Overview
I realize the first post on this was rather dry and boring; however, I think it provided a number of controls which I needed to do more unique testing. For those bored by the first post, I encourage you to read this one more carefully, as I assure you it will have unique and profound implications on the game - this post goes over testing which can prove and quantize hate.
The problem with enmity testing so far has been the lack of real quantifiable numbers. I think most agree that there has to be some real number behind everything, but it's so difficult to test because the only observable thing you can get from this game is who a target has hate on. I think the really elegant part of this testing is that through the use of some very very simple and easily reproducable tests, we have found a way to actually quantify something that players have been trying to for years. Things people might be most interested in seeing in this post...
1) Some enmity decays while some does not (think JP wiki Model).
2) There is a hate cap and we prove it.
3) We have found what we believe to be the BASE UNIT of hate.
4) We can prove how many base units of hate any ability grants
5) We can prove the number of base units at the enmity cap.
All of these things are unique to these tests. I have been and will continue to be incredibly careful in detailing how each test was done so that any person can reproduce this to verfiy the accuracy. I think the best part about this testing is I started this to disprove the JP Wiki; however, as it turns out, it is conceptually pretty accurate. The key point of this post is quantizing enmity, though.
I would like to thank Ashira for doing all these tests with me so far. I personally find it pretty exciting and I think she does a bit as well. We're both really excited to get this information out. I hope this is something we can build upon quickly in the near future - enmity has been quantified, so the rest should not be quite so difficult. Pics are sparse and probably not that amazing again since we're doing something pretty theoredical. I'll supplement with a good pic of the day section.
Previous: Enmity Testing (Part I)
Next: Enmity Testing (Part III)
Enmity Table
EDIT: Thanks to you guys who carefully read the post and tried to find any errors in methodology. I've changed this post to correct the errors people have found so far. I'll cite if the poster was not /anon.
Eamon (Lakshimi) - Dispel does have decaying hate component; I originally stated it did not. Added test 5B.
________________________________________
________________________________________
Does the enmity generated by Dispel (any action can be used) fully decay?
H0: The Enmity generated by Dispel (or action in question) does not fully decay
H1: The Enmity generated by Dispel (or action in question) does fully decay
Testing Setup:
- 1 puller character with Silence access (BRD/WHM)
- 2 level 75 members with Dispel or the ability in question (RDM/NIN, BLM/RDM for Dispel)
- 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 10 times consecutively
- Wait 2 minutes for any possible long term decay to occur
- Test character 2 casts Dispel until hate is FIRST SHIFTED
- To clarify, the mob may shift towards player 2 but then back to player 1 - we are looking for the shift
- Whether the mob shifts to player 2 then stays or goes back to player 1 is irrelevent to this test
H0: The 2nd player will get hate on exactly the 10th Dispel
H1: The 2nd player will get hate on the 1st to 9th Dispel
Note that this test does not tell you if Dispel has a decaying portion - it only tells you if it has a significant residual hate portion. There is a seperate test to prove if an ability has a decaying portion. Based on the results, we summize that Dispel does have some nice residual hate remaining - we don't know yet how much, but we do know that if we sit for awhile after a Dispel is cast, there is some left over hate that doesn't go away!
So the big keys to this test -
1) Dispel has a significant non-decay (residual) hate component
2) This cast can be used on any action that generates enmity to test possible residual hate
- e.g. If you run this test with Blind or Provoke, 2nd player gets hate after 1 action
Ashira and I ran this test for more spells than just Dispel. The following is a summary table of spells we tested and the number of actions it took player 2 to receive hate.
| Blind | 1 | Decay |
| Dispel | 10 | Non-Decay |
| Paralyze | 1 | Decay |
| Poison | 1 | Decay |
| Provoke | 1 | Decay |
| Silence | 1 | Decay |
| Slow | 1 | Decay |
Of the 7 actions we first tested, Dispel was the only one that took longer than 1 cast to get hate from player 1. This shows that the other 6 obviously have some form of decay to them and not much residual hate. Test 6 attempts to shed light on the form of this decay - most notably if all of the hate decays or only some of it. Before moving on to decay though... I want to first show how you can prove that decay does or does not exist for Dispel or any other move that had more than "1" as the answer to test 5A.
Test 5B
Does any part of the enmity generated by Dispel (any action can be used) decay?
