Elitist Jerks WotLK Healing Compendium v3.2: Same Old Thing

05/16/09, 7:00 PM   #301
Biffins
Glass Joe

Dwarf Priest

Bronzebeard
 Originally Posted by Hegen Now, to get back to the main point: if you are certain that you can frequently save lives by making flash heals a few percent larger, then it's irrelevant how much more statistical healing BR provides
I am sorry, but this is just a wrong statement in so many ways.

I just showed that Flash Heal with 3/5 EH, 0/3 BR will be 5232 and Flash Heal with 0/5 EH, 3/3 BR will be 5047. That means that chosing to put points in EH over BR will make your Flash Heal 5232/5047 = 3.6%.

Are you seriously telling me that ANYONE can be "certain" that they save lives from a 3.6% stronger flash heal??! How? Just by intuitive sense? How can it be possible to feel that a 3.6% stronger flash heal helps save more lives? How can you not analyze everything analytically and mathematically to determine the highest healing output for your chosen playstyle.

This is not about whether you like to use renew or not, whether you like to use Gheal or not, whether you like to have SoL procs or not.......... the point is no matter what your chosen playstyle is there must be some spec that maximizes your healing output with that chosen playstyle. One spec MUST be better than another. Period. Whatever way you decide to play your priest, there MUST be some spec that is the best for it. And I am providing math to back up that analysis.

Even if your role is to do a LOT of Flash Heals, my claim is that Fheal is not going to make up 45% of your total healing and hence you will be better off putting points in BR instead. You will still save as many lives with a ~3% weaker flash heal and your overall healing output will be higher. Can anyone dispute this?

Last edited by Biffins : 05/16/09 at 7:12 PM.

05/16/09, 7:05 PM   #302
Biffins
Glass Joe

Dwarf Priest

Bronzebeard
 Originally Posted by Hegen Also, you do not factor in individual overheal for each spell - this can shift the numbers quite a bit. If 50% of your flash heals overheal (which is not the same as 50% overheal, of course), and only 25% of your PoH or CoH hits overheal, then your break even point will be different.
Can you please elaborate on this? I don't understand. How is this relevant?

05/17/09, 3:42 AM   #303
Jamora
Glass Joe

Dwarf Priest

Die Aldor (EU)
 Originally Posted by Biffins Can you please elaborate on this? I don't understand. How is this relevant?
Basically, when you overheal the gain from the talents is lost anyway. So in reality on those heals which do not overheal benefit from the extra healing of the talents.

And, as others have said, it is not a healers job to output the most HPS, it is a healers job to keep people alive. For example, EH is simply superior to heal sag pot victims on Ignis. Sure, other healers might be better suited to do so, but perhaps the paladin healer is a 55 year old guy who is nice and all, but his reflexes are not that great any more. If you use FH to cover this, then EH might be better even if you do more group healing than single target healing.

On a more general note, this debate gave me the idea to write a simulation for this. Given a damage pattern and healing pattern, what is the survival rate of individual people. That way it would be possible to measure how often different builds/spell choices decide between life and death. It would not be so terribly difficult, but I fear that I will be to lazy so actually do so. Still I like the idea because it would quantify the "feel" we healers often express.

05/17/09, 6:24 AM   #304
eliott
Glass Joe

Human Priest

Sylvanas (EU)
 Originally Posted by Jamora And, as others have said, it is not a healers job to output the most HPS, it is a healers job to keep people alive. For example, EH is simply superior to heal sag pot victims on Ignis. Sure, other healers might be better suited to do so, but perhaps the paladin healer is a 55 year old guy who is nice and all, but his reflexes are not that great any more. If you use FH to cover this, then EH might be better even if you do more group healing than single target healing. .
I do not agree with this statement: Even given that your healing mates have slow reflexes or whatever, and you are called to shoot around a couple of unnecessary FH in order to "cover" his target, the difference between life and death of a single person is really hardly made by a 3% stronger FHeal; Or, at least, in my ulduar experience just few times i saw people getting overkilled for less than 1k dmg after someone landed a direct heal on them; If they take damage twice is often due to their own error and really hardly a single heal of a single healer can save them; For instance, a situation where EH should win is mimiron p1 on napalm shells; but in this given situation most likely you will be supported by 3 other healers, and a slight improvement of your FH won't surelly make the difference in 99% of the cases, while that 3% more output will be really temptating during p2 and p4, even more if you are doin hardmode and you might run without a healer to match up the dps.
Of course there will be a situation where healing for 500 more would have saved a life, but it is irrational to downgrade your performance just for a not-so-much-likely-to-happen case;
Also because i see as a priest FH isn't the best situation often to save a life: generally i prefer to throw a shield on a single person who took a heavy spike (if there are no disc priests in raid ofc, but that's usually me so..) to allow him to run out of the aoe if damage is due to that and give other healers a bit more time to save him, but that's personal feelings.

