Why We Do What We Do (An in depth explanation of EHP and ICC 3.3.3 tanking mechanics)

[MellvarTank]

Is its more of a gear scaling issue? or just how the boss fights work?

It's the culmination of a lot of factors. Gear scaled higher than they had originally thought (way more tiers of gear in this expansion) and the way they designed encounters favored EH. Remember Gormok? That was probably the biggest nudge in the direction of EH, but even Ulduar and Naxx tended to favor big health and armor over avoiding hits on account of healers having more mana than they knew what to do with.

[Itsover]

Probably wrong but sounds like the spell's/ability's dont scale with gear at all. So later on down the road your so over geared for your spells/ability's that it almost doesn't matter what ability's you have you can spam what ever you want with out consequences.

Probably went that direction do to some like rogues in TBC getting 110% avoidance and tanking Illidan.

[felhoof]

The main reason that downranking was fun was that it gave you options, and rewarded you for playing well based on how you used those options. Choices that are meaningful and give rewards based on making the correct choice tends to be something that is fun. This is something that apparently Blizzard agrees with, as they're adding multiple choices for healers again. Note that downranking wasn't like using every single rank of a spell; you'd only have a few ranks you'd use regularly. You had your slow, efficient heal (GH1), a bigger heal that was somewhat efficient (GH4 or 5, I forget) and your big heal (GH...7?).

It was also fun as a priest to have fights where you simply kicked ass - things like CoH spam on Bloodboil were fun in the same way that Gorefiend DPS was fun - it was a measure of raw power.

[Aggathon]

I've added two disclaimers. One to section 4 (the stochastic nature of incoming damage) and one to section 5: the avoidance table.

The section 4 one reads:

***DISCLAIMER***
This following section is somewhat speculative on the details. No one has done an in depth analysis on boss damage output to determine the exact distribution of hits. I hypothesize that it is a gaussian curve here, but it very well could be a uniform distribution. Either way it is distributed the theory remains the same, just the shape of the curve is shifted.

The section 5 one reads:


***DISCLAIMER***
A table like this is NOT how it is ACTUALLY programmed. In programming they run a random variable on the range 0 to 1. The calculations get reduced based on level difference as explained below. This is merely a VISUAL REPRESENTATION that helps me visualize how this all works in a way that is, in my opinon, easier to understand. It is probably easier for me to understand because I've played DND a lot and generally think of things like this in terms of lookup tables, not actual programming random variable decimals. If it helps you to understand it better as a random variable and what it REALLY is, then just know that it's not an ACTUAL table with 10s of thousands of spots, it's really a bunch of decimals on the range 0 to 1 but it is still a sort of lookup table. If you have any more questions feel free to PM me and maybe I'll add more detail to this.

I updated a few other things, the link to satorri's post, etc.


I also changed the bold thing in section 1 about not debating EHP vs. Avoidance in the thread and insted to send me a PM (since well, that already happened) to this:

A side note when reading this: I would really prefer that if it's a specific contention that you read the entire thread because perhaps the question has already be answered/responded to or send me a PM first. If you have good evidence then I may add your concerns to the thread, but there have been a lot of people that come forth with merely assumptions and don't really have any proof to back up what they say, especially in terms of the EHP vs. Avoidance debate. If you do have data to support your claims than I will be more than willing to listen.

[Delmonte]

N.B. this is not an argument that avoidance is as good as EH, I fully advocate stacking EH as the OP describes and appreciate other people's views, this is simply a question about the logic used in one section.

Apologies if this specific point has been raised, I have read through the whole thread but may have accidentally skimmed over this question and maybe you are already planning to edit this as I saw you mentioned concerns about Gaussian curves. I think this is a little bit more specific than what others have posted.