H0: The Enmity generated by Dispel (or action in question) all remains - that is to say no decay
H1: The Enmity generated by Dispel (or action in question) does not all remain - there is some decay
Testing Setup:
- 1 puller character with Silence access (BRD/WHM)
- 2 level 75 members with Dispel or the ability in question (RDM/NIN, BLM/RDM for Dispel)
- 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 1 time followed by Cure1 for 0 HP
- Wait 2 minutes for any possible long term decay to occur
- Test character 2 casts Dispel 1 time
H0: The 2nd player will not get hate at all
H1: The 2nd player will get hate, but then lose it back to player 1
The result of this test for Dispel shows that hate was taken by player 2. This means that even though player 1 was at a hate level of 1 Dispel + 1 Cure for 0 (however much hate that grants - it does grant some though), player 2 took hate. If there was no decay component to Dispel, then player 1 would have held hate throughout. The fact that player 2 took hate off that 1 DIspel proves that there is at least some hate generated by Dispel that does not remain, even though it has a signifincant part of it that does.
Does this mean DIspel is still useless for testing? Not really! If you try this test, you'll realize the decay fully occurs in a very short period of time, on the order of seconds. As long as you wait between your actions, you can disregard the decaying component of Dispel - this will be done in later tests.
Conclusion
There is a decaying component and non-decaying component to hate in actions.
(Note we have tried other spells - we have found that spells affecting movement have different properties, so this test should be avoided when using Bind)
________________________________________
________________________________________
Do decaying enmity actions generate any form of residual or non-decaying hate?
H0: Decaying actions (use the 6 found in test 5) leave no residual non-decaying hate.
H1: Decaying actions (use the 6 found in test 5) leave some residual non-decaying hate
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 Blind 10 times
- Wait 2 minutes for any possible long term decay to occur
- Test character 2 casts Blind until hate is taken
- Document the number of casts it takes before hate is secured - i.e. does not shift back to player 1
- Wait 2 minutes after you think hate has completely shifted and will not return to player 1 to verify
H0: The 2nd player will either never fully secure hate or do so after the first Blind cast
H1: The 2nd player will fully secure hate on the 10th Blind
This test is exactly like test 5, with the exception that in test 5, we documented when hate first shifted - we did not care if hate shifted to player 2 but then went back to player 1. In this test we do care about that shift back to player 1 and document the number of casts to have the hate placed solely on player 2 for an indefinite period of time. The results show that there is some type of residual or non-decaying hate associated with these decaying actions. Of the 6 decaying actions we tested in test 5, all 6 showed the exact same result.
Now let's mix this test up just slightly...
Test 6B
Is the non-decay hate associated with decaying spells the same amount? Yes
Run this again but this time, use any spell on that list of 6 we tested - Poison, Blind, Provoke, Slow, Silence, or Paralyze. For instance, after doing the normal pulling on 3rd character and getting the same distance, have player 1 cast a random assortment... let's say Blind x3 > Provoke > Slow > Paralyze x2 for a total of 7 actions. Now wait 2 minutes then have player 2 do this - cast cure on himself for 0 HP 7 times.
What you'll find is that it takes the same number cures for 0 as the number of actions player 1 performed, regardless of the order or number of those 6 actions used! This leads to the idea of the BASE UNIT of hate - that is even the most basic and useless actions (like curing yourself for 0 damage) grant you some form of non-decay hate - the smallest unit seen in the game.
Conclusion
All actions must leave at least 1 BASE UNIT of non-decaying hate
We arbitrarily call this amount of hate the BASE UNIT of hate, i.e. 1 unit of hate/enmity
________________________________________
________________________________________
How many BASE UNITS of enmity equals 1 Dispel's worth of enmity?
- 1 puller character with Silence access (BRD/WHM)
- 2 level 75 members with Dispel or the ability in question (RDM/NIN, BLM/RDM for Dispel)
- 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
- You can actually avoid using Stoneskin or Phalanx and a level 0 mob will hit for 0 if it doesn't critical hit
- Puller pulls with Silence while the 2 test characters move to 0.7 distance (same distance)
- Test character 1 casts Dispel 1 time
- Test character 2 casts Blind 1 time (1 action or 1 base unit of hate)
- Test character 2 continually casts cure 1 at MAX HP until hate is taken
- Record the number of base units or actions needed to take hate (remember the first cure is action 2)
Number of base units or actions to overtake hate: 320
That's right... I made Ashira cast Cure 1 on herself for 320 straight times. She told me I was insane the entire time but this actually did work! It was a bit daunting at first because we didn't know when or even if it would eventually even out. This the best part of our testing so far - the implications for this are really great I think! We actually set up a macro using Windower to loop so she didn't actually have to cast it herself each time.
The keys to this very important test:
1) Dispel is 320 base units of non-decaying hate - the decay component is unknown from this test
2) Every action must at least produce 1 unit of base hate
2) This cast can be used on any action that generates enmity to quantify non-decay hate in base units
- e.g. Very simple case - Provoke (and the other 5 decays tested) only have 1 unit of non-decay hate
We have now done our first quantitative analysis of hate (really the first every test that is experimentally reproducable). We can now attack the concept of a hate cap. This is done in test 8.
Conclusion
Dispel provides 320 units of non-decaying hate.