05/17/09, 6:33 AM   #305
Elimbras
Don Flamenco

Dwarf Priest

Eitrigg (EU)
 Originally Posted by Jamora On a more general note, this debate gave me the idea to write a simulation for this. Given a damage pattern and healing pattern, what is the survival rate of individual people. That way it would be possible to measure how often different builds/spell choices decide between life and death. It would not be so terribly difficult, but I fear that I will be to lazy so actually do so. Still I like the idea because it would quantify the "feel" we healers often express.
This in actually the right approach. And depending on your damage and healing pattern, it may be mathematical tractable, in which case you even don't need any simulator.
However, you'll quickly find a limit of this approach : you need a damage (and healing) pattern, and results will be highly correlated to these patterns. How can you choose them accurately ? More importantly, damage pattern is bound to be valid only for one (or a very few) bosses. That's especially true for tank healing (where the frequency and size of hits on the tank is crucial), but that's also true for raid healing.

 Originally Posted by Biffins I am sorry, but this is just a wrong statement in so many ways. [...] Are you seriously telling me that ANYONE can be "certain" that they save lives from a 3.6% stronger flash heal??! How? Just by intuitive sense? How can it be possible to feel that a 3.6% stronger flash heal helps save more lives? How can you not analyze everything analytically and mathematically to determine the highest healing output for your chosen playstyle. This is not about whether you like to use renew or not, whether you like to use Gheal or not, whether you like to have SoL procs or not.......... the point is no matter what your chosen playstyle is there must be some spec that maximizes your healing output with that chosen playstyle. One spec MUST be better than another. Period. Whatever way you decide to play your priest, there MUST be some spec that is the best for it. And I am providing math to back up that analysis.
First of all, you can have "more than reasonnable" reason of this claim. Because you have enough AOE healing power, and never loose anyone during aoe (when you raid with four healing priest and 3 shamans, for example ;-) ). Or more likely, nearly never, compared to RST situations.

Or because you know that you use flash (or bind) heal when your target is low, and POH / COH only as a "top up" spells. Basically, you need to realize that not every point of healing has the same value. One point of healing is more important when it prevents immediate death, or heal a very low person, than when it put's your DK from 27k999 HP to 28k HP...

Ultimately, I perfectly agree with your conclusion. For anyone, there must be a spec that maximizes healing output. And they will be a spec that minimizes the death probability. The key point is that they are not necessarely the same, and computing mathematically one won't therefore give you the other (except of obvious points, like take meditation, SOR, COH...). Unless you prefer to play kikimeters, but that's another game.

Another way to look at this difference, which dps nearly never have : we play with finite horizon, when dps plays almost with nearly infinite horizon.
Basically, our horizon is the life of a person. With 20/ 30k HP, and the present damages in raid, players can die really quickly. On the contrario, dps tries to kill a boss, which last a few minutes, often more than 4. We all know that some specs are better on trash than others, but equivalent on bosses. The difference is really the fight lenght. When you need 15s to start your real dps, you're in bad position for trashes. For bosses, it doesn't really matters, that will be 4% of the fight length (for a 6 min fight). As long as healing is considered, we don't even speak of 15s time length, but more likely 5s.
If we had players with 300k hp, ie. if we have a long time for healing, then maximizing hps is equivalent to minimize death probability. That's the case when healing Razziavious adds, for example. But not for most bosses. The frontier conditions are not negligeable.

