[MoP] Protection Paladin Guide

[lakhesis]

Interesting stuff, it's all linking back into the second condition I was trying to get at with a gap between hitcapped + no expertise and hitcapped + hardcap expertise.

I hadn't appreciated that there was a variation of the superior stat along the way. I could see that the two would work off each other, but I assumed one (expertise) would trump at all points due to its value for the shortest CD generator ability.

I very much agree with the reliability argument personally. I enter into semi-risky situations banking entirely on the fact that i've got 5+1 holy power to play with, and I can think of a lot of situations where a dodge/parry on that +1 would've been horrible. It's a bit like the DP versus HA/SW talent argument tho, I know what I firmly believe is the better approach but I could see why some might disagree.

I've got somewhat the same-but-opposite reaction to Firefly on the crossover points too. When you take into account a haste raidbuff, I have difficulty imagining a haste stacker who wouldn't have so much haste that they'd be needing to push towards the expertise cap to keep up. I'm running 25.58% haste unbuffed with a stamina trinket in currently (second trinket is a haste trinket). You'd have to be very low on haste not to be breaking 21% raid buffed as a haste stacker imo.

And to cover what Nairobi was asking, that's mentioned in the Matlab thread that got linked earlier. From softcap -> hardcap expertise, haste pretty unquestionably wins the higher dps side of things... the holy power generation points we're talking about are purely for survival.

[Nairobi]

And to cover what Nairobi was asking, that's mentioned in the Matlab thread that got linked earlier. From softcap -> hardcap expertise, haste pretty unquestionably wins the higher dps side of things... the holy power generation points we're talking about are purely for survival.

I guess my question was not perhaps as clear as it could have been. What I was considering was that because of haste's hands-down winning over expertise in the DPS realm assuming 7.5+ expertise, is it ever more valuable to stack expertise to hard cap (from a DPS perspective)? Would reforging OUT of expertise into, say, Crit be more effective DPS if you can keep 7.5%?

Likewise, how much "effective DPS" would you lose, going from 30% Haste/13% expertise to ~27% Haste/15% exp?

Basically, I'm interested in the sliding scale of stat weights for this tier, assuming ~530 ilvl worth of stats to play with. As we all have noted, we are rapidly approaching the point where we have a bolus of stat values to move around, so I am trying to gauge the best allocation or probe for some outside the box DPS opportunities.

[Deleted]

A reason why i always favored SW as t5 talent further decreasing exp value as I mentioned in the thread I linked earlier.
Now I have started favoring DP a lot since it is really wonderful. But if i need an extra cd SW falls back in quickly. Never been a fan of HA except for emperors and empress in t14.

But yeah, would be lovely if theck could consider running some of the simulations i mentioned in the above post since that would truely be interesting. Feels like the only fight that current simulations applies to was emperors in t14, and even then a simulation lasting 30-40 seconds starting out with 5 hopo would be more relevant since the dance allowed you to bank up hopo.

Always fun talking with an open mind of someonr with different methodology/opinion, since that spurs the greatest discussions and conclusions

[Theck]

That said, I find the increased SS speed, hps, dps and Rppm procs to favor haste strongly over expertise reliability (which is easily countered by smart hopo use) atleast up to the point in your stats where exp hoporegen= hastes.

....

Some things I would love to see you sim is.

Haste vs exp with SS and shiftqueue included. (with reasonable values of exp and haste, not 0 exp 30 haste vs 15/20)
100% SS uptime, to find how much more valuable mastery is for those periods you can keep sotr on 100% effectivr uptime.
simulating short periods of between 3 to 15 attacks starting with 5 Hopo to simulate tank switches and times you can bank Hopo
30-120 scenarios of above to simulate tank swap figths. A fight with tank swaps every 60 seconds give you 8%~ more effective uptime on sotr favoring mastery.
A simulation where every 10-20th attack hits 200-300% harder to simulate hard hitting abilities. Smart sotr use predicting the attack
Simulations with physical and magic damage dots.

Thats just on the top of my head simulations i think would really help gain a broader picture of tanking and combined give a better opinion on the stats compared to the current patchwerk sim as only source.