________________________________________
________________________________________
Does a hate cap exist?
H0: A hard cap does not exist for enmity
H1: A hard cap does exist for enmity
Testing Setup:
- 1 puller character with Silence access (BRD/WHM)
- 2 level 75 members with Dispel or the ability in question (RDM/NIN, BLM/RDM for Dispel)
- 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 50 times consecutively
- Test character 2 casts Dispel until hate is taken
- Record the number of Dispel casts it took player 2 to get hate
H0: The 2nd player will get hate on exactly the 50th Dispel
H1: The 2nd player will get hate before the 50th Dispel (observed 31st or 32nd)
This is essentially a longer version of test 5 - instead of spamming 10 Dispels we're spamming 50 first in order to see if some type of hate cap is reached. What we see is that hate is shifted to player 2 after 31 Dispels; player 2 will have hate for a second, then it will move back to player 1; the 32nd Dispel will secure hate on player 2 forever. The explanation for this can be found from test 5B - 31 non-decay hate portions of Dispel is below the hate cap, but 32 non-decay hate portions of Dispel is above the hate cap. In addition, 31 non-decay hate portions of Dispel + the decaying portion of 1 Dispel is above the hate cap. Think about it that for a bit and it should make sense.
There is a hate cap occurring around 31-32 Dispels (non-decay portions). If a cap did not occur, we would expect the results to be the same as test 5A and 50 Dispels would be required to shift hate. The fact that this does not occur suggests hate was capped at some point. We can quantify this cap in terms of Dispel. Because we know the number of base hate units for a single Dispel's non-decaying hate (320 by test 7), we can calculate a range of possible hate values the cap resides at...
(Number of Dispels for Cap) x (Number of Units per Dispel) = (Number of Units at Cap)
(Apply for Low End Range) ... 31 x 320 = 9920 units
(Apply for High End Range) ... 32 x 320 = 10240 units
The cap must reside between 9920-10240 units. The exact cap of this can be found by applying 30 Dispels then slowly building up hate smaller units at a time using Cure 1 for 0 HP... This will be done and posted at a later time. For now, we have merely established that it does exist and the range of values it can fall in.
I was actually quite surprised by the number ranges. I expected from the start that if a hate cap existed, it would occur at a power of 2, i.e. 256, 512, 1024, etc. Most other things in this game such has haste and damage reduced follow this base 2 system. Despite my gut feelings that this should occur, I feel confident that the cap is within this range - obviously the nicest numbers being 9,999 or 10,000. The exact value still needs to be verified.
Conclusion
A hard hate cap exists
The cap can be expressed in base units at a range between 9600-10240 units of hate
________________________________________
________________________________________
Implications of Quantifying Hate
I believe that making that leap from an abstract idea or model to actual numbers opens up a great deal of possibilities when it comes to enmity testing. I was extremely careful to model everything prior to this to lead up to this quantizing of enmity - establishing that base unit of hate. I think it's possible to find the enmity values for just about any ability and even start getting into decay rates, the effect of enmity gear, damage implications, etc., so this is a very big deal in my opinion.
Even with the added possibilities of quantifying hate, I think it's important to remain focused on making each test clear, reproducable, and accurate. Again, every single test here has been done using the basic priniciple that the only observable piece of hate information the game provides you is who is at the top of the current list. It remains pretty difficult to easily prove harder concepts, although, I think the hardest part is over now.
Citation and Information Flow
At this point, I don't really want to get into the 'politics' of information sharing on the internet too much; however, I think it's important to be wary of them. Ashira and I want this information to be readily available (afterall, that's why we did it in the first place), yet we still want our due credit for having gone through things like spamming Cure 1 320 times. I ask that anyone who uses or posts this information, as raw as it is, to at least cite her or me.
Later on, I feel it will be necessary to branch out and include help for exaustive testing of every ability in this game. At that point in time, I think the politics of this will become more important. We don't want to be assholes about who owns what information or test, but I think we have to be careful nonetheless. Numbers are very easy to copy - the testing behind them is not. We just ask that people don't start copying numbers and calling them their own, as well as respecting our wishes if we wish to temporarily withhold information from major informational sites.
________________________________________
Conclusions
I think this post has really completely disproven my own post just months ago on modeling enmity...
- I said there was no hard hate cap - there in fact, is one
- I said there were not 2 types of hate - there are actually 2 types (a decaying and non-decaying)
- I said the JP Wiki was flawed - it's not perfect, but it's surprisingly accurate at least conceptually
I set out to prove the JP Wiki wrong, and I've basically just helped to prove its conceptual accuracy over the first 2 posts - actually just adding the quantitative data for it. I think that's actually a nice thing as it shows the testing I've done has been unbiased towards my own original thoughts on enmity. This is not to say JP Wiki was entirely accurate though. For instance, they seem to split actions into 1 of 2 catagories, but based on the testing here, we believe each spell actually fits into both catagories, just different percentages of it.