 05/17/09, 7:07 AM #306 The Not So Evil Piston Honda     Brundtland Night Elf Priest   Trollbane (EU) And we're starting to discuss healing granularity. This is not useful. - Some classes have more hp than others. - Some classes have talents that reduce magic & physical damage. - Gear can change physical & magic damage taken. - Some raids have another healer setup than yours. - Some players stand in fire. You CANNOT start talking about healing granularity except for the specific raid setup and fight you are working on. And even then it comes down to your own playstyle. Discussing this will just lead to 500 different viewpoints, most being correct, and it will just confuse the hell out of the people who don't already understand it. I'll try to put it as simple as I can once more: If entire raid is going down because they are too hard to hold up, you might want to take Blessed Resilience for the extra throughput. If single targets are going down but raid is being kept up fine, you might want to take those extra points in Emp. Healing and Test of Faith. Or just go Disc. Rawr - Coder of HolyPriest (Healer) and ShadowPriest (DPS) Modules. Get Your Rawr 2.3.x!
05/17/09, 8:50 AM   #307
Promethia
Piston Honda

Blood Elf Priest

Kilrogg
 Originally Posted by Biffins the point is no matter what your chosen playstyle is there must be some spec that maximizes your healing output with that chosen playstyle. One spec MUST be better than another. Period. Whatever way you decide to play your priest, there MUST be some spec that is the best for it. And I am providing math to back up that analysis.
No. Your calculations may be "correct", but they are not really solving the problem.

You are assuming that "maximizing your healing output" is the goal, when it is not. Instead, we are interested in minimizing deaths. Until you provide some quantitative model for just how much each of these talents leads to a reduction in deaths, you really have not provided any real mathematical analysis of the problem. Sorry.

 Originally Posted by Elimbras This in actually the right approach. And depending on your damage and healing pattern, it may be mathematical tractable, in which case you even don't need any simulator. However, you'll quickly find a limit of this approach : you need a damage (and healing) pattern, and results will be highly correlated to these patterns. How can you choose them accurately ? More importantly, damage pattern is bound to be valid only for one (or a very few) bosses. That's especially true for tank healing (where the frequency and size of hits on the tank is crucial), but that's also true for raid healing.
Yeah, exactly. Simulators can be very useful, but they are still models and have limitations. All simulations make assumptions, and when those assumption do not apply, neither do the conclusions. Some people seem to assume that numbers produced by simulators are inherently more accurate than others, and that is not necessarily so.

05/17/09, 8:58 AM   #308
The Not So Evil
Piston Honda

Night Elf Priest

Trollbane (EU)
 Originally Posted by Promethia Yeah, exactly. Simulators can be very useful, but they are still models and have limitations. All simulations make assumptions, and when those assumption do not apply, neither do the conclusions. Some people seem to assume that numbers produced by simulators are inherently more accurate than others, and that is not necessarily so.
This is why I have implemented a more advanced calculator into the Rawr HolyPriest module (Custom Rotation). This lets you ask Rawr what it thinks given an EXISTING fight you had to deal with, with your raid variables. I have used this quite extensively myself, and it has really made me change some things around.

Rawr - Coder of HolyPriest (Healer) and ShadowPriest (DPS) Modules.

05/17/09, 9:48 AM   #309
Hegen
In gear/DCT lock pin

Human Priest

Alleria (EU)
 Originally Posted by Biffins I am sorry, but this is just a wrong statement in so many ways. I just showed that Flash Heal with 3/5 EH, 0/3 BR will be 5232 and Flash Heal with 0/5 EH, 3/3 BR will be 5047. That means that chosing to put points in EH over BR will make your Flash Heal 5232/5047 = 3.6%. Are you seriously telling me that ANYONE can be "certain" that they save lives from a 3.6% stronger flash heal??! How? Just by intuitive sense? How can it be possible to feel that a 3.6% stronger flash heal helps save more lives? How can you not analyze everything analytically and mathematically to determine the highest healing output for your chosen playstyle.
Please read carefully. I do not claim it to be so. I simply state that your conclusion is not valid. I am not sure I would make the claim that slightly larger FHs save more lives. But perhaps they do for them. Each raid is different. Perhaps they do for me, and I just don't know. Havoc12 hasn't backed up his claim by providing combat logs, true. But you haven't proven the contrary either. Personally, I think it's absolutely possible to provide combat logs to prove either point, and even for same boss fight.

