level 51-60 kill a level 45 mob or something. We know what the experience should be on a level 45 mob using the old xp formula. Take what they now get from that level 45 mob, and you will get multipliers based on each level. Use those multipliers on each levels experience requirements and you get new experience requirements. Add the new ones up and check the total required to the old total required. If they match, this theory is pretty water tight.

The only thing that sucks to me is that levels 50 through 54 got shit.

-=Wedge=-

[DeltaHuey]

****edit****

I'm a dumbass. Was looking at Both Ratt's numbers and Fee's numbers, when they got two different numbers for a level 60 on a level 56 mob. Obviously different ZEM's, which fucked me up.

This is a good thing though, because now we have 2 sets of numbers and can verify any patterns that hold true.

****starting over****

Fee has 156% XP gain for a level 60 on a level 56 mob with new XP changes.

Using same assumptions (an xp modifier was added based on YOUR level to smooth things out, an xp modifier was added to MOBS 55 and higher to add a bonus and that the bonus has nothing to do with YOUR level).

75% * X = 156% ... X = 208% ---- FEE's numbers.

So, level 56 mobs give a 208% bonus to XP.

RATT's numbers:

1. Level 60 (10/10 race/class) Regular Exp for level 56 Mob - 415,808
2. Level 56 (10/10 race/class) Regular Exp for a level 56 Mob - 495,720

See if using both their numbers can make any sense.

BASEXP = old xp formula with correct ZEM, etc.

BASEXP * 0.75 * 2.08 = 415,808 at 60
BASEXP * 0.895 * 2.08 = 495,720 at 56

BASEXP * 1.56 = 415,808 at 60
BASEXP * 1.8616 = 495,720 at 56

BASEXP = 266543.58 at 60
BASEXP = 266287.06 at 56

Obviously I'm seeing some rounding errors, but these are close enough where I feel confident about them. If the zone these are from has a ZEM of 85, you get 266,560... not shabby.

I think this proves that using both Fee's and Ratt's numbers, I prove my assumptions (really Baelin's with some of my own added) are correct.

There is now a modifier based on YOUR LEVEL to smooth things out (75% for 60, 89.5 for 56).

There is now a modifier for MOBS 55 and higher which does not depend on your level (208% for 56).

With more numbers, this can be proven to hold true, and we can fill out a table with all the new XP values.

-=Wedge=-

[Junu Peeth]

To DeltaHuey:

In your first post you performed the following operations and obtained these results. Note that I substituted figures with pure algebra so that you can see the final correlation without any association by numbers.

Assuming that:
415,808/0.75 = 554,410
a/y=c

And we know:
495,720/X = 554,410
b/x=c

Then Solve for X:
x=b/c

where:
c=a/y

Therefore:
x=b*y/a


================================================== =
Now you assumed the calculations from your previous post were correct and continued on:

BASEXP * 0.75 * 2.08 = 415,808
z*y*d=a

BASEXP * 0.895 * 2.08 = 495,720
z*x*d=b

BASEXP * 1.56 = 415,808
z*y*d=a

BASEXP * 1.8616 = 495,720
z*x*d=b


BASEXP = 266543.58
z=a/(y*d) (**1)

BASEXP = 266287.06
z=b/(x*d) (**2)

This was where you ended your conclusions but you could (and should) have gone further to its final conslucion.

Since we are already working on the calculation made in your first post (which were based off your intial assumptions) which gave us "x":
x=b*y/a

Substituting this into (**2):
z=b/([b*y/a]*d)
z=1/([y/a]*d)
z=a/([y]*d)
z=a/(y*d)

So you have effectively proved that:
1=1

Or also demonstrated:
1/4=0.25
0.25*4=1

It's a common mistake but nevertheless your two posts prove nothing because they are a loop. They support each other ONLY if you assume that the other calcualtion is correct.
IE. Assuming A=B then B=A, because B=A then A=B. Conclusion: Gimme phat Lewtz.

[Junu Peeth]

In Fee's 2nd post in this thread he mentioned a formula:

Prior to the patch AAXP was gained using the following formula:
code:--------------------------------------------------------------------------------
(moblvl^2) * ZEM * classmod) = A
(moblvl^2) * ZEM = B

B - (A - B) = aaxp reward
--------------------------------------------------------------------------------

This can be simplified. It does not change the formula although it kept glaring at me wheneve I looked at it.

B - (A - B) = aaxp reward
aaxp reward = {(moblvl^2) * ZEM} - {[(moblvl^2) * ZEM * classmod] - [(moblvl^2) * ZEM]}
aaxp reward = [(moblvl^2) * ZEM] * {1 - (classmod - 1)}

aaxp reward = [(moblvl^2) * ZEM] * (2 - classmod)

EDIT: I just want to say also that SwedishChef's post has been the most concise and correct statement on what has been demonstrated so far. Your usage of 3 modifiers was excellent.