For those of you who were too lazy to read the testing portions, I will go over the key points and findings shown in the tests outlined in the post...
1) Dispel has a signifincant non-decaying hate portion, but the other 6 spells we tested did not
- Blind, Poison, Provoke, Slow, Silence, Paralyze were the other 6 spells - all decayed near fully
2) The other 6 spells decayed, but left 1 "unit" of non-decayed hate - we called this the Base Unit of Hate
3) All actions in this game produce at least 1 unit of non-naturally decaying hate (the Base Unit)
4) Dispel produces 320 base units of non-decaying hate
5) A hard cap on hate does exist - it ranges from 9920-10240 base units of non-decaying hate.
We assert that this is the first "proof" that a hard cap on hate exists; this is also the first experimentally proven method to show that hate can be expressed in some form of accurately determined unit. A lot of possibilities are opened up with this. I think the best way to go from here is to collect more general data on abilities. Dispel is the only spell we've found to have significant non-decaying hate - a property which makes it incredibly useful in testing. Finding other ways to accurately and precisely grant non-decaying hate (more than 1 unit at a time) would help tremendously in future testing.
I personally find this testing business really exciting for a video game (the nerd within me). For those of you who really don't like to do math online, I'll try to counter with some actual content related posts - I've been told I am overdue for a Black Mage Guide update. I'll try to balance things out in the future.
________________________________________
Pic of the Day
Random assortment of extremely old 2005 pics.
First Limbus the day the event came out (October 2005) - Apollyon NW (Teamed with LB).
Zone went from just came out to officially soloed in 1 year.
Tarutaru Line! (October 2005)
I miss a lot of those guys :(
Again, from October 2005.
DO NOT REPLY OR POST THE NAME OF THE PERSON IF YOU KNOW IT - I will remove this if someone does.
This was probably the first time anyone had ever flipped out on me over the internet. I think this person held/holds an eternal grudge against me and TK over the fact Ashira remains in TK. I remember years after this incident it was clear he still had not let this go. I actually had 0 power to recruit, recommend, or kick members at the time of this pic, so I find it odd that he would single me out to say this. Regardless, I think the person still feels it's my fault; although he's probably quit the game by now. He actually accused me of Speed Hacking too I think a long time ago - he shut up after he realized what Herald Gaiters were.
I don't think I've ever showed anyone this pic. I honestly feel terrible for the person - not because he was removed about 3 days after he got into TK (long story I won't talk about it) - but because he can't seem to let it go. I don't think anyone in TK ever really told him why he was really kicked (wasn't RMT related really) in an effort to spare his feelings. Obviously that didn't turn out too well. On one hand, I know some people take this game very seriously sometimes, myself included, but to hold a 2+ year old grudge over a misunderstanding in a video game is quite sad. It's another example of why everyone who plays online should remain grounded in real life.
Drama Thread of the Day
http://killingifrit.com/forums.php?m=pos
I started reading KI BLM forums recently. I'd like to say it was to learn something new or to get different takes on styles from other career BLMs, but honestly the best thing I've found are good drama threads. Seeing as this post dealt with the concept of experimental testing, reproducable and transparent evidence, and proof, I think this was a good time to throw this one out.
It's about a mage asking whether the Tier II ancient magic 2 spells for BLM grant a magic accuracy bonus with extra merits on only MBs, or anytime - the wording of the description is pretty ambigious for both Japanese and NA versions. Despite this, you'll have people swearing on the Bible it works in different ways with no evidence outside of "I seem to resist less with it!". Black Mage discussions are always so heated; I find it really amusing people would care to debate something so mildly trivial and without proof either way.
For anyone that happened to care - I merited 1 point in all 6 spells. My reasoning is I don't know how the merits work, but I do know I like messing with useless spells sometimes I would never use in practice like Quake II. I don't think it would hurt performance too much either way. There are more important things to dwell on.
________________________________________
________________________________________
Page Summary
![[info]](http://l-stat.livejournal.com/img/userinfo.gif)
sevataru : (no subject) [+0] - xhene : (no subject) [+0]
- (Anonymous) : (no subject) [+2]
- (Anonymous) : (no subject) [+8]
- (Anonymous) : (no subject) [+0]
- (Anonymous) : (no subject) [+3]
- (Anonymous) : Please rewrite 'Enmity' entry on FFXIclopedia [+4]
- (Anonymous) : Thanks [+1]
- aurik_ffxi : (no subject) [+0]
- aurik_ffxi : (no subject) [+13]
- aesthetik_dekay : (no subject) [+5]
- (Anonymous) : A few comments [+3]
- (Anonymous) : FFXI base calculations. [+0]
- (Anonymous) : Interested in helping [+7]
- (Anonymous) : Barspells and Terminology [+2]
- (Anonymous) : Some more testing [+1]
- (Anonymous) : (no subject) [+1]
- (Anonymous) : (no subject) [+0]
chasethestars
Comments
Anyway here is what popped into my mind.