Dont get me wrong, love your work and effort. Just think it could be broadened. And hate all the sheeps that ctrl c. Ctrl v you without understanding your work

I already ran some sims with Sacred Shield and the shift queue, and "Control/Haste" still beat out "pure Haste" in spike prevention. But I agree, it may be time to go back and try a wider variety of exp/haste combinations. I'm almost finished updating the sim to include Seal of Insight as well, since the prevailing wisdom is that this will favor haste over stamina (I predict the opposite, actually, because I think people misunderstand the magnitudes of the effects they're considering).

Rather than a "every Nth attack hits for 300%," I was planning on adding a predictable but pseudo-random spike that was independent of the swing timer. For example, a random cooldown of ~15-20 seconds. The player would still have access to that information for preparation reasons, of course. It will complicate the finisher queue logic a bit though.

Holy Avenger wouldn't be too hard to implement either, which would give the 100% uptime information you're interested in.

Tank swaps can be done as well (I did something similar for warriors in the past). It's all a matter of how much time I have to work on the code. Luckily, classes are over in a few weeks, which means less time grading and more time for side projects.

I have never been a fan of using your current sims as hard proof in an argument, rather a foundation, something to grow from, since in my opnion simulating a bosd mindlessly hitting you and putting =tanking behind it is not right. Again, because I rely less on math and more on logic.

Well, they're only "hard" proof for the situation being simulated, which is fairly limited to begin with. We have to extrapolate from that to figure out what it means for real situations. Which is pretty much what you said - it's a foundation, a base from which we begin further deduction.

That said, you mentioned a pretty common criticism of my simulations, and I don't think it's entirely valid. I think a boss mindlessly hitting you is a pretty good model for the bulk of our death scenarios. You're not going to die when you don't have aggro, obviously. And you're probably not going to die while you have Holy Avenger up and are coasting along on 100% uptime of a 50+% damage mitigation buff. Or a big cooldown, for that matter. You're not even that likely to die right after a tank swap, since as you noted, you'll have extra SotR uptime during that period if you've pooled holy power.

You're most likely to die when you get a big damage spike, and that really only happens when you're deprived of all of those safety nets. In other words, when you just get unlucky and take a couple big unmitigated melees because you're rebuilding Holy Power.The "big, predictable spike" death scenario is pretty dangerous too, but less so than many seem to think if you're using active mitigaton properly - shaving 50% off of the spike attack and one of the neighboring melees usually makes it quite manageable - once I add that mechanic to the sim we'll see exactly how that works out numerically. But in any event, by that logic the simple, Patchwerk simulation much more relevant than it would initially appear.

Also note that while there are edge effects for tank-swaps and such, those are going to be relatively small. An 8% increase in uptime sounds like a lot, and it will give you a lower incidence of spike representation across the board for sure. But it probably won't give you an 8% increase in mastery's value, for example. Because when you have SotR up, you're not that susceptible to spikes anyhow, which is one of the reasons haste performs so well - it helps minimize the downtime, which is the danger period, rather than making the already-safe period safer like mastery does. (To be fair, mastery also does this via block chance, which is why it's able to keep up in the simulations at all).

This is also, not coincidentally, why expertise does so well in smoothness metrics. Even though it's lower HPG, the reliability means that your SotRs are more evenly spaced, which means you eliminate strings of unmitigated hits caused by HP starvation. The best analogy is a picket fence - expertise gives you a fence with slightly wider spacing, and haste gives you a fence with narrower spacing but a bunch of missing posts randomly distributed along the length. The spike danger increases dramatically in those gaps where the posts are missing.

I don't fully agree with you that it's possible to mitigate that through careful pooling - even the best player can't do much with a string of 3-4 missed holy power generators. You can certainly react to that problem with other cooldowns to mitigate the problem, though. It all comes down to luck and reliability, and how quickly you think on your feet. I think a novice player is less likely to react appropriately and pop DP or AD when they have a string of missed HPGs, which makes me lean towards the reliability end of things for them. On the other hand, I think you're also correct that a sloppy rotation (like accidentally using J>CS at a clash) favors haste over expertise. I'm not sure what failure mode is more likely and more dangerous for a given player. I prefer the reliability angle for planning 5-10 seconds in advance, but that's not a rigorous argument that it's the optimum choice - just that it may be optimum for my play style.

---------- Post added 2013-04-22 at 04:33 PM ----------

I know your above data is really for HPG simulations, not DPS, but I am wondering if the above would refute that Haste > Exp above soft-cap for DPS. Additionally, would it be beneficial to drop some haste to return to hard-cap (survival benefits notwithstanding).