The point here is that both ways of thinking are valid. Rating some spells as more important (read: life saving) than others is a totally valid methodology. You try to take that into account by looking at the amount healed - but that's not the whole story, and that's why your analysis won't provide an answer to the issue at hand.

Your question regarding the overheal has already been answered, but let me state that it's really the number of targets hit that you would have to analyse, not the amount overhealed, which makes it pretty difficult with current tools. Anyway, if you were to believe you in your methodology, you would need to take this into account. It think it's a waste of time, though.

We're really deep in simulator territory here, a flat model won't help anybody.

Last edited by Hegen : 05/19/09 at 5:10 AM. Reason: Typo

 Originally Posted by Jeremy Clarkson The simple fact is this. We are told to concentrate more. But we can only do that if we are allowed to go considerably faster.

 05/17/09, 3:22 PM #310 Havoc12 King Hippo   Shaarra Night Elf Priest   Silvermoon (EU) Just to clarify my position 1) BR will clearly have bigger return than empowered healing in the almost all cases (unless you have to support a tank healer or take over from a dead tank healer) 2) This return is in the order of 1-2% more healing. 3) Small increases in single target healing will not frequently save target lives, but they will save them occasionally. Very small increases in aoe healing, can save lives extremely rarely. I justify this for the following reasons a) The lower the amount of healing the lower the chance that it is significant and as you get below 5% of a DPS health the chance is no longer linear, as the ways you can get to such low health get reduced by the possibility of death (thus lowering the probability that under healing will kill). b) DMG abilities which are meant to be healed by single target heals are generally designed to match or exceed single target HPS to make them challenging. Abilities which are designed to be tackled by aoe heals generally are well below HPS output of benchmark aoe spells ---> With single target heals the size of the heal is critical with aoe heals, hitting all the assigned targets is critical. c) Single target healing often requires multiple repeats per occasion, aoe healing requires few repeats per occasion. Thus maxing BR at the cost of empowered healing will increase your healing by a very small margin. Maxing EH instead of BR will increase the rate at which you save lives by a very small margin. The amounted added by BR is significantly less than the normal variation in the amount our aoe heals hit for. Incidentally can anyone confirm with data from PoM that BR stacks multiplicatively and not additively? =========================================== On a side note I only have one example I clearly remember where EH has saved my life. I had to use binding heal to save me and another of my assigned targets who were low after heatwave and I got rapid burst in the middle of the heal. I hit myself with bh twice and after rapid burst I had 500 HP. Without empowered healing I would be dead. To post positive proof one would need to scan death logs and find cases where underhealing by fh, bh or gheal has lead to deaths and see what is the smallest deficit one can find. Last edited by Havoc12 : 05/17/09 at 3:27 PM.
 05/18/09, 1:38 AM #311 BobTurkey Glass Joe   Turkelife Blood Elf Priest   Nagrand GJ constantius Very nice OP constantius. Useful information for priests. Your Int/Spi numbers look spot on to me. I disagree that Spell Warding is a useless talent. I prefer it to Divine Fury for my Holy spec. If i'm not casting Gheal why retain Divine Fury? Lightwell is a strange one. Those raid teams that know how to use it get a lot of benefit from that 1 point. Unfortunantly most players do not even realise what to do with the 'lolwell'. Renew or not to Renew seems to be a valid and personal preference decision, which adds good variety to cookie cutter builds. Body and Soul looks a waste to me, but I don't have a lot of Uldar expereince to comment on. It would be excellent lower in the tree so Disc could get it, but as Holy I can't yet see the benefit being worth it. You are right that in 3.1 Disc priests have less variability in their good builds than Holy. Gobble gobble.
05/18/09, 4:12 AM   #312
Alucardu
Glass Joe