Dispel is non decaying hate based on time. It might be worth while to determine if it will decay based on damage taken. I would like to say it would but hey if it doesn't that would be amazing. I was shocked to see a spell that doesn't decay over time. This also interests me from a Pld perspective to see if there are other spells that generate enmity that doesn't decay.
I think it would also be very important to test if this is a base hate instead of a hate cap. I think thats important because what you may or may not be seeing is a base cap of non decaying hate. What I mean by that is, what if there are two different types of hate like the JP wiki first presented but handled a different way. 1 type that can exceed the base cap by a large margin but quickly decays down to your base, and another type of hate that slowly raise your base until it hits its cap.
So I present a test option for you. 2 players dispel says 32 times. So they're both at what is currently believed to be the hate cap. I would say have each one dispel back and forth a few times so you can be sure you're "at the cap". Now have the rdm spam blind, paralyze, slow, etc and if there is indeed a hate cap and no distinction of hate caps between spell types, every time player 2 casts dispel regardless of how many spells player 1 is spaming, it should turn to him every time a dispel is cast.
If it does not return to him, then we know that there are two different hate caps based on the types of spells. This could be very important for a tank to know. Building your base hate at the same time as your max hate could give you a steady flow over the course of a fight.
Anyway that was just a thought I had based on what you're presenting.
Oh and If you're thinking of a way to test "cure" hate decay. I suggest a berserkers torque on two pld/rdm's. Phalanx so you take no damage and just cure a bunch of times to see what happens. It may not be the best way, but 50 damage a tick would provide quick damage with no hate implications. You could use poison potions or something but that would take forever >_>
I use berserkers on pld from time to time to help spike hate with Cure 4. So if you need a pld for the testing I'll be happy to help out but you would also need someone with capped enmity to match or it will skew the testing.
Thanks for taking the time to test all this information. Enmity may not be the most exciting thing to everyone but its a hot topic for some. Best of luck and I look forward to seeing more info!
~Elipse
I think the term "non-decaying hate" is mis-leading. I think the better desciptive term is "decay that does not decay naturally over time". That hate you get from Dispel can still be lowered if you get hit - it just won't decay naturally over time if you sit there like the other ones we tested would.
For the cure thing, I think the easier way is to continually equip and unequip Zenith gear to lower your HP. Torque can't be worn by a RDM, which is annoying.
By the way, if there is a hate cap, it should be a power of 2 minus 1, this is from the nature of binary coding. So for example: if hate is capped at say 2^8 (which your test have shown it does not), then the hate cap would be 256-1 = 255. 0 hate still requires a binary equivalent.
So my guess is that hate has a cap of 2^14 - 1 = 16383 and that dispel has a very slow decay hate loss property. I mathematically modeled both your cap and my guess at the cap, if you were do the example I suggested above.
Here is what I got assuming that dispel has a slow-decay hate loss property using the example I discussed above:
It would take 56 casts of dispel from player #2 to get hate. Whereas if your cap is correct, I assumed 10240 hard cap, then it would take 39 casts of dispel from player #2 doing the example I suggested above.
Non-Decay:
It would take 52 casts of dispel if I am correct about the hard level cap of enmity and 32 if you are correct about the enmity hard cap level.
Outside of this dispel issue, I think you have done a good job experimentally showing how enmity works. I look forward to reading more on your enmity report.
- Eamon of Lakshmi
The question about binary quantification is a big one. I personally felt that it had to be a binary number as well. When I first ran this, I told Ashira that it would have to be 2^X at the hate cap; so I was just as surprised to see the value fall in a range not anywhere near a 2^X value (8000ish and 16000ish are the 2 closest ones). You can debate how to interpret the findings here, but this is what we found using our methods. I also agree it should have been binary, but our results say otherwise.
The key to any experiment is to remain unbiased at all times. I do not want to be biased by the binary system, although I strongly believe it should be that way. If our results do not show this to be true, I will not force it to be true. Until someone points out serious flaws in the methods we used, I won't waver from the range of units the hate cap fell in.
Also, regarding the use of binary. There are situations where people choose to use less than the maximum alotted values given by binary. For instance, sometimes people choose to use 3 1/4 bits - meaning they use 4 bits of data, but don't use all the numbers you could. A digital clock is a good example of this. I am not an SE programmer so I do not want to speak for how/why they did something. I only extract the results based on our personal testing.
Easiest way I have found to test Dispel's Decay without taking a long time...
