Enmity Testing (Part II)
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.
____________________________________________________________________________________________________
____________________________________________________________________________________________________
Test 5A
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)
____________________________________________________________________________________________________
____________________________________________________________________________________________________
Test 6A
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
____________________________________________________________________________________________________
____________________________________________________________________________________________________
Test 7
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.
____________________________________________________________________________________________________
____________________________________________________________________________________________________
Test 8
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=posts&q=161838
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.
____________________________________________________________________________________________________
____________________________________________________________________________________________________