Forgive me if I have misunderstood but the implication of the section seems to be that stamina is 'not RNG' and stats like dodge are. My concern is with this logic. I agree with your conclusion that stamina reduces the chance that you will die on a given melee swing and therefore reduces RNG and is not itself RNG. However, I do not think it follows logically that therefore stamina is not RNG and a stat like dodge is. I think you need to clarify your use of the term RNG somewhat as you seem to have assigned them different meanings when referring to stamina and avoidance depending on whether you are talking about the stat itself or the effect it has on the outcome of a weapon swing. I hope this explains where I am coming from:

I'm going to refrain from using the term 'RNG' as much as possible as it is often used very loosely and taken to mean different things. I think it is confusing to talk about a stat itself as being 'RNG' as the outcome that is random is the result of a melee swing made against you.

One statement you made is that stamina is not rng in the sense that, excluding multipliers, one point of stamina always gives the same amount of HP. However, this is also true of stats like dodge and parry, one point of dodge rating will not give you a random amount of dodge %, it will always give you the same amount.

You correctly argue that the effect of stamina is to reduce the % of weapon swings (assuming you are at a level of health where one swing may or may not do enough damage to kill you), however, this is also true of dodge and parry.

The use of Gaussian curves when representing EH seems reasonable for the purpose of having a graph to look at to illustrate a point but it would also be equally reasonable to draw a representation of dodge outcomes and mark a line on it at the % point at which you will die or survive depending on which side of the line your random swing outcome lies. This line is also moved as your dodge % increases and reduces the % chance that you will die on any given melee swing. The dodge distribution would have a different shape, however, the distribution shape does not affect the result that in both cases increasing EH and dodge will both move the line in your favour by a predictable amount (if you have the raw numbers and know the shape of the hit's damage distribution).

In both cases the stat in question is reducing the % chance (reduce 'RNG' as you put it when referring to stamina) that you will die on any given melee swing. In both cases neither stat in itself is random in that they always give the same predicted increase in stamina or % dodge, however, in both cases the outcome - whether or not your hit will fall within the part of the probability distribution that results in your death is random.

I hope this is not beating a dead horse too much, I noticed you recently reiterate the disclaimer for that section but I believe it remains phrased in such a manner as to be misleading i.e. implying that 'stam is not 'RNG' and avoidance is' follows logically from the curves. I would recommend avoiding phrases like 'stamina is RNG' or 'reduces RNG' if you are trying to use rigorous mathematical logic to analyse something as they are quite vague and could be taken to mean different things.

P.S. I hope this is not coming over as offensive or overly critical, it isn't meant to be at all and I found your post very interesting and thoughtful. It's just my 2 pence from a mathematical point of view, maybe I am wrong anyway! (I'm sure some of my terminology is :P).

[Aggathon]

RNG means, "Random Number Generator," pure and simple.

Avoidance is RNG because a random number is generated and that determines whether you are hit, missed, dodge, parry, or block an attack. The reason I can't draw a Gaussian curve is because the curve for avoidance isn't gaussian. The curve for avoidance is stated in section 6.

Yes, both increase survivability, but in different ways. EHP never generates a random number, Avoidance does, hence, RNG.

[MellvarTank]

One statement you made is that stamina is not rng in the sense that, excluding multipliers, one point of stamina always gives the same amount of HP. However, this is also true of stats like dodge and parry, one point of dodge rating will not give you a random amount of dodge %, it will always give you the same amount.

Stamina does not suffer DR's, however Dodge does. So no, 1 point of dodge will not always give you the same amount.

[Kojiyama]

Actually, it will in many ways. The DR on Dodge primarily prevents it from having increasing returns, despite the nomenclature.

(In a linear scenario each point of Dodge Rating would be worth more than the previous point. The DR formula keeps this from happening.)

[Aggathon]

Well it's increasing returns just like 10% more dodge from 80% dodge to 90% dodge is a significantly better return (since that 10% means you avoid 50% more of the attacks) vs. 10% dodge from 0 to 10% being only 10% flat, for example. There is a point though where you're just not gonna get that increase because the DRs are too high, blah blah, it's pretty academic, whatever. We never really face that scenario currently.