1) Puller Pulls with Silence
2) Player 1 casts Dispel and takes hate
3) Player 1 casts Cure1 for 0 HP on himself
Player 1 is now at Dispel Hate + 1 base unit
4) Player 2 now casts Dispel...
If Player 1 maintain's hate, that means he has not even decayed 1 base unit of hate. This implies no hate was lost, not even 1 unit.
If player 2 takes hate, that means at least 1 unit of hate had been shed from Dispel. This means Dispel does shed hate, albeit only a tiny amount.
If you run this, I think you'll see that Player1 keeps hate. This implies that not even a single unit of hate has been shed by Dispel's hate.
Misin
#1 dispel 50x
wait 10s
#2 dispel until target change
#1 dispel 50x
wait 200s
#2 dispel until target change
however i feel it's likely there's no decay, i'd like to know if you have tested with finale already?
1) Puller Pulls with Silence
2) Player 1 casts Dispel and takes hate
3) Player 1 casts Cure1 for 0 HP on himself
Player 1 is now at Dispel Hate + 1 base unit
4) Player 2 now casts Dispel...
If Player 1 maintain's hate, that means he has not even decayed 1 base unit of hate. This implies no hate was lost, not even 1 unit.
If player 2 takes hate, that means at least 1 unit of hate had been shed from Dispel. This means Dispel does shed hate, albeit only a tiny amount.
If you run this, I think you'll see that Player1 keeps hate. This implies that not even a single unit of hate has been shed by Dispel's hate.
/bow
/kneel
/beg
/plead
...
/pout until you agree.
http://wiki.ffxiclopedia.org/wiki/Enmity
- Itazura
The information about enmity on these posts I guess are "made and owned" by Ashira and myself. At this point in time, I personally don't feel it is a good idea to put something on a database with such limited knowledge - although the information is pretty cool and new, it only scratches the surface of the entire enmity concept.
Also, Ashira and I are BG forum members (her being much more active than myself, but we both consider ourselves a part of that community). If either of us were to transfer our information after extensive testing, it would be towards the newer BG wiki, not the old Wiki which you linked. I personally do not feel slated against the old Wiki, but I know Ashira does; in addition, BG was the original source that helped to popularize my own blogger, so I do feel it would be better to for us to contribute there.
Unfortunately, he didn't have a single shred of experimental data, method, or explanation with any of those numbers, so I can see why the numbers would seem so useless and made up. I don't think anyone should claim any quantifiable number to enmity without an extension explanation and test to prove it since it is so difficult to control and observe. I question how he got any of those numbers - in the discussion he doesn't seem to be very clear about his methods.
In the end, I think all that 99% of the players would care about is the numbers. I personally feel the testing is the most important part though. No matter what comes out, we will probably never know how SE programs hate - any information or models we choose to use come from our own testing and observations. Because no one is going to hand us a clear answer, the testing and extrapolation of data becomes incredibly important.
Whoever wrote that little "Enmity Numbers" guide was probably doing something right, but seemed to focus purely on the numbers, not the methodology, which is why no one probably took him seriously. I think our own tests so far as shown that he was at least partially correct - although because he never said how he came up with them, I don't think he can really do much with them.
I'd especially like to see how your conclusions mesh with the idea of an emnity curve as you suggested in your previous blog entry about emnity. Plus, yknow, graphs are awesome :D
ubermonkey - Fenrir.
It isn't clearly the case that a SE programmer would use a binary value as the hate cap, since it is very likely integer values which store nondecaying hate (NH) and decaying hate (DH).
In this case, you see one writing a block of code that calculates the addition of NH as such:
#define NH_MAX 10000
...
NH += action.NH
if ( NH > NH_MAX )
{
NH = NH_MAX
}
The main time you will find arbitrary limits to conform to 2^X values is
1) using fixed-point math. IE, 1 unit is actually treated as 1/256th
2) when storage size reaches the upper limit of the number. For example, single bytes can only store 256 different combinations of bits. Therefore if some stat (say, INT) was stored using only a single byte, there would be no possible way to represent any value above 256, therefore the game may explicitly restrict INT from exceeding this value.
Regardless, I am looking forward to more tests on enmity.
- Eamon (Lakshmi)
One thing that I think could use some more clarification is how you arrived at cure for 0 being equal to the base unit of hate. You use this basis somewhat extensively, but only mention at how you arrived at that in a small section of your test (where you measure it with the non-decaying portion of other spells).
Also, you mention that Dispel as the spell with the least decaying properties found thus far, but what about the spells of choice for tanks? Flash and Stun?