[eq_freak]

I noticed Fee updated the charProfileStruct in everquest.h today to include the current altexp offset. So hopefully we can get some new numbers soon

[DeltaHuey]

I don't think it's loop, let me look at it another way, and you let me know.

2 equations, 2 unknowns ....

Here is what I start out with.

BASEXP * X * L = 415,808
BASEXP * Y * L = 495,720

BASEXP is known (level^2*ZEM*mod): 266,560.
X is known from Fee's data: 0.75

L is the MOB level modifier, right now unknown
Y is the PLAYER level 56 modifier, right now unknown

So There you go, two equations, two unknowns.

415,808/(X*L) = 495,720/(Y*L)

multiply both sides by L

415,808/X = 495,720/Y

415,808 * Y = 495,720 * X
Y = (495,720 * X ) / 415,808
X = 0.75...

Y = 0.894

Okay, so now Y and X are both known, pick one of the equations...

BASEXP * Y * L = 495,720
266,560 * 0.894 * L = 495,720

238304.64

L = 2.08

Of course, I can show a loop too, just, like you say, carry it one step further:

415,808/(X*L) = 495,720/(Y*L)
415,808/(0.75*L) = 495,720/(0.894*L)
554,410/L = 554,410/L

FUCK ME, 1=1

This time I didn't even use Fee's numbers but I get the same 56 MOB mod (208%). I had to use Fee's stuff the first time, because I didn't know the BASEXP, but with a ZEM of 85 I do know it.

Let me know if this is still flawed, but I think using my assumptions to setup the equation as:

BASEXP * LEVELMOD * MOBMOD = NEWXP

I get numbers that fit both sets of DATA out there. Now, with only two sets, it needs more to be proved.

-=Wedge=-

[Junu Peeth]

I didn't really make my point cleraly in my first post DeltaHuey and I apologise for it. I will elaborate further.

In your first post you proved that X = 0.894 (to save confusion it is noted that you use X in your first post and use Y to last post to define the same number). Where X is the multiplier a level 56 player receives for killing a lvl 56 mob. This was obtained by knowing:
(1)the XP received for the mob from a lvl 56 char;
(2)the XP received for the mob from a lvl 60 char;
(3)the multiplier a level 60 player receives for killing a lvl 56 mob is 0.75.

These facts I will not dispute. Nor will I dispute that the multiplier for a lvl 56 mob being killed by a lvl 56 player is 0.894.

Therefore I agree with your first post up to this point in your workings.

What I disagree with at this point in time but forgot to even mention in my post (it was 3am when I posted) was your assumtion that:

Also, I beleive 55+ mobs give a straight bonus based on THEIR level, having absolutely nothing to do with YOUR level.

Now you may believe this but it has not been proven. It may well be the case but we have at this time only one point of data available so no comparison nor conclusion can be made, only possilbilities can be discussed. Your belief is indeed one of those possibilities and is therefore an assumtion in your calculations. You can not use an assumption to prove itself.

How do you think Verant would add a level dependant experiene modifier for a mob when you are dealing with groups.

I think (based on the data set we have available for a lvl 60 player) that they could have used the formula set forth by SwedishChef in his forst table and adapted it for a lvl 56. For a lvl 60 we see that the 2nd multiplier could be represented by:
Mul2 = 2 - [(playerlevel - moblevel)/10]
which would make Mul2 dependant upon the player's level or Mul2 could be represented by:
Mul2 = (moblevel - 40)/10
which would make Mul2 independant of the player's level.

Both of these formulaes are simple and both are, at this time, valid possiblities for the description of Mul2. There are of course an infitie number of possible formulae but in my mind these two should be explored first as they are the most likely.

Getting back to your assumption that Mul2 is independant of player level. Your whole second post starts with that assumption and then goes on to prove that assumption based upon itself.

BASEXP * 0.75 * 2.08 = 415,808 at 60
BASEXP * 0.895 * 2.08 = 495,720 at 56

Here we know that 0.75 is correct but you assume that because the 2.08 (ie a 108% gain) is the correct figure for a lvl 60 killing a lvl 56 mob then 2.08 is again the correct figure for a lvl 56 killing a lvl 56 mob.

You show the same thing in your final post.

BASEXP * X * L = 415,808
BASEXP * Y * L = 495,720

You inherently assume that L = L when this has not yet been proven and they should be tagged as L60 and L56 (I wish I cold find a way to subscript 60 and 56). If it is proven that the modifier for a lvl 60 player and lvl 56 player killing a lvl 56 mob is the same then L60 = L56 and the constant L (L = 208) can indeed be used for the killing of a lvl 56 mob by any player.

