?

Log in

No account? Create an account

Previous Entry | Next Entry

Enmity Testing (Part IV)

Overview

The concept of Enmity has really been blown wide open from the first 3 parts of our testing.  I've started a running Enmity Table of all actions that do not involve damage or curing in an effort to get a complete list of CE and VE values.  I hope that that list can be completed rather quickly - it will only grow after the new jobs come out for WotG in a few weeks.  If you're interested in helping, just find something not listed or with missing numbers and try to run the test yourself and post back results.

In the meantime whlie the list is being filled, I'll continue to go over more concepts regarding Enmity.  In this part, I want to focus on 2 more major parts of Enmity - the nature of Volatile Enmity (which I have largely been avoiding) and the effect of Enmity Gear.  Thanks to some very quick work by groups other than myself and Ashira (we are honestly a bit burnt out by the testing and just feel like playing a bit online this weekend), the nature of both of these were quickly found.  Most of this data was not tested by us - but the description and controls are good nonetheless.

To summarize the goals and questions to be answered in this section...

      -  Is the decay of Volatile Enmity (VE) static?  If so, what is the rate of decay?
      -  What is the Volatile Enmity (VE) of an action? (a test for any spell or ability)
      -  How does Enmity Gear affect Cumulative Enmity?


Special thanks to 2 groups for setting up and conducting these tests - Elipse and Suwiside (Odin) for the Enmity Gear testing and Ryoii and Nomomii (Lakshmi) for the VE testing.  The setup for testing will not be exactly as these 2 groups conducted, but represent methods of reproducing similar results.


Previous: 
Enmity Testing (Part III)

Next:  Enmity Testing (Part V)

Enmity Table

____________________________________________________________________________________________________
____________________________________________________________________________________________________
 
Test  12A

What is the Volatile Enmity (VE) of Provoke

Testing Setup:

- 1 puller character with Silence access (BRD/WHM)
- 2 level 75 members with ability in question (RDM/WAR, BLM/RDM)
- 2 characters must have 0 enmity merits (+/-), be stripped of all gear, and have access to "distance plug-in"
- Use a "level 0" monster - we chose Bumblebee in West Sarutabaruta

- All characters must have either Stoneskin or Phalanx buffed prior to starting to ensure 0 damage
- Utsusemi shadows must be off when this experiment takes place

- Puller pulls with Silence while the 2 test characters move to 0.7 distance (same distance)
- Test character 1 casts Sleep x1, Cure 1 for 0 HP x1, and Dispel x5 (CE = 1781)
- Test character 2 uses Provoke (1 CE) - player 1 has a +1780 CE difference
-
 Document if Player 2 takes hate immediately after Provoke

- Repeat the test but have player 1 gain a +1801 CE difference
-
 Document if Player 2 takes hate immediately after Provoke

Results

CE DifferenceHate?
  +1780 CEYes
  +1801 CENo


Based on these simple results, we can deduce that the VE of Provoke must be between 1780 and 1801.  You can actually ramp the CE up to narrow down this range; however, be aware that your results may start to vary as you approach the true value.  This is because the mob seems to choose targets only every so often while VE seems to decay constantly.  If the VE decays beyond the point of taking hate before the enxt time the mob chooses targets, the player will not get hate off provoke - this causes variations in the results.

Another good question about this test is how would you know what range to look in?  It's not like the range of 1780 to 1800 VE is intuitive.  There are better ways to tell from a bigger range, but for the purposes of this test, they are irrelevent since our concern is to just nail down the VE of Provoke for the latter two tests.  I will go over how to conduct a VE search on an action later on.

I think if you run this, it will be come more and more likely that the VE of Provoke is 1800 - it's the nicest number in the range we applied, everything else we've seen so far has been in an increment of at least 5, and if you try it enough times, you can actually find it in the test itself.  While not an "exact" test, I see this one as more of a "reasonable assumption".


Conclusions

The Volatile Enmity (VE) of Provoke is 1800 VE.

               ________________________________________________________________________________

 
Test  12B

What is the rate of Volatile Enmity Decay for Provoke?