Trollbane (EU)
 Originally Posted by BobTurkey Very nice OP constantius. Useful information for priests. Your Int/Spi numbers look spot on to me. I disagree that Spell Warding is a useless talent. I prefer it to Divine Fury for my Holy spec. If i'm not casting Gheal why retain Divine Fury? Lightwell is a strange one. Those raid teams that know how to use it get a lot of benefit from that 1 point. Unfortunantly most players do not even realise what to do with the 'lolwell'. Renew or not to Renew seems to be a valid and personal preference decision, which adds good variety to cookie cutter builds. Body and Soul looks a waste to me, but I don't have a lot of Uldar expereince to comment on. It would be excellent lower in the tree so Disc could get it, but as Holy I can't yet see the benefit being worth it. You are right that in 3.1 Disc priests have less variability in their good builds than Holy. Gobble gobble.
You might B&S isnt that awesome. But it is. Period. Get it, love it.
Lightwell, its a good talent for only 1 point. I put it at ppl who dont need to move and tank, warlock tank on mimiron. Off tank on Iron Council.

05/18/09, 11:04 AM   #313
tedv
Observation: I am awesome

Goblin Priest

Mal'Ganis
While there isn't a 100% correlation between total healing done and deaths prevented, the two are very closely related. We should take it as axiomatic that increasing the total amount healing is the better choice unless proven otherwise. In other words, if Havoc12 claims that Blessed Resilience is around 1.5% more healing than Empowered Healing but Empowered Healing is still better for preventing deaths, the burden of proof is squarely on his shoulders. There is no need for us to justify the benefit of 1.5% extra healing.

That said, let me address his argument.

 Originally Posted by Havoc12 3) Small increases in single target healing will not frequently save target lives, but they will save them occasionally. Very small increases in aoe healing, can save lives extremely rarely. I justify this for the following reasons a) The lower the amount of healing the lower the chance that it is significant and as you get below 5% of a DPS health the chance is no longer linear, as the ways you can get to such low health get reduced by the possibility of death (thus lowering the probability that under healing will kill). b) DMG abilities which are meant to be healed by single target heals are generally designed to match or exceed single target HPS to make them challenging. Abilities which are designed to be tackled by aoe heals generally are well below HPS output of benchmark aoe spells ---> With single target heals the size of the heal is critical with aoe heals, hitting all the assigned targets is critical. c) Single target healing often requires multiple repeats per occasion, aoe healing requires few repeats per occasion.
Argument a) is based on granularity, and The Not So Evil gave a good explanation for why granularity is a meaningless point in these discussions. The basic argument is "Flash Heal is better on really hurt targets than Circle of Healing, and the really hurt target is in more danger of death." But the counter argument is that the greater healing throughput of group heals could have prevented the target from even getting in that situation in the first place. That enables you to use another round of efficient group heals to keep them up, and other people as well. Rather than granularity we should focus on total throughput.

I don't think argument b) matters much. I don't believe Blizzard designs the encounters "intending" for us to use specific heals in most situations. Their model seems to be, "toss out a bunch of damage and let the healers figure out the optimal way to deal with it". But even if there's a situation where one person takes 15k to the face, you'll still top of the raid faster if you have 3 healers bounce group heals off that healer than if they all cast their version of flash heal on the target. Assume other people in the raid can use the healing from circle of healing, chain heal, beacon of light, and so on.

I confess I don't understand what point you're making with argument c). What is a "repeat per occasion"?