I guess my question was not perhaps as clear as it could have been. What I was considering was that because of haste's hands-down winning over expertise in the DPS realm assuming 7.5+ expertise, is it ever more valuable to stack expertise to hard cap (from a DPS perspective)? Would reforging OUT of expertise into, say, Crit be more effective DPS if you can keep 7.5%?

Not really. For DPS purposes, all you need to do is consult the MATLAB DPS simulations I host on maintankadin. They're fairly rigorous and obscenely accurate thanks to using a much different simulation method. The forum won't let me post a link, but googling "theck matlab stat comparison" should take you to the table of contents.

Short version: Haste, Crit, Hit, and Expertise all scale essentially the same way. They all scale with Vengeance. You can see from the hit and expertise plots in that post that below hit or exp soft-cap, hit>exp>haste>crit. That will remain true at any Vengeance level.

Once you reach expertise soft cap, exp falls behind haste for DPS, but still edges out crit. so the DPS ordering becomes hit>haste>exp>crit.

There is one caveat here, which is that the stats obviously doesn't scale with themselves. That doesn't have a lot of relevance for the two cap-able stats, but it does matter for crit and haste. The very last plot in that post shows that crit will slowly creep up on haste as you stack more haste. There's definitely a break point, but judging from the slope of that graph it's not going to happen until >50% melee haste, at which point other considerations take over (GCD capping, primarily).

[Theck]

No idea why this double-posted. It was only showing up as one post yesterday. *shrug*

[Deleted]

Very noticeable you raid 25 theck. Sadly i raid 10, even though i prefer 25. As a 10 man tank, i am not really sure what kills me, in the terms of nothing really does. Intentionally stand in shit, on fights like lei-shen, just 1 tank and eat decapitate to the face. Which is why 10 man tanks prefer dps stats so much.
Just beeng meleed down is not the same issue in 10 as in 25, as SS, soi and healer hots are enough to keep you at 100% even if you are not using sotr. What kills you is singlehandly the predictable damage burst together with the attack before it and the two after it. 2-4 string attacks around a big hittinh attack is what is most often relevant in 10. And for that matter it is easy to bank up 5 hopo allowing for 100% sotr coverage. What happenscin between the big hits is not as relevant in 10.

I gotta admit I have no time to read up that much about what changes on the forums, so not entirely sure how your new implementation of sta/healing works. Guess my question is.

In you current sims, if you were to increase or decrease healer throughput (say you under or overestimated it), how would that affect the stats? Would some stat gain/lose more than another?

---------- Post added 2013-04-22 at 10:02 PM ----------

@Pooling solving reliability issues.
I would agree with you if it was not for judgment. You know that your judgment will always hit, even if your cs misses 20 times in a row, judgment still provides enough hopo to cover the big attacks which i mentioned earlier is the worry of a 10 man tank.
The thing is also, as with many other stats, some stats you notice, some you dont. I have raided below exp cap since release, and have not yet seen 1 time that I can say 'that parried cs killed me'. I actually recall 1 time it almost killed me but a swift AD saved the situation. Though even lets say you say 'shit that parry killed me'
Therw is no way to tell how many timed the haste saved you! Since it works morw subtle, less noticed. Maybe being below exp cap killed you twice but the haste saved you 3x. That is still +1. these are the hard questions that are not answered so easily.

[Drummerboy]

For God sake Firefly you should know when to stop. Some of these thoughts should be messaged directly to Theck instead of out in public otherwise it just makes you look irresponsible.

First, playing the "cool guy" trying to dis the "smart guy" causes you to loose credibility. Second after a rebuttal you pull the "ah I'm a 10m tank so your rules 25m don't apply" which again causes you to loose more credibility. Then your last rebuttal to paraphrase: "Haste makes me feel like it I survive more" is just nonsense. A feeling represents nothing but your own satisfaction. And just because you feel satisfied with your style of play doesn't mean you're breaking some barrier or doing something amazing. Raid make-up may allow you to play the way you do but that doesn't mean your style of play is optimal for me.

If you're really that interested in what the sims show then come up with your own sim and prove it. Then start a new site called fireflyduty.net and publish your results. I'm sure you'll get some followers. You won't gain followers by calling people sheep.