Testing Setup:

- 1 puller character with Silence access (BRD/WHM)
- 2 level 75 members with ability in question (RDM/WAR, BLM/RDM)
- 2 characters must have 0 enmity merits (+/-), be stripped of all gear, and have access to "distance plug-in"
- Use a "level 0" monster - we chose Bumblebee in West Sarutabaruta
- This test requires a stopwatch!

- All characters must have either Stoneskin or Phalanx buffed prior to starting to ensure 0 damage
- Utsusemi shadows must be off when this experiment takes place

- Puller pulls with Silence while the 2 test characters move to 0.7 distance (same distance)
- Test character 1 casts Cure 1 for 0 HP twice (2 CE)
- Test character 2 uses Provoke (1 CE) - player 1 has a +1 CE difference
- Immediately start timing as Provoke is used
-
 Player 1 will retake hate after a certain period of time due to higher CE
- Find the time in seconds needed for player 1 to retake hate.

- Repeat the timing sequence; however, this time have player 1 gain a +301 CE difference
- This is easiest to achieve by have player 1 use Dispel and player 2 use 19 cures (320 CE vs. 18 CE)
- Player 2 uses Provoke (320 CE vs. 19 CE)
- Immediatley start timing as Provoke is used
- Player 1 will retake hate after a certain period of time due to higher CE
- Find the time in seconds needed for player 1 to retake hate.

- Repeat at higher CE differences in increments of 300 CE


Results


CE DifferenceTime Needed (s)
 +1 CE30
 +301 CE25
 +601 CE20
 +901 CE15
 +1201 CE10
 +1501 CE5
 +1801 CE0 (no hate)


Obviously, the data has some error associated with it since we're manually timing with a stopwatch.  The results I've shown are estimates as well - but if you try this, I think you'll find that it's generally accurate.  The fact that no hate is taken at +1801 CE is consistant with Test 12A in that Provoke's VE is 1800 CE.  Based on the results, it takes roughly 5 seconds to decay 300 VE; in addition, this time needed for decay is static regardless of the amount of VE, meaning the decay is linear!  This makes things much easier to deal with in general.

If we take the amount of VE lost and divide it by those 5 second increments, we get an estimate of -60 VE / second rate of decay.  Due to the imprecision of the testing methodology here (using a stop watch to determine time), we will further try to back this result up.  The biggest key to this test is to show that the rate of VE decay is linear and constant regardless of the current VE.


Conclusions

The rate of VE decay is roughly -60 VE per second for Provoke.
The rate of VE decay is not related to the current total VE of a player.


               ________________________________________________________________________________


Test  12C

Is the VE decay of -60 VE per second unique to Provoke or accurate for all Volatile Enmity?

      H0:  The Volatile Enmity decay of -60 VE per second is the same for all actions.
      H1The Volatile Enmity decay of -60 VE per second is unique to Provoke and varies with actions

Testing Setup:

- Simply repeat test 12A and 12B for different spells known to have Volatile Enmity. 
- Document the estimated VE decay rates for each action
- In this case: used Blind (640 VE), Barstone (300 VE), and Boost (300VE)

      H0:  The VE decay rates for each spell tested were all close to -60 VE per second
      H1:  The VE decay rates of at least one of the spells tested was not close to -60 VE per second

This test cannot conclusively prove that the decay rate is constant regardless of the spell or action, but to do that would require specific testing of every single ability - something I don't think anyone is willing to do.  Again, this is one of those "reasonable assumptions" that we're just willing to assume is true.  The implications of this are great - the VE decay rate is something constant and has nothing to do with the current amount of VE (Test 12B) or the action used (Test 12C).  This makes future calculations much easier.


Conclusions

The natural Volatile Enmity Decay Rate is -60 VE per second for all actions with Volatile Enmity.

____________________________________________________________________________________________________
____________________________________________________________________________________________________ 


Test 13

What is the effect of Enmity Gear on Cumulative Enmity?