We arrived on this based on an assumption - since many actions seem to decay but leave back this 1 unit. So many things we've encountered so far seem to use this unit; Provoke, Gravity, Silence, Slow, Blind, Paralyze, etc. (I could really name you 50+ we've tested) all decay and leave only this 1 unit of hate left. Regardless of which we used and how many of each we used, you could always balance the number of actions you performed with the same number of cure for 0s. Based on this observation, we deduced that this may be the lowest amount of hate you could produce in any single action.
This is something that is really easy to try out for yourself. Just get someone to pull a mob for you and then have 2 characters sitting at the mob. 1 casts random decaying spells while the other balances with cure1 for 0HP. You can literally pull off any combo you would like - you could spam protect 4 50 times and balance with 50 cures. Likewise, you could do 20 blinds, 5 refreshes, and 25 provokes - it would balance at 50 cures again.
Choosing a "unit" is arbitrary at this point since we do not (and may not ever) know the true machine code "unit" of hate. Our choice to use this cure for 0HP as the "unit" was based on the fact that it was very small and so many things seem to leave this unit behind. When we say "Dispel is 320 units", it does not mean the game necessarily sees Dispel as 320 units - it just means if you spam cure 320 times you get the same hate as Dispel. We have started to quantify hate; however, we have not shown that our quantification and numbers are the ones the game actually uses. There is a difference which I should probably clarify.
Regarding other spells - I'm guessing you're interested in Stun, Flash, Sleep, Bind, and Kurayami - we've done a few tests for them, but nothing all that formal. Of those 5 I listed, I can tell you that Sleep II produces the most cumulative hate; we think it's roughly 1.5 times higher than Dispel so around maybe 450-500 "units". We haven't really done too much with those since their recasts are really high.
Going back to reading now.
Lakshmi:
Ryoii - SAM/WAR/RNG/DRG/THF
Nonomii - RDM/WHM/BLM/BST
I've started some investigation into Provoke, but I'm finding it rather difficult without three characters with Stoneskin.
Provoke Initial Test Results posted on http://www.ffxionline.com/forums/paladin/6
Dispel x 5 > Initial Provoke > Dispel x 4
Provoke > Dispel x 4 for ~9 sec (does not fit with Hypothesis A; should be 6 sec)
Provoke > Dispel x 3 for ~15 sec
Provoke > Dispel x 2 for ~21 sec (does not fit with Hypothesis A or C; should be 18 sec)
Provoke > Dispel x 1 for ~27 sec
Given: Fully Decayed Dispel = 320 Units of Non-Decaying Hate or final Hate/Enmity Units (fHU)
1600 HU > Initial Provoke > 1280 HU
Given: Fully Decayed Blind = 1 fHU
Provoke x 3 > Blind x 3 > Provoke x 2 > Blind x 2 > Provoke x 1 > Blind x 1
Blind was used before Provoke, so the person who had the highest HU before HU became equal keeps hate.
Fully Decayed Provoke = 1 HU
Hypothesis A:
Provoke = 1500 initial Hate Units (iHU) which decay at a rate of 150 HU every 3 sec down to 1 final Hate Unit (fHU)
Hypothesis B:
Provoke = 1500 iHU which decay after 3 sec at a rate of 166.67 HU every 3 sec down to 1 fHU
Hypothesis C:
Provoke = 1400 iHU which decay after 6 sec at a rate of 200 HU every 3 sec down to 1 fHU
Hypothesis B holds against all of my initial data, but I plan to do some more testing now that I have some ideas to test more thoroughly.
Note: '>' symbol should be read as greater than
Comment: My method for determining the time at which hate was lost was to keep track of who the enemy was targetting while watching the Provoke recast from Windower. I made sure to do this 5 times since the enemy's attack animation made this difficult. I'll probably up this to 10 times for future testing. I've also found that if longer durations are needed I can use my 2 hour recast compared against the recast of the ability being tested.
I assume you just "zoned" the hate, but actually zoning does not fully get rid of hate (if you can believe that). We've done it already, but have yet to post it. Just be aware that if you zone, don't expect a full hate clear. It may not have influenced the data, but in order to get something postable, you have to go through all the controls I think.
I'm glad others are interested in this. We've actually gotten a ton of number in the last few days that we're nearly ready to post. Right now, we have 5-6 spells down, Utsu shadow loss hate loss, the actual hate cap (it's 10,000, very nice number), and some other findings. I may move this information to a formal forum later on once enough information is verified - this way more can get into the testing procedure.
Also, what terms are you now using? I started using iHU and fHU to attempt at keep aligned with your terminology.
As for terminology, from next post and on, I will probably refer to the non-decaying type of hate as "Cumulative Enmity (CE)" and the decaying type of hate as "Volatile Enmity (VE)" in an effort to be consistant with JP Wiki.
Suwiside and I (Elipse) were out doing some enmity testing today.
Zoning hate
TEST:
Player 1 for pull of bumblebee
Both players move to 0.7 distance. Both players are naked with no enmity gear on or merits.