If there's anything I've learned with this expansion it's that style of play of every class is highly customizable. There are min/max's for each style of play. There are conditions where one style of play is better than another. There are a lot of variables with the intent to remove the "stamped" talents and styles of play. Finding your own way of play is the best part of this game right now.

Heck I remember having a conversation about how your DPS was crazy on Tortos when I realized my team was tanking the boss completely different than your team. One way of killing Tortos isn't any better than another way as long as the result is the same.

[lakhesis]

Don't forget that the haste raid buff doesn't count, because it's not real haste. It's a melee attack speed buff, not a melee haste buff. It shows up on the character sheet as melee haste, presumably because the character sheet is coded to sample the haste applied to your melee swings. But you can easily test that it doesn't affect Sanctity of Battle (which is good, actually - if it did, spells like AS and SS would have a longer GCD and cooldown than their melee counterparts, leading to some odd delaying effects and clash possibilities in the rotation).

I was actually still thinking that it was influenced by spell haste, I'd missed that change. I took a break for a bit at the tail end of the beta.

Very noticeable you raid 25 theck. Sadly i raid 10, even though i prefer 25. As a 10 man tank, i am not really sure what kills me, in the terms of nothing really does. Intentionally stand in shit, on fights like lei-shen, just 1 tank and eat decapitate to the face. Which is why 10 man tanks prefer dps stats so much.

My guild swapped from 25 -> 2x 10's with this expac, and personally I actually prefer the format. I like the individual responsibility.

What I find kills me in a 10s environment is healer cc/overload/inattention. I definitely feel squishier with a build that focuses haste over stam, but I also feel more able to respond. What I'm actually interested in is the ability to stack+maintain bastion x5 ASAP, and then recover enough Holy Power to use WoG after using 2x SotR while both my healers are being chased by frostbeams on Megaera & I'm eating a breath attack + poison bomb.

Not necessarily the best example considering most of that is magic damage so you'd be less likely to chain SotR & more likely to rotate CDs, but I hope you get my point. Possibly solo tanking Tortos while my healers are derping it up with spinning tortoises would be a better example

But in those situations I most definitely want reliable as well as quick. That's also why I tend to like HA actually. Between that, AD & LoH we can provide an awful lot of emergency burst self-healing. I found if I stacked stam, while I was less squishy, the healers were also much more likely to just let me hover at 70% hp. Now they keep me full & I have more gas in the tank to react if they drop the ball - but reliability is key to that ability.

[Deleted]

Battle healer nerfs! Sky! Falling! Etc

[lakhesis]

Battle healer nerfs! Sky! Falling! Etc

Meh.. maybe 1-5k randomly allocated HPS lost, half of which was typically overhealing.

Rather not lose it, but if it shut up some of the "pally tanks are OP, they healzorz everyvunz!" QQ then it'd be worth it.

[Rivin]

if it did, spells like AS and SS would have a longer GCD and cooldown than their melee counterparts, leading to some odd delaying effects and clash possibilities in the rotation)

Instead, spells are the ones with a shorter GCD due to the spell haste raid buff, so it leads to the same issue either way. :P

[Deleted]

@drumme

So you are denying that 10 and 25 is fundamentally different? I only stated that because theck mentioned some problems that I do not feel as a 10 tank, and vice versa i have some problems in 10 that is not shared with the 25 counterpart. Seems silly to state that 10=25 as you seem to incline.

And when did I ever say that 'haste makes me FEEL like i am surviving more', if anything i said the opposite by stating that exp has a more notable effect than hastes more subtle effect (as you wont tell as easily that getting your spells of 0.01 sec earlier saved your life as with a parried attack)

You seem to be grasping thin air, but if you dont like me, it is fine


And no, I am not gonna start simulating since I do not have thr skillset for that.
But just because I do not sim doesnt mean I cant give constructive criticism to someone that does. By your logic noone can agree or disagree with theck because no other paladin tank sim? Unless you sim, you arent entitled to an opinion either right?

I am however planning another project in the future

---------- Post added 2013-04-23 at 12:13 AM ----------

And I dont remember discussing tortos dps, granted I took a rank 1 but was not really impressive as not many people solo tanked it.
Still, just tanked the bats on top of the boss. Nothing special really, is there any other way (other than kiting) to do it?
Still, high dps as tank on tortos is about dpsers doing low dps. So nothing impressive there at all really.