Testing Setup:

- 1 puller character with Silence access (BRD/WHM)
- 2 level 75 members with ability in question (RDM/WAR, BLM/RDM)
- 2 characters must have 0 enmity merits (+/-), be stripped of all gear, and have access to "distance plug-in"
- Use a "level 0" monster - we chose Bumblebee in West Sarutabaruta

- All characters must have either Stoneskin or Phalanx buffed prior to starting to ensure 0 damage
- Utsusemi shadows must be off when this experiment takes place

- Puller pulls with Silence while the 2 test characters move to 0.7 distance (same distance)
- Test character 1 casts Dispel (CE = 320)
- Test character 2 casts Dispel using +1 Enmity
- The player without hate casts Cure 1 for 0 HP until hate is taken

- Repeat the test using different amounts of Enmity Equipment
- Include both positive and negative enmity enhancements


Results

Enmity ChangeCE Change
  -10-32
  -5-16
  -2 -7
  -1-4
  00
  +1+3
  +2+6
  +5+16
  +10+32


The most obvious thing from the data above is that negative and positive Enmity Gear have nearly exact opposite effects.  But what is the deal with the number increases themselves?  Many have felt that gear enhancement for enmity is simply a percentage increase... If we look at the data it turns out it fits the results very well.  Enmity Enhancement on Gear is an exact percentage increase...

                    Final CE = [Base CE] x { 1 + [Enmity Enhancement] / 100 }

The key to understanding this is the rounding error.  If we look at the case of Enmity -1, -1% of 320 is -3.2, but you cannot have a decimal value of enmity in this system.  The decimal truncation occurs after the addition or subtraction occurs - no rounding occurs, the decimal is simply "cut off".  This math would work like this...

                    [320 CE] - [3.2 CE] = [316.8 CE] >> Truncation of Decimal >> 316 CE or -4 CE loss

If you run this same math sequence for the +1 Enmity case, you will get +3 CE.  This "decimal cut-off" sequence is the reasoning as to why the + and - enmity cases do not exactly match always.  Generally speaking though, this would only cause a 1 unit difference at most, which is pretty minor.


Conclusions

Enmity Gear grants a direct percentage increase on Cumulative Enmity of actions.

____________________________________________________________________________________________________
____________________________________________________________________________________________________ 


Determining Volatile Enmity

The process for determining Volatile Enmity is unfortunately much more complicated than Cumulative Enmity.  I feel that this 3 step process is the simpliest way to accurately determine Volatile Enmity of an action not involving Curing HP or Damage.  For those interested in helping the community find these values, I encourage you to use this method for determining VE of an action.  Please post any findings on comments of in the BG forum thread right now.  I will add it to the running Enmity List on this blog.

               ________________________________________________________________________________

Step 1 

  - Determine the Cumulative Enmity of the Action in question.  Follow examples shown in Part III

Step 2 

  - This requires a stopwatch
  - Have a 3rd player pull a mob using a low enmity spell - we always use Silence
  - Have both testers reach the same CE using any method - 1 Dispel each works fine
  - Player 1 must gain a CE amount equal to 1 CE greater than the CE of the Action requiring testing
          -  For many spells and abilities, the CE is 1, so the amount of CE needed to be gained is 2 CE
          -  Another example of this is Dispel is 320 CE, so the amount of CE needed is 321 CE

  - Have player 2 use the ability or spell to be tested.  Player 2's CE will remain 1 below Player 1's CE.
  - Start timing exactly when the action in question is used
  - Document the time needed for Player 1 to regain hate in seconds
          -  Note that if Player 2 fails to even take hate, the VE of the action is 0
  - Multiply the time taken for player 1 to regain hate by 60 VE.
          -  Example - if it takes 5.5 seconds to regain hate, Volatile Enmity is roughly 330 VE

  -  Use this test to get an approximation of the true VE value of the action
  -  If player 2 failed to take hate, then the VE is 0 and you are done!
  -  If the VE value was above 1 (aka hate was shifted), move to step 3
  -  This test alone is not good enough to determine the exact VE of an action - it only estimates

Step 3

  -  Conduct a test similar to Test 12A - the test used to determine the exact VE of Provoke
  -  It is safe to assume that the final VE value will be a multiple of 5

               ________________________________________________________________________________

I think that you can really cut a couple of corners when testing abilities.  There seem to be a lot of repetitive values out there even with just our limited information at this point.  This is how I would formally conduct a study of an action's Volatile Enmity though.