If you were to recalculate your final post but substitute L for L60 and L56 where appropriate you would find that Y (as represented in your final post) is not solvable. It ends up as:
415,808/(X*L60) = 495,720/(Y*L56)
Y*L56 = 495,720/[415,808/(X*L60)]
Y*L56 = X*L60*495,720/415,808
Y*L56 = X*L60*1.19218
we know: L60 = 2.08 and X = 0.75
Y*L56 = 0.75*2.08*1.19218
Y*L56 = 1.8598

IF at a later time L56 is shown to be the same as L60 (=2.08) then Y does indeed calculate as 0.89414 or 89.4%. Personally I hope it doesn't as this means there is no bonus given for a lvl 56 killing an even con mob versus a lvl 60 killing a mob that it 4 level below him. I hope that VI introduced bonuses to mobs based upon the difference in levels between the killer and killee.

And before some asks again: "how could this theory apply to group XP being split being players of different levels".

Each mob is equivalent to an amount of XP calculated by: level^2*ZEM*classmod (what DeltaHuey terms BASEXP)

BASEXP is then split between the group using the formula:
Player's_split = PLayer's_Level / Sum_of_group's_level

So far nothing is different and this is where group XP goes it separate ways and is then calculated on an indivual basis.

Multiply this by the hell level modifier for the player receiving the XP (shown as Mul1 in SwedishChef's table) and then by Mul2 and Mul3 as shown in SwedishChef's table. Voila, individual player XP taken from group XP and adjusted for hell level removal and for a risk_vs_reward scale.

Again DeltaHuey, my apologies for correctly singling out the factor of your arguement that I disagreed with in the first place. I do not agree with your assumption that xp given by mobs is the same for all players until this is proven to be the case. In your second post you did assume this to be true and did input this assumption into the two formulas which proved nothing more than: If L = L then Y = 0.894.

My conclusion is that until more XPdata is gathered by players under level 60 in zones whose ZEMs are well known then the only thing we can say is what works for lvl 60 only. And even then we can not give an accurate description of mobs of level less than 47 or greater than 56. As SwedishChef has shown in his work there appears to be a pattern involving three multipliers the last of which is equal to 1 at levels 47 to 54 and changes to 1.3 at level 56 and 57 (and likely 58 if the 4/9 modifiers have not been changed to much), but what happens at levels greater than these we can only extrapolate.

Also unknown at this time is if Mul2 is dependant on player level and Mul3 is not OR Mul3 is dependant on player level and Mul2 is not OR both Mul2 and Mul3 are dependant on player level OR neither Mul2 nor Mul3 are dependant on player level.

I would suggest everyone refer to SwedishChef's post as the most concise and accurate summary of what is known at this time. I do. (Because I babble too much)

EDIT: DeltaHuey, I do agree that the formula you postulated:
BASEXP * LEVELMOD * MOBMOD = NEWXP
is the most likely one with possibly these clarifications/modifications:
* LEVELMOD is determined only by the level of the player;(assumed in your theory but I want to state it here)
* MOBMOD may or may not be determined by the level of the player; (in your theory you assume it to be solely independant of player level)
* MOBMOD is very likely to be made of two parts as described in SwedishChef's post. (you have made no mention of this possibility either, the use of two modifiers though does describe a logical step in the data we have seen gathered so far)

[DeltaHuey]

I see what you are saying.

X*L
Y*L

I assume L is a constant, so X and Y can be figured out that way.

If you make L dependant on level, then bam, 3 unknowns, 2 equations.

I did make it perfectly clear that I made these assumptions and that more data was needed to see if things fit. That is how you do it, you make a model and your assumptions and see if the data fits. The data does fit, but it is very little data.

I was a bit overzealous in posting that I was right.

So, in my opinion, I found a theory that works with both sets of data, but that more data is needed to prove it. And it will be proven =)

-=Wedge=-

ps. The easiest way to see which is right is to have a level 56 kill a level 54 mob. In my theory, 56 would get 89.4% of BASEXP, in Swedesh's, it would be something else entirely. Of course, the more I think about post 9/4 and pre 9/6 things, I think Swedesh has to be right...

[Junu Peeth]

Originally posted by DeltaHuey
So, in my opinion, I found a theory that works with both sets of data, but that more data is needed to prove it. And it will be proven

I believe you didn't find a theory, you only solved for a variable by assuming that another variable was a constant. I am not taking that away from you as no one had done it previously and I am beng a little pedantic. But you did not prove anything.

You took two equations that contained two variables and then by an assumtion you added another equation that said L = L.

You solved for 2 variables in 3 equations, this is taught in high school maths.

As for whether or not your assumption that L is a constant will be proven or now I do not know obviously. But I do hope that VI coded such that your assumption is incorrect. I would rather receive a larger reward for a larger risk. IE. a lvl 56 taking out an even con vs a lvl 60 taking out a mob 4 level lower than him. Whether this would skew levelling would have to be tested and balanced of course, it might be that because the XP required to level when 56 is so much lower than at 60 that the 56 player receives a good enough reward anyway. Although this logic is a little skewed in itself it is possible VI took this path.