Player 2 does 5 dispels
Player 3 does 5 dispels
Player 3 zones.
Player 3 waits 1 full minute in town and returns to saruta. Player 3 casts 1 dispel and hate is returned to player 3. Hate did not reset by zoning out.
What this proves is that non-decaying hate is not reset upon zone out. It is possible decaying hate is but I don't have a better way to test this. My guess is that hate is actually not reset upon zone out or log but simply the time waiting while logged out gave time for the decaying hate to decay.
We did multiple other little tests with all the same results. We did not however have time to wait longer. Someone should try this same test with a wait of 5 minutes or something and see if it changes the result.
Enmity + Gear/merits:
I believe suwi and I have found out how enmity gear works from a math perspective. I believe now that it is 1 enmity = 0.01 multiplier for hate. So 5 enmity would equal 0.05. I'm sure its probably like 1 enmity = 5/512 but lets keep it simple for now.
Now here is the thing. Everything works out dead on perfect in our tests except for one thing that doesn't seem to make sense. That is when the mob first turns then turns back. You'll see below.
Test 1:
I'll use UH = Units of Hate
Player 1 pulls bumblebee with silence.
Player 2 is wearing 2 enmity.
Player 3 has no enmity gear.
Player 2 casts dispel. Player 2 should have 320UH x 1.02 = 326.4UH (In fact I believe the game ignores decimals see below so this is really 326UH and I do not believe its a rounding function, in test 2 you'll see why)
Here comes the oddity that doesn't make sense.
Player 3 casts dispel. Mob turns to player 3 for a split second then returns to player 2. Player 3 should have 320UH.
If this is all correct then player 3 should have to cast 6 cures in order for the mob to turn back and stay solid on him.
Player 3 casts 6 cure 1's for 0 and now has 326UH and the mobs attention. Now player 2 and 3 can bounce the bee back and forth with Cure 1's showing they now have equal hate.
Test 2:
Player 1 pulls with silence
Player 2 now wearing 4 enmity gear casts dispel. This player now has 320UH x 1.04 = exactly 332.8UH (you'll see when player 3 starts casting cures that this number is 332UH and not rounded up to 333UH)
Player 3 casts dispel. Again the oddity the mob turns for a split second, then returns to player 2.
Player 3 casts exactly 12 cure 1's for 0 and hate shifts to player 3 solidly. Both players sitting at 332UH.
Test 3:
Player 1 pulls with silence
Player 2 is wearing 5 enmity and casts dispel. Therefore has 336UH (320UH x 1.05 = 336).
Player 3 casts dispel, mob turns for split second as seen in the first two tests then instantly returns to player 2. Player 3 currently has 320UH.
Player 3 begins casting Cure 1 for 0 and upon the 16th cure the mob turns exactly as predicted towards player 3 and hate is solid.
So from this testing today I can only conclude that +1 enmity = 0.01 hate added to a hate action.
Note about the oddity: 2 things come to mind. Either
1) that Dispel has a decay of more than 16UH, we can figure that out really fast by stacking on enmity gear and seeing if the mob turns for a split second upon casting the second dispel. I doubt it has any decay but it should be tested at some point.
2) that the game in fact has a delay in calculating +enmity from actions. I didn't wait (I should have) before casting dispel on the 3rd player. So its possible that had I waited 10 seconds or something the enmity would have been calculated and it wouldn't have shifted. Maybe the game is refreshing the enmity calculation every 3 seconds (like a tick of refresh) instead of every second. Who knows, right now it seems odd and I can't explain it.
I can't think of any other reason to explain it considering how completely perfect the numbers work out for +enmity gear.
I like the enmity testing you did. I'll be sure to cite this when I start going into +/- Enmity gear. The 1% multiplier is a nice fit. Also, yes, when you do machine coding, you rarely ever round and you just ignore the decimal place. This has to do with how most fixed point math involving multiplication and division is calculated.
So assuming everything here is correct, the effect of enmity gear is a a 1% effect for CE. It still won't tell you anything about VE though, which is honestly really hard to set up a test for.
We still have more to do, obviously, and will have to recruit other people for various abilities and such that we just don't have access to.
But so far, lotsa stuff... we got some other pretty interesting stuff just last night regarding the exact hard cap numbers, Cure V's hate level, etc. Good times. ;D
He's threatening to make me cast Cure another 300+ times, too.... need to you know, repeat and verify....
And don't cite anything to me besides "crazy bored Mithra who cast Cure a shitton," Kaeko is the one who does all this wonky math. :) I make a witty comment now and then, and I guess I've made a couple statements that made some actual sense.. but otherwise, all Kaeko. ;)
BTW, thanks so much for that tell log pic of the day, Kae... the mistake that will never die, I swear. ._.
-Ash
just an fyi.