____________________________________________________________________________________________________

Conclusions

I want to thank Elipse/Suwiside and Ryoii/Nomomii again for their help in conducting these tests.  I think we're really making great strides in testing now ever since the first numbers were matched.  We now know how both the Cumulative and Volatile Enmity systems work and how Enmity Gear works for at least Cumulative Enmity.  The second finding alone leads to a great deal of strategy discussion as mages almost always stress negative enmity enhancement and tanks certainly stress positive enmity enhancement from gear.

Here is the summary of our findings this post...

      -  The natural decay rate of Volatile Enmity is a static  -60 VE per second
      -  Enmity Gear provides a direct percentage modification of Cumulative Enmity


                    Final CE = [Base CE] x { 1 + [Enmity Enhancement] / 100 }

      -  A method of testing Volatile Enmity values of actions is given


I encourage anyone interested in these tests to help contribute!  There are a total of 360 spells or abilities in the game currently that do not involve Curing HP or Damage - we need to complete this table for CE and VE.  Even contributing 1-2 actions would help a great deal!

____________________________________________________________________________________________________
____________________________________________________________________________________________________

Comments

( 32 comments — Leave a comment )
Page 1 of 2
<<[1] [2] >>
(Anonymous)
Nov. 11th, 2007 10:25 am (UTC)
I'm fairly certain that the rate of decay of 'VE' is affected by your CHR.

It's not a huge change, but I believe more CHR causes a slower decay, perhaps 55 VE/sec, whereas very low CHR causes about 65 VE/sec. Unfortunately I don't have a 'base' CHR amount to suggest, it was quite a while ago and I no longer have my results.

(And yes, it's a serious comment, I know most people laugh at the idea of CHR and enmity being related.)
(Anonymous)
Nov. 11th, 2007 12:34 pm (UTC)
People may laugh but really it doesn't hurt to test. Grab a bard, pack up some chr and run a VE test. See what happens.

On that note, we also need to test the effects of +enmity or -enmity on VE decay. My guess is that it has no effect but its still worth testing.

~Elipse

(I should probably get one of these livejournals but its more fun to post all over kaekos.... /sorry ><)
ext_69631
Nov. 11th, 2007 12:54 pm (UTC)
great work
This is a bit of a useless comment ^^; but just felt i have to say that what you've found out here is quite amazing. I havn't had chance to read through it all yet but it really clears things out.

great work!
(Anonymous)
Nov. 11th, 2007 09:27 pm (UTC)
Nice info
Hey guys, great work, hella useful info as both a dd, and a tank. You mentioned needing some help with testing from a BLU, please let me know if can be of any assistance, just mention it in ure next post I would glad to help in any way. Thanks again for the great info.
khoisan
Nov. 11th, 2007 11:41 pm (UTC)
Wow that is some amazing work. It must have taken a long time to come up with the tests.

Keep up the work!
(Anonymous)
Nov. 12th, 2007 12:30 am (UTC)
Nifty
Well, that's pretty cool. It would be interesting to discover whether taking damage reduced CE, or how much cumulative enmity nukes give, and whether it's proportional to their damage. I'd bet they give only 1 Cumulative Enmity, but a huge amount of Volitile Enmity. The only way I can think of to be absolutely sure your nuke will do 0 damage would be to use Double Weakened status and ES + Nuke -> Dispel with your bard puller. Then you have your partner Dispel and Cure up to see when they hit the cap.
Also, when I was dual tanking to 37 on NIN, my friend used to always complain that I pulled so I started out with a hate edge, which seemed probable but neither of us had ever heard anyone say. It's nice to see confirmation.

Good work d00ds, ciao~!
(Anonymous)
Nov. 12th, 2007 12:41 am (UTC)
Re: Nifty
Oh, PS. The number you came up with for Provoke's volitile enmity makes sense. I mean, 1800 = 30 seconds/Provoke * 60 Hate Units/Second (proposed rate of decay). That would mean that Provoke's generated Volitile Enmity finishes completely decaying just as the ability comes back up.
akroma_ffangel
Nov. 12th, 2007 04:17 am (UTC)
I would go out with a WHM? And see if having more than -50 Enmity does anything. It probably caps @ 50% like everything else in this game, but It would be something to test.
(Anonymous)
Nov. 12th, 2007 06:53 am (UTC)
Me (Suwiside) and Elipse have already doen this test... kinda.