Also, I must have missed my rebuttal. We only agreed to disagree. I value hopo regen, dps, rppm procs, SS ticks and hps over reliability of haste. You are measuring 6 different factors, all with individual value. Who is to say who is right? Personally I dont value reliability as i feel it is reliable enough and being below exp cap has never yet cost my life since mop release. That is reliable enough for me. But please enlighten me where I got my rebuttal.

[zloitima]

patch 5.3
Glyph of the Battle Healer now heals a nearby injured friendly target, excluding the Paladin, for 20% of damage dealt while Seal of Insight is active, down from 30%.

long awaited nerf strike. 33% nerf to our raid healing, still doesn't change our superiority.

[Xs]

patch 5.3
Glyph of the Battle Healer now heals a nearby injured friendly target, excluding the Paladin, for 20% of damage dealt while Seal of Insight is active, down from 30%.

long awaited nerf strike. 33% nerf to our raid healing, still doesn't change our superiority.

Needed to be done to keep Haste balanced as our DPS scales further and further


Also on the subject of you 3(4) discussing the various ideologies, keep it up personally I tend to sit on Firefly's preferences but with exp rating higher than he does, as I prefer to remove as much randomness as possible from my character, freeing up my cooldowns to cover real issues

[Deleted]

Well xs, youvknow how much i hate randomness aswwell from the avoidance discussions.
Which is kinda my poibt here. From playing below exp cap entire expansion, I dont feel that there is much randomness about it. I stillcover every boss special with sotr, most of the time not even noticing the misses.
Granted this can be a different story in 25s when boss melee matters more, which should make the randomness factor higher. But as i got no experience from 25 this expansion I cant comment on that.

---------- Post added 2013-04-23 at 08:21 AM ----------

Though to sau again.this discussion is a tier late. Since you can exp cap bwith haste>exp prio. Im on about 13-14 exp now, was on 15 earlier.

[lakhesis]

In some ways the element I find more interesting about the gearing in this tier are all the secondary stat oriented trinkets. Especially the expertise/hit ones.

Personally, I'm not prepared to use trinkets to hit / expertise cap. I want to be able to swap in dual stam trinkets and/or dual dps trinkets and/or whatever other fight specific trinket(s) seems appropriate.

Those who're less concerned with reliability in their survival stats would presumably be happy enough with using a trinket for expertise, but even for them using a trinket for hit would be an issue. And remember it's not just the default stats on the trinket, maintaining swappability in your trinkets locks you out of reforging non-hit/exp trinkets to hit/exp.

Personally I have Soul Barrier in fairly constantly, normally paired with Spark of Zandalar, but I also occasionally swap in my 2/2 upgraded Lao-Chin's Liquid Courage & I keep Stuff of Nightmares floating around in my bags for if I need a mastery trinket (haven't used it currently, contemplating trying it for H-Tortos when I get onto it in the next week or two). If I had Fortitude of the Zandalari then I'd regard that as a much more compelling mastery option for some fights.

I've basically dismissed Ji-kun's Rising Winds & all the other expertise/hit trinkets because I'm not prepared to lock them in permanently & 60% of their stats would be wasted if they were a swap-in piece. I'm not particularly happy about doing that, but I feel it's the best approach if you maintain flexibility without routine reforging.

Curious how other people have been approaching their trinket slots.

[Xs]

Curious how other people have been approaching their trinket slots.

I share a similar ideal world situation, but due to #OSworldproblems I'm not able to be as picky, I wouldn't be as adverse to the idea of an expertise trinket though as I feel when I'm swapping out to a Stam trinket in that slot, its usually going to be for a much heavier magic fight, thus losing a little ShotR uptime is reasonable, and as we've seen above you're not missing much from a percent or less exp, specially when I like to run SW diminishing that need for exp even further

[britishbubba]

Curious how other people have been approaching their trinket slots.

I typically run Ji'kun's Feather and the SPA str DPS trinket. If I need to put in a stam trinket for a fight, I hearth, reforge and get summoned back. or if we're in a part of the instance where I can mount, I just get our DK to get on his grand expedition yak and I reforge.

Granted, i'm running 10M currently and I don't feel like I need a stam trinket... Ever. When I was doing 25s earlier this teir, I ran one Accuracy trinket always, and one Stam trinket/non accuracy trinket (fort of the zandalari)

[Drummerboy]

The only difference between 10 and 25 is strategy. When it comes to tanking fundamentally it's still the same: be healable, keep threat, minimize raid damage. Tank theory-crafting targets being healable.