The thing is though that the "stochastic" part goes down to refer to the "final" hit that would kill someone. The entire point is that a boss hit isn't always for the same amount so a boss might hit for 10.2k one attack but then might hit for 9.8k on another and if you get a weak one you might survive.

Avoidance gets WAY more tricky, mainly because the purpose of avoidance (imo) isn't to prevent that last hit in the hit streak from killing you if you get a hit streak, but it is to decrease the odds of that lethal hit streak from happening in the first place by breaking it up and allowing healers to catch up very easily. So with avoidance we're not really talking about the final hit probability because that's easy. You have the same % chance as your avoidance that you will avoid any given swing. It's not tricky. What section 4 is talking about is merely a musing and an explanation of the argument that stamina is RNG because of the nature of incoming damage when really stamina just increases the buffer by moving the probability of a specific hit from killing you further up the curve.

You can't really compare the two in this case because they're referring to two different things, 1 being mitigation (be it from armor or decreasing the margin by stamina) and the other being completely avoiding an attack.

Idk how else to explain it really.

[Delmonte]

As you say 'RNG' stands for random number generator, the phrase 'stamina is not random number generator' does not mean anything. The point is that you cannot say stamina is any less 'RNG' by the definition that you are using than dodge because you have drawn a curve with a line on it that moves up and down with probability because you can do the same thing with dodge. It is also not a binary case of something either being random or not in the way you have looked at it, the outcome of the swings of both stats are dependent on probablitily, the extra % that the stat swings the numbers in your favour is not.

If you are going to say that using your definition of RNG that stamina is not RNG because it reduces the probability that you will die on any given swing, by the same definition, dodge and parry are not RNG because they also reduce the probability that you will die on any given swing.

The curve in section 6 is not the equivalent of your gaussian curves for EHP because it is looking at different scenario - the probability of a string of hits rather than the probability that one given hit will kill you. I agree that the stats cannot be directly compared and serve different purposes, however, you could draw a diagram to illustrate the same point in section 4 for dodge or parry, looking at one hit with a line on which you die on one side and live on the other which moves up and down as your dodge% changes. This is the point, that you cannot say by this logic therefore that 'stamina is not RNG any more than you can with dodge or parry', whether or not one looks like a normal curve and one does not (remember your curve in section 6 is not the required one for this process) does not alter this fact.

In both cases you can say that the result of whether you live or die as a result of increasing either stat is still decided by the outcome of RNG, in both cases you can say that increasing the stat reduces the probability that you will die and therefore if you want to say that this means that RNG is reduced rather than the stat itself being 'RNG'. If you are going to use different definitions for 'RNG' for EHP (that you will or won't survive one hit) and dodge (the chance that a number of hits will occur in a row which prove to be fatal) which are quite different situations, your statement that 'stamina is not rng' in section 4 is meaningless as you have not correctly compared it to dodge in that situation.

I'm not criticising your general advice but you should reread my post more thoroughly and perhaps consider the point I am trying to make about the logic you have used in section 4 and the statement that you have made at the end of it because you have misunderstood what I said, it is perhaps more subtle than you are suggesting. I'm saying that by exactly the same logic you could say that 'dodge is not RNG, it reduces it'. I totally agree that it is not that simple though and different stats are different, you just can't justify section 4 with the logic you have used (if you are going to say that dodge is RNG which wasn't stated specifically but is presumably implied). I'm assuming that that is the point of section 4 as this thread is comparing EHP vs avoidance. You could pretty much interchange the word 'stamina' with 'dodge' in section 4 and an appropriate representation for dodge and reach the same conclusion (within that given context).

Again, I am in no way suggesting that dodge is as good as stam. What I am asking you to do is just to consider the cold logic of what you have tried to do in section 4 and think about the precise nature of the statements that you have made. Forget everything else for now, does it really follow, purely from your arguments in section 4 that 'stamina is not random and dodge is'? I would argue that the answer is 'no' so, while illustrating a good point, section 4 is not a valid proof of this statement.

[MellvarTank]

Actually, it will in many ways. The DR on Dodge primarily prevents it from having increasing returns, despite the nomenclature.