Basically we have found that the enmity gain from sentinel is 100%, which was never affected by stacking more enmity in gear. So I think its safe to assume that the enmity cap, in both directions is 100 enmity +/-

so basically, a dispel, with sentinel on is 640CE, rather than 320CE.

~Suwiside~
(no subject) - kanican - Nov. 12th, 2007 06:55 am (UTC) - Expand
(no subject) - (Anonymous) - Nov. 13th, 2007 01:18 pm (UTC) - Expand
(no subject) - (Anonymous) - Nov. 13th, 2007 03:12 pm (UTC) - Expand
(no subject) - (Anonymous) - Nov. 13th, 2007 03:13 pm (UTC) - Expand
kanican
Nov. 12th, 2007 08:40 am (UTC)
Some Enmity values were changed on test 13. The rounding or cut-off sequence is different than I had originally posted.
(Anonymous)
Nov. 12th, 2007 01:18 pm (UTC)
when you remove the units of a number beyond a certain point which you call 'cut-off', that is truncation. if you're just saying 'cut-off' for people unfamiliar with the maths term, it doesn't matter~

SE limit stats in three ways, hard caps, available caps and calculation caps;
hard caps you can see on haste, subtle blow, attack, defense...
available caps you can see in vitality, agility, -+enmity, in some cases strength...
calculation caps are things like fSTR cap, -enmity greater than 100 possibly.
so there's no particular reason to put a hard cap on -enmity, since you can't normally get that number anyway, and no need to cap it over -100 because the server would probably output the same result past that point.

ciz.
kanican
Nov. 12th, 2007 07:34 pm (UTC)
Editted it to call it "truncation" instead of "cut-off", thanks!
(Anonymous)
Nov. 12th, 2007 03:03 pm (UTC)
VE + CE on Provoke
Quick question: When you say the VE generated by Provoke is 1800, don't you mean that the Enmity generated by Provoke is 1799 VE and 1 CE?
kanican
Nov. 12th, 2007 05:34 pm (UTC)
Re: VE + CE on Provoke
Generally SE seems to like to make both CE and VE either "1" or a multiple of 5, not TE a multiple of 5. So based on that, I just said Provoke is 1800, not 1799. Honestly, it is pretty impossible to tell based on the testing - there is a margin of error to it.
Re: VE + CE on Provoke - (Anonymous) - Nov. 12th, 2007 06:45 pm (UTC) - Expand
Re: VE + CE on Provoke - kanican - Nov. 12th, 2007 07:18 pm (UTC) - Expand
(Anonymous)
Nov. 12th, 2007 09:12 pm (UTC)
Sentinel?
Anyone attempted testing sentinel yet?
and on top of that Sentinel with guardian merits which should lower the rate of decay on VE?
I think that will be a pretty complicated one
kanican
Nov. 12th, 2007 09:30 pm (UTC)
Re: Sentinel?
I believe someone tested that Sentinel is

1) The equivalent of +100 Enmity while active
2) The cap of Enmity enhancement is 100 so using Enmity gear with Sentinel is pointless (i.e. you will only ever get +100 anyways)

Guardian I would assume lowers the -60VE/sec decay rate. We have tested and already found another ability - Foe Sirvente - which does not alter the +/- enmity, but does lower this decay rate. Guardian likely does something similar?
fullburn
Nov. 12th, 2007 11:08 pm (UTC)
So the big burning question remains: how much Enmity results from dealing damage with physical attacks or spells, taking damage, and curing damage?

I may start helping with these tests myself, because this has huge implications for every job at learning to manage their pace.

In particular, consider this: Provoke may be effectively completely composed of VE, but it is enough VE to mask all other VE decay. This means that if you are provoking every 30 secs, in addition to the VE "spike" all your other VE will never be allowed to decay.

My suggested test method will be pretty simple, and similar to yours. First, determine if damage dealt or HP cured contributes to CE, and how much, then repeat for VE. The only difference is that the amount of damage dealt (which CAN be controlled) is bound to be factored in. Based on this, we need to create a large spread of damage values in order to determine whether the hate produced for each is linear (most likely) or something else.