As far as my "feeling" usage, as long as you don't provide empirical data to support your theory then all it is is a feeling. In my line of work (computers) I am quite familiar with people's "feelings". Generally something "feels" better or faster when you receive a highly animated feedback (like a fast moving progress bar vs a slow moving progress bar). Haste tanking feels faster so therefore it can have the feeling of being better. And you could play better when you feel more comfortable with or enjoy your style of play.

Who is to say who is right? Personally I dont value reliability as i feel it is reliable enough and being below exp cap has never yet cost my life since mop release. That is reliable enough for me.

This was my point which you seem to understand. Let me re-emphasize:

That is reliable enough for me.

Theory-crafting is great and Theck's proofs are well thought out. The only arguments used against his proof should be another proof. Unfortunately it's difficult to create a proof regarding human reliability. Come up with a proof that takes reliability variability into account and we can come closer to creating a "human sim". Philosophically though, do we want a sim that plays like a human or a sim that maximizes benefits based on calculations? Which model is better? And will that human sim play like you?

In the end, who cares? Play the way you enjoy. Take the bits and pieces that you want from the current theory-crafting and bend it to your will.

[Theck]

Very noticeable you raid 25 theck. Sadly i raid 10, even though i prefer 25. As a 10 man tank, i am not really sure what kills me, in the terms of nothing really does. Intentionally stand in shit, on fights like lei-shen, just 1 tank and eat decapitate to the face. Which is why 10 man tanks prefer dps stats so much.
Just beeng meleed down is not the same issue in 10 as in 25, as SS, soi and healer hots are enough to keep you at 100% even if you are not using sotr. What kills you is singlehandly the predictable damage burst together with the attack before it and the two after it. 2-4 string attacks around a big hittinh attack is what is most often relevant in 10. And for that matter it is easy to bank up 5 hopo allowing for 100% sotr coverage. What happenscin between the big hits is not as relevant in 10.

I think what you're hitting on is the major difference between 25 and 10: tank throughput. 10-man tanks deal with much smaller melee attacks because of the healer constraints in that format, and as a result, they naturally have a little smoother intake than 25-man tanks.

That does have a subtle effect on how each stat impacts your survivability, but not a lot. For example, Stamina still gives you smoother intake than haste whether the boss hits for 150k or 350k. The size of the gap just shifts a little.

The more important point, which is something I've been trying to hammer home for a long time, is that "stamina > haste for survivability" is not the same as "stack stamina to the exclusion of all else." While having more survivability is always nice, it's not always necessary either. Good tanks learn to recognize when they already have enough survivability for an encounter, and at that point are free to allocate stats wherever they see fit. Which generally means shifting to haste for more DPS nowadays. Which is especially efficient since the survivability sacrifice of stacking haste instead of stamina is relatively small.

So it doesn't really surprise me that you find yourself nearly invulnerable in 10-man raiding, because the survivability threshold is lower in that format. As such, dropping below expertise cap isn't really a problem. Sure, you lose the reliability, but if you're already above a survival threshold you feel comfortable with, that isn't an issue. The fact that expertise might be slightly better doesn't make what you're doing wrong - it just means you're making a calculated trade.

I gotta admit I have no time to read up that much about what changes on the forums, so not entirely sure how your new implementation of sta/healing works. Guess my question is.

In you current sims, if you were to increase or decrease healer throughput (say you under or overestimated it), how would that affect the stats? Would some stat gain/lose more than another?

I have a blog post later today that gives a more thorough overview of how I perform my simulations and why I do them the way I do. The short answer is that there is no healer throughput, because there is no healer in the simulations (nor is there tank health). That said, adding healer throughput in the form of small short-term absorption bubbles tends to strengthen stamina more than haste or mastery - Meloree gives some insight into why that makes sense in the comments on last week's blog post.

---------- Post added 2013-04-23 at 10:42 AM ----------

But in those situations I most definitely want reliable as well as quick. That's also why I tend to like HA actually. Between that, AD & LoH we can provide an awful lot of emergency burst self-healing.