I'm just referring to his post in that 1 point of dodge will give you %dodge increase always. This part specifically:

one point of dodge rating will not give you a random amount of dodge %, it will always give you the same amount.

From a numbers point of view (X dodge rating = X% dodge always) it won't.

From the "value" point of view (ie. what Agg and Koji said.) it will.

[MellvarTank]

If you are going to use different definitions for 'RNG' for EHP (that you will or won't survive one hit) and dodge (the chance that a number of hits will occur in a row which prove to be fatal) which are quite different situations, your statement that 'stamina is not rng' in section 4 is meaningless as you have not correctly compared it to dodge in that situation.

This statement is confusing. RNG (being a random number generator) is applicable in that stam does not have any "RNG" mechanic, however dodge does. This is shown in the combat table.

You are reading the comments that someone made and the discussion in section 4 a different way than I did. Section 4 is based on disproving that stam is RNG, which someone else stated. A boss hit being variable damage is, in fact, RNG. The definition Agg has used has not changed in any section throughout the document. It is, and will always, mean "Random Number Generator".

[Delmonte]

Section 4 could also be used to disprove that dodge is RNG, you are not talking about RNG in the same sense for stam in section 4 as you are for dodge in section 6. There is an RNG component in section 4 which is about stamina, namely the probability that you will survive the attack, this has parallels in the hit table which also contains probability that can determine whether or not you survive an attack.

My understanding of section 4 is that it is discussing stamina within the context of 'stamina versus avoidance', if that is not the case then what is the value of the conclusion in section 4 if no similar conclusion is being made about avoidance or any other facet of tank survival?

Look, you simply can't say 'stamina is Random number generator' or 'stamina is not random number generator', you cannot say 'stamina has no RNG mechanic' without defining what you mean by this. There is a statement in section 4 'stamina reduces RNG', which I have taken to mean 'stamina reduces the probabality that you will die from a given melee attack' this is an 'RNG mechanic' if you want to call it that and in the same sense you can say that 'dodge reduces the probability that you will die from a given melee attack'.

People keep making statements like:
'Stamina is not random because the stat itself it does not give you a random amount of health, even though the outcome of the swing can be random'

juxtaposed with:
'Dodge is random because the outcome of the swing is random, even though the stat itself does not give you a random % to dodge'.

[MellvarTank]

Section 4 is talking about the incoming damage from a boss, and whether or not that makes stamina RNG. The reality is, it does not. The hits incoming with variable damage may be RNG, but not your health. That is the point. It is presenting an argument that someone made, and disproving it. He is talking about why you use stamina to reduce the chance of dying on a hit, and it can reliably do that.

I think (reading BruisedOoze's post) the biggest problem is looking at the point the right way. Avoidance is based on an RNG mechanic, Stamina is not. If there is variable damage from a boss hit, that is based on an RNG mechanic. The whole thing is arguing wording, which is so convoluted that a certain amount of understanding must first be had by the reader. This topic is not for someone who doesn't already understand the proper use of the word, and trying to make an "EH for dummies" just won't work.

[BruisedOoze]

Stamina does not suffer DR's, however Dodge does. So no, 1 point of dodge will not always give you the same amount.

Yes but that point will never give a random avoidance number either which is the bigger point. If you know how much avoidance your character has you know the %.

RNG means, "Random Number Generator," pure and simple.

Avoidance is RNG because a random number is generated and that determines whether you are hit, missed, dodge, parry, or block an attack. The reason I can't draw a Gaussian curve is because the curve for avoidance isn't gaussian. The curve for avoidance is stated in section 6.

Yes, both increase survivability, but in different ways. EHP never generates a random number, Avoidance does, hence, RNG.

His issue is a point of the language being sound. For Example. You explicitly make the statement that if stamina were 'RNG' then it would give you X or Y hit points. Which is not true. Delmonte then contests that he can then equally well make the statement, if Avoidance were 'RNG' then it would you X or Y% chance to be missed. But that would imply Avoidance gives a non-deterministic % chance to be missed. Which is not true either. So following this, if Stamina is not RNG, then Avoidance can't be either.