05/18/09, 12:11 PM   #314
Elimbras
Don Flamenco

Dwarf Priest

Eitrigg (EU)
 Originally Posted by tedv While there isn't a 100% correlation between total healing done and deaths prevented, the two are very closely related. We should take it as axiomatic that increasing the total amount healing is the better choice unless proven otherwise. In other words, if Havoc12 claims that Blessed Resilience is around 1.5% more healing than Empowered Healing but Empowered Healing is still better for preventing deaths, the burden of proof is squarely on his shoulders. There is no need for us to justify the benefit of 1.5% extra healing. Argument a) is based on granularity, and The Not So Evil gave a good explanation for why granularity is a meaningless point in these discussions. The basic argument is "Flash Heal is better on really hurt targets than Circle of Healing, and the really hurt target is in more danger of death." But the counter argument is that the greater healing throughput of group heals could have prevented the target from even getting in that situation in the first place. That enables you to use another round of efficient group heals to keep them up, and other people as well. Rather than granularity we should focus on total throughput.
I really don't buy your idea of maximizing healing output axiomatically. We both agree that there is positive correlation between both and no equivalence. But this mean that small increases for one metric are not necessarily increases for the other. Big increases will be : you want to take the 10% more healing talents. But small increases, especially when they impact different spells, are another case. You suggest to stick to maximizing healing output, because it is mathematically tractable. But solving and sticking "blindly" to an approximative tractable model has never been a good idea of science. The model is useful, but only to indicate trends. Not little details / variation. Especially, you need to know when your model is nearly exact, and when it is only gross approximation. Therefore, you can't let the proof of better death mitigation on Havoc shoulder. You should also proof that 1.5% more healing output, with blessed resilience and w/o empowered healing, is better death mitigation. You're basically assuming equivalence where you know there is no equivalence.

About the second point, The Not So Evil doesn't say that granularity is a meaningless point. He just says that healing granularity is not mathematically tractable, because it depends on a lot of variables. But it plays a role, vital, and we can't ignore it. Read carefully his conclusion :
 If entire raid is going down because they are too hard to hold up, you might want to take Blessed Resilience for the extra throughput. If single targets are going down but raid is being kept up fine, you might want to take those extra points in Emp. Healing and Test of Faith. Or just go Disc.
Basically, there are cases when Emp. Healing is better, and cases when BR is better. Do as your experience and feeling tells you.

 05/18/09, 3:20 PM #315 Sgat8516 Von Kaiser   Victomizer Night Elf Priest   Garona I'm kinda curious as to what simulator's yall are using to come up with these numbers? And regardless of how exact most of them are, isn't it kinda a matter of semantic's at this point? Going back through a couple WMO reports, and using RAWR to estimate the actual values, EH is more closer to 10% healing lost on FH/GH/BH (although BH not used as much), which is after taking BR (IE if you took neither, it would be a 13% loss). So on a WMO report for the night of all bosses (+attempts, short night in ulduar, First 4, + Kolo, and Sarth+3d, Maly): CoH = 3,426,460 FH = 2,788,577 PoM = 2,390,309 Which = 8,605,346 (which those 3 total about 84% of my total healing) Had I of taken BR, instead of EH, assuming I did exactly the same behavior in those fights: CoH = 3,529,538 FH = 2,498,565 (Rawr says its a 10.4% loss) PoM = 2,462,018 Which = 8,490,121 (and lets say all of the other 16% healing were spells benefiting from BR, so 16% x 3%+ = 0.4% bonus to those) Which is a 1% (1.4% from 3 main, -4% gain from the other 16%) loss of overall healing. We could do the same math, on a full night worth of Thorim Hard mode attempts, top 3 abilities = 79% of total healing: CoH = 6,357,611 / 6,548,339 FH (23% btw) = 4,238,631/ 3,797,813 PoM = 3,669,084 / 3,779,156 14,265,326 vs 14,125,309 = 0.98% difference And we'll say (there was a couple of Gheal's/Bindings, but less than 1%) all of the other 21% were abilities that gaining a bonus from BR: (21% + 3%) 18,057,374 Total / 18,031,118 (total, assuming the other 21% got the 3% bonus)= 0.1% difference But what real difference does that make? Does it change gameplay style? Probably. Does it drastically effect healing potential? Not really Would taking BR over EH, drastically reduce your effectiveness when having to single target heal? yes Would it be game breaking, or wipe causing? Probably not. IMO, the priest class is the best healing class, because of our flexibility. No other healing class can do the variety of tasks, on the level that priests can. Priest's can adapt to any boss fight, and there's never any consequence to bring more of them, over another healer (you can't say that, about trying to bring 5 holy paladin's to a heavy AoE fight). Personally, I'll stay with EH 5/5 BR 0/3, just because it fits into the gameplay style that I've always used. IMO, just take what benefit's your gameplay the most.

 Elitist Jerks WotLK Healing Compendium v3.2: Same Old Thing