I'm sure I've said this before (though probably not here), but I think HA is heavily underrated by most players. Divine Purpose may give higher average SotR uptime (and higher DPS above a certain haste threshold, but it's about 30% or so), but I think that's a misleading metric. On any fight involving a tank swap, half of that uptime is wasted. Holy Avenger lets you concentrate it's 100% uptime into a period where you're definitely tanking, and much of the fallow period is occurring while someone else is tanking. Depending on the specific timing of the fight, that can make Holy Avenger's effective uptime as high as 50% or more.

---------- Post added 2013-04-23 at 10:46 AM ----------

Instead, spells are the ones with a shorter GCD due to the spell haste raid buff, so it leads to the same issue either way. :P

Similar, but less problematic problem. Our top two holy power generators (CS, J) are melee attacks. Since they're at the top of the priority, most people are used to mapping their internal metronome to the melee GCD. AS, Cons, and HW may come off cooldown earlier and have shorter GCDs, but since they're lower-priority, we're far less likely to cast one of them when CS is only 100-200 ms from being available.

Or at least, that's how I play. Maybe others organize their thought process differently. It would be nice if they were made equal, but that would require SoB going back to spell haste. Which I always thought was a cleaner implementation, personally, if not for spell haste debuffs in PvP. If those debuffs were scaled by cast time (i.e. increased cast time by X% rather than decreasing spell haste), it would remove the problem. Though spell haste is sort of out of control too, so we'd probably have to lose the 10% spell haste on Seal of Insight if that change happened.

---------- Post added 2013-04-23 at 10:55 AM ----------

The thing is also, as with many other stats, some stats you notice, some you dont. I have raided below exp cap since release, and have not yet seen 1 time that I can say 'that parried cs killed me'. I actually recall 1 time it almost killed me but a swift AD saved the situation. Though even lets say you say 'shit that parry killed me'
Therw is no way to tell how many timed the haste saved you! Since it works morw subtle, less noticed. Maybe being below exp cap killed you twice but the haste saved you 3x. That is still +1. these are the hard questions that are not answered so easily.

Oops, I missed this part. While you may not be able to quantify hastes effects on deaths in a given combat log, you can make estimates of the effect in stochastic cases. That's exactly what my simulation does - it looks at the magnitude and frequency of spike events over a long period of time. If that shows expertise removing spikes better than haste does, it means that the reliability trumps the increased holy power gains for overall smoothing (the fence post analogy).

Of course, that doesn't include your ability to react with AD and what not, and I think that trying to build a comprehensive set of cooldown/reaction logic is no easy task, and probably beyond the scope of the simulation I've written (partly because, like you, I have ... plans; let's just say if all goes well this summer may be a very interesting time for tanks of all varieties). However, the point is that while you were able to respond with AD to counteract that situation, perhaps with expertise you wouldn't have had to in the first place.

So while you might not know how many times the haste saved you, I also don't think it's fair to assume a value without evidence. It may have saved you 3x or it may have saved you 1x - who's to say? The best estimate we have is to compare it to overall spike prevention, which is what the stochastic simulations I run are modeling.

---------- Post added 2013-04-23 at 11:04 AM ----------

Theory-crafting is great and Theck's proofs are well thought out. The only arguments used against his proof should be another proof. Unfortunately it's difficult to create a proof regarding human reliability. Come up with a proof that takes reliability variability into account and we can come closer to creating a "human sim". Philosophically though, do we want a sim that plays like a human or a sim that maximizes benefits based on calculations? Which model is better? And will that human sim play like you?

Just for clarity, keep in mind that what I'm offering is not a "proof" by any formal definition. It is a set of data, and the analysis accompanying that data. Any good scientist would balk at calling that a proof, because it isn't a universal truth derived from fundamental equations and axioms.

It's data from a model of how a very complex and intricate system works. Sort of like how statisticians model population flow, or the stock market. While it is rather sophisticated, and may very well be the best model we have available, it is in no way complete. It's an abstraction that tries to do the best job it can of modeling the situation in a way that produces useful results.

It's also not infallible. I'm human after all, I make mistakes. But I try to be as open as I can with the code so that if I do make mistakes, they can be identified and corrected and the model improved (see the last two weeks - I found a bug with my Sacred Shield implementation that was skewing results, and posted a follow-up to make sure people knew about the change). That's very important, because I don't do this stuff so people will worship me as the patron saint of paladin tanking. I do it because I want to make sure our class has as accurate and reliable information as possible (partially for selfish reasons - I want to be able to answer the questions I come up with while analyzing my own performance, and often those answers don't exist before I work the problem out or simulate it).