More precisely what's happening is that Avoidance is directly affecting a result produced by the hit(and it's variants)/miss RNG. Your EHP then determines the outcome after the RNG, which you appropriately assert is the biggest factor in determining your survivability.

The misuse of the term RNG results in people making the inconcise statement that bothered you initially. "Stamina is RNG." A statement like that likely came from someone thinking "Stamina effects the results of a random event, so Stamina must be random." Which you then correctly call them out on (ie if it was random then Stamina would give you a random amount of hp). But now the contention is that the same misuse is being used in the statement "Avoidance is RNG." It's an issue of clarity.

[Delmonte]

Yeah what BruisedOoze is saying, the point is then that if you are going to use a logical system to compare 2 things, the language cannot mean 2 different things depending on the stat being discussed. I would avoid the term because a random number generator is a device used for generating random numbers, neither stamina nor dodge are devices or programs that do this and people use the term to mean both 'probability' and whether or not something has an element of probability in the way it functions which leads to confusion. You also need to make the distinction in models between steps of the process that involve probability and those which do not. You are talking about 2 different steps of the process when you are talking about stam and dodge being random or not.

[Aggathon]

That post isn't comparing dodge to stam though. The words dodge and avoidance aren't even in that section.

There are several "random" events that occur in a fight. You avoiding or not avoiding an attack is one of them.

Another is how hard a boss hits when he does hit. Stamina decreases this mattering and can push you into the "okay I can survive one more hit" category or even make it so that just 1 small heal pushes you there. All section four is talking about is the stochastic nature of incoming damage and stams relationship to it, it is NOTHING to do with avoidance and I never compare it in such a way. I think you're drawing a conclusion that I simply don't make.

However, if you want to make the argument that avoidance pushes the curve one way or another then the answer is, "well kinda." Over a very large sample size your avoidance % will match what's on your paper doll, but go back and look at your logs. How many times is your exact % avoidance the exact same as on your character? The game doesn't go "oh well he got hit the last 3 times since he has a 40% chance to avoid I guess I'll let him avoid the next 2," no, not at all. Probability does NOT work that way so even though I agree that increasing dodge percent still increases the probability that you'll avoid something (dur) it doesn't mean you will if you've already been hit X times.

Avoidance is still a dice roll where stamina is not, it may be a marginal increase to a small part that only decreases the probability of the stochastic nature of damage and heals, but it is not in and of itself a dice roll. If you have 1000 more hit points you have 1000 more hit points. If you have 1% more dodge you have 1% more chance that you'll not get hit but it's still a marginal chance where as the small increase in stam is a marginal certainty. What's uncertain is how much you'll get hit for or how much you'll be healed. You have no control over that.

[Delmonte]

If you are not comparing stam to anything else in section 4 then what's the purpose of the section? Your thread is about why stam is preferable to avoidance, you are not refuting that the same is true of dodge and parry. Perhaps we are focusing too much on the purpose of section 4 because I think that it does illustrate the benefits of stamina well but it is written in the context of a thread comparing avoidance with stamina and saying that one of these is not random and therefore good. If it is not your intent to imply that it follows, purely from the logic in this section that is less random than dodge or parry, I would clarify this. I even acknowledged in one of my posts that you do not directly state this so I am not sure why you are pointing this out, I merely stated that I assumed this was what you were implying.

All you have demonstrated is that stamina may reduce the probability that you will die on a given swing made against you. How can you not see that this part of your argument involves chance? Dodge and parry also reduce the chance that you will die on any given swing made against you. You are comparing the step of the process in which you find out what your additional % chance to survive is in stamina which is not random (and neither is it with dodge) to a step where you ask the question 'will I survive this melee swing?' in dodge which is random (and you are saying that this step is also random with stamina with your curves as the boss hits for random amounts).

In both cases the % increase in chance to survive a melee swing is a certainty and in both cases answer to the question 'will I survive this swing' is a chance.