I also will try to determine if different weapon classes (at least 1h/2h/ranged) show any variation, and what added effects contribute if anything.

This would be made much easier of Phalanx and Stoneskin were present in the table. Not sure why these spells so critical to the testing procedure were not among the first ones to be tested? Casting them in combat will be necessary against any mob with enough HP to use in damage tests. So if nothing else, I'll get those out of the way first.
(Anonymous)
Nov. 13th, 2007 02:56 pm (UTC)
Enmity Testing
What about Counterstance? This seems to be AoE hate because when I enter Operation Desert Swarm, I pop Focus Dodge and ocunterstance and when the mobs wake, they come after me opposed to the BLM that slept them.
(Anonymous)
Nov. 13th, 2007 03:32 pm (UTC)
2 hr testing: Benediction ~ Dmhlucky
Will you be planning tests on 2 hrs as well? And if so, how will you do so. Some 2 hrs will naturally draw more enmity than others, some 2 hrs are immediate effect, some 2 hrs are a variable effect and some 2 hrs are a "Base" effect which sets up other possibilities. EX. Benediction, Cor 2 hr, Manafont. How will you test on these? Cor 2 hr will technically vary based on what it does, and manafont its self does not actually DO anything to anyone. Inversly Benediction will vary based on how much it sures and how many stats it takes off. I can try testing a "Dry" Benediction with my Bro. I'll saruta on a bee, benecide after a 3rd person pulls w/ Silence, and have my Brother dispel the mob til he has hate. Also will benediction be VE ot CE? Since its a 2 hr, it could take a long time to test the diff possibilities. I'll post back with the details and what i find out.
kanican
Nov. 13th, 2007 08:34 pm (UTC)
Re: 2 hr testing: Benediction ~ Dmhlucky
I don't see what's different about 2hrs that make them harder to test. We already tested PLD 2hr using the same/similar method and it turned out very well.

To my knowledge, whether a spell is resisted, has no effect, or lands does not matter as far as enmity gained is concerned. The only exception has been Bind, which has some weird decay properties when it wears.
(Anonymous)
Nov. 13th, 2007 04:57 pm (UTC)
VE and enmity gear
Noticing that voke is 1800 and depletes in 30s at a rate of 60E/sec it seems that voking a slept mob is useless in an attempt to build hate up, and you'd be better of just waiting until its about to be woken. However if enmity gear does effect the VE as it does CE voke should last longer. Test was a smn/whm pld/war drk/war smn earthen wards the party, and pulls with silence, DRK with 0 enmity, PLD with 37 enmity. 1.37 * 1800 = 2466 or a difference of 666 VE, aprox 11 seconds.

Test 1 (no enmity gear)
Smn pulls << takes hate
Pld vokes << takes hate
1 second later Drk vokes << takes hate

Test 2 (+37 enmity gear)
Smn Pulls << takes hate
Pld voke << takes hate
1 second later Drk vokes << doesnt take hate

Test 3 (+37 enmity gear)
Smn Pulls << takes hate
PLD voke << takes hate
11 seconds later DRK vokes << doesnt take hate

Test 4 (+37 Enmity Gear)
Smn Pulls << takes hate
PLD voke << takes hate
15 seconds later DRK vokes << takes hate

Clearly Enmity gear affects VE aswell, meaning with +37 enmity, voke will last 11 seconds longer then its timer, making it possible to build hate on a slept mob by voking it every 30s.

Thanks again for the great info, cant wait to see how you guys plan on doing dmg ^^
kanican
Nov. 13th, 2007 08:36 pm (UTC)
Re: VE and enmity gear
The previous test on Enmity Gear was only "the effect of Enmity Gear on CE". We've found results similar to yours and Enmity Gear probably does the exact same to VE as well. We know it does not affect the decay rate, but it will increase/decrease that hate spike.

Will probably be in the next post whenever that is.
(Anonymous)
Nov. 15th, 2007 07:00 pm (UTC)
60VE/sec
60VE/s can also be expressed, 1VE/frame, or 1VE/delay

Maybe a useful number in the future.
Page 1 of 2
<<[1] [2] >>
( 32 comments — Leave a comment )