If you have 1% more dodge it is a certainty that you will have 1% more dodge not a chance, it is not based on any dice roll or random number generated by the server in exactly the same way that if you have 1000 more HP it is a certainty that you will have 1000 more HP. These are just trivial statements that are true for both stats and in no way prove that stamina is less random than dodge.

In your Gaussian curves, the game also does not go 'oh I hit him for 94% of my average damage so I will even it out on the next swing by hitting softer'. This is still based on chance and may not average out over the course of a fight. There is a chance that 1000 more hit points will not save you depending on how hard the boss hits, similarly there is a chance that your 1% dodge will not save you. It is not the case that one of these chances is random and one is not.

Every single argument you are making can be applied to dodge, you are just choosing to look at different stages of the effect of both stats as it suits you. You seem also to have forgotten that while RNG results in dodge are independent events as you correctly state, so too are the RNG results that determine how hard the boss will hit you on each swing. The flaw in the logic is not something that will bring EHP crashing down and make it unreliable so do not worry. Please, before you post again, read my posts very carefully and think about it. Are you really talking about the same thing every time you compare randomness in stam to dodge? The answer is 'no'.

Comparing the benefits of both stats in terms of survivability is much more complicated than this anyway as you well know.

[Aggathon]

Mellvar already explained what the original purpose in section 4 is, but here's the quote from section 4:

Someone was talking on the forums about the stochastic nature of incoming damage (that is the randomness of boss hits), and was using it to say that stamina is RNG.

Obviously increase your % chance to avoid increases your % chance to avoid a hit. That's like saying tautologies are tautological. Originally someone was saying that stamina was just as RNG as dodge but when in fact it is not, the incoming damage is RNG and stam/EHP helps even that out.

What avoidance does (or at least what I contend it is best at other than the obvious; as I've stated many many times) is decrease the probability of a lethal hit string from occuring. If you are relying on a % chance of the final hit every time from killing you or not then you're doing it wrong (which you say you're not so I don't even get why you're arguing this, at BEST it becomes a semantics debate and is really just an academic argument). Like... the argument here is that if you're going to die in the next hit and you're increasing your chance to avoid it the only way to push it over into 100% survivability is to push your avoidance to 100%, whereas if you're stacking stam you just have to make the marginal jump to the next hit, which is way easier and way more feasible to do.

Also: refer to my statement in my last post that 40% chance to avoid does NOT mean you'll avoid 40% of all attacks in a small sample size even one as large as a single boss fight (which is still relatively small, imo), which is another reason that avoidance is "RNG."

I even say in section 4:

Stam isn’t RNG, it diminishes RNG. Stacking stam is a way to decrease variability in a boss fight!

Anything else I say would just be re-hasing other EHP vs. Aviodance debates.

And yes, I thought long and hard about what you were saying, I re-read your paragraphs several times.

As always, I never say avoidance is bad, far from it, avoidance is very good. However in ICC based on current mechanics sacrificing EHP for avoidance is suboptimal.

Edit: whoops, mixed up my last sentence.

[Delmonte]

Obviously increase your % chance to avoid increases your % chance to avoid a hit. That's like saying tautologies are tautological.

TIf you have 1000 more hit points you have 1000 more hit points.... ....the small increase in stam is a marginal certainty

Uhuh....

All section four is talking about is the stochastic nature of incoming damage and stams relationship to it, it is NOTHING to do with avoidance and I never compare it in such a way. I think you're drawing a conclusion that I simply don't make.

Originally someone was saying that stamina was just as RNG as dodge but when in fact it is not, the incoming damage is RNG and stam/EHP helps even that out.

Right, so, section 4 is not a comparison of dodge and stamina in any way but at the same time, it answers the utterly vague question 'is stamina is less random than dodge?'.....

...using a logical process where the words 'stamina' and 'dodge' can essentially be interchanged.

You do not directly respond to any point I make and have not been able to refute any of the actual statements I have made as to why your logic is incorrect.