It's a common myth that because Discipline priests have a talent that affects spell crit, crit must be good for them. This is similar to people thinking Spirit is good for Shadow Priests because they have Twisted Faith.

The first thing to consider is what your raid role is. If your raid needs someone to do nothing but slug heals into the tank(s), a Discipline priest isn't the best choice. They would be better served by a Paladin or Shaman. Preventative shield spam is a large part of why discipline priests are so good. Exclusive raid healing is best served by a holy priest. If you need a healer that can take care of a lot of raid healing while still spending a good portion of time on the tanks, then discipline is the best choice. You can see this in a typical log as well: World of Logs - Real Time Raid Analysis

On this Algalon kill, Worldeater has 55% of his healing from shield absorbs and 45% from standard heals. Typically speaking, only half of your healing was even eligible for a crit. The absolute high end should be around 70% non-shield healing, for a mostly focused tank healing situation.

What exactly does Divine Aegis provide? Your crits heal for 150% of normal healing, and it provides 30% of that value as a shield. This is effectively 45% of the base heal. It changes your crits from 150% to 195%, and it only applies to the percentage of healing that could crit. In other words, Divine Aegis has this average effect on your heals:

Crit Chance x Percent of Heals that can Crit x 45%

If half of your heals can crit because they aren't shields, the formula is:

Crit Chance x 22.5%

So a 1% increase in crit rating produces a 0.225% increase in effective healing. At around 3800 HPS, that 1% crit will give you an extra 8.4 HPS. And that 1% crit cost 46 crit rating. If you got those item points in spell power, you'd have 53 extra spell power. With a 1.2068 coefficient on Power Word: Shield after the borrowed time talent, that 53 spell power provides an extra 71 points of shielding to each shield alone. It will give you 41 extra healing per second on your penance and 37 to your flash heal. All totaled, that much spell power will give somewhere between 40 and 60 extra healing per second.

That makes spell power at least 5 times better than crit for discipline priests, probably even better than that. Crit rating really is atrocious.

No, that's not true. Your spell power increases the amount healed by the spell before any multiplier effects are added. In other words, they scale the base heal by the specified percent and the coefficient by the same percent. Talents like Empowered Healing only modify the coefficient, not the base. Here's some data from a log to prove it: World of Logs - Real Time Raid Analysis

That's the data on Flash Heal during an Algalon kill. Here are the starting variables.

My buffed Spell power: 3600
Multipliers on Flash Heal: 10% from Spiritual Guidance, 3% from Blessed Resilience, occasionally 12% from Test of Faith
Flash Heal base: 1887 - 2193 = 2040
Flash Heal coefficient: .8068 (No empowered Healing)
Bonus Healing: 3600 * .8068 = 2904

The multipliers are either 1.133 or 1.27 depending on whether Test of Faith applies.

Option A: Multipliers only affect base heal

Average non-crit Flash Heal (no Test of Faith)
2904 + (2040 * 1.133) = 5215

Average non-crit Flash Heal (with Test of Faith)
2904 + (2040 * 1.27) = 5495


Option B: Multipliers affect base heal and bonus spell power

Average non-crit Flash Heal (no Test of Faith)
(2904 + 2040) * 1.133 = 5602

Average non-crit Flash Heal (with Test of Faith)
(2904 + 2040) * 1.27 = 6279


Browsing through the log, you can see that my typical non-crit flash heal is over 6000 healing. So it's clearly the second option.

I think it depends on how you look at this...

When you compare SP, Crit rating, and Haste rating. Basically both Haste and Crit are never going to compare to SP. SP is just that good. So when it comes down to it you shouldn't ever choose to gem/enchant haste or crit over spell power. Crit isn't that atrocious when compared to Haste which is normally the trade in stats that occurs. Depending on spell composition and current gear the desire for either could be negligble or could lean in favor of one over the other.

Saying Crit is atrocious for Disc would in general push towards gearing Haste which can be equal to more atrocious. It would be safer to say that "SP is always more valuable for throughput than Crit" and leave it at that.

First of all, I agree that a Paladin is a better primary tank healer than a disc priest, and in a perfect world when every guild has access to many skilled, geared and experienced players of every spec of every class I would never advocate a disc priest being the primary tank healer. But lets be realistic, most disc priests are going to often find themselves in the position of primary tank healer, so it's important to prepare for that. Although, I do agree that a disc priest is best utilized by doing a combination of raid sheilding/healing while also helping on the tanks.

In my original argument I stated that crit is not atrocious in the setting of tank healing, this is because while exclusively tank healing only 1 spell every 15 seconds will not take advantage of crit (I'll ignore that the glyph can crit). The other 13 seconds are all going to be direct healing spells which can all take advantage of criticals (and maybe another 1 or 2 PWS on yourself or another raid member). The WoL you link shows a disc priest almost exclusively raid healing, and in that scenario I agree crit will not be very valuable because of all the PWS casts. But I'm talking about while tank healing.

I think you are mistaken in believing that because half of the effective healing done by a disc priest is though shielding that therefore half of the spells cast by that priest are PWS. Because shields are much less likely to overheal than a direct heal, they will show up as a disproportionate percentage of effective healing.

Also, let's remember that the extra healing from DA is in the presence of a shield and not a direct heal, so it is much more likely to be effective.

I would argue for adjusting your equations to more like 80-90% of casts can crit (tank healing scenario) and instead of 45% additional value, some higher value due to the higher likelihood of the DA shield being effective. Given those adjustments, the HPS value of 1% crit will be much higher than 8.4, more like somewhere in the neighborhood of 25 to 30. Still not as much as an equivalent amount of spell power, I admit, but my argument was never to sacrifice spell power for crit. I simply think that it is incorrect to believe that crit rating is atrocious.

*Edit* I do agree with TheDoctor, spell power is always more valuable than crit.

*Edit 2* Props to l337n00b's post below, I missed that the value of crit should really be 95%, not 45%, which will further boost the perceived value of crit.

I believe we basically agree. I think you have to balance your stats relative to one another, and that such balance changes, both as your gear progresses and from one encounter to another. I also believe it would be possible to theorycraft a "best" budget for holy for each particular encounter. And of course, we should all be seeking to maximize throughput.

But practically speaking what we probably are best able to do is to have some items of gear that we can swap out depending on the specific encounter so as to most closely approach that theoretical "best" relative weighting for that encounter.

What I postulate above that, however, is that from the practical in-game raiding aspect, the relative value of Haste is higher than its on paper theoretical value, relative to both the numbers you end up actually producing on the meters and the ability you thereby gain to do more and do it quicker at key crisis points of the encounters you attempt.

An experienced and skilled holy priest with sufficient minimum other stats is in my postulation better enabled by virtue of adding Haste to do more and do it quicker to mean success for the raid than by doing the same thing with any other single stat.

Further, when seeking to maximize your effective healing in a competitive raid, that additional Haste will imo translate more often and more certainly into more effective healing than adding any other single stat past that minimally adequate level.

While SP on paper is arguably the single max throughput stat, what I see in actual game raiding conditions is that past some point more SP more often translates into increased overhealing than increased effective healing, while Haste more often translates into increased effective healing.

This last does bring into consideration the issue of heal sniping. But worrying about sniping another's effective healing is second to having been able to get to targets about to die fast enough that they instead live, when very possibly they would not have otherwise.

Your reasoning is flawed in several points.
First of all, all coefficients that apply to your total healing modify not only your spellpower value, but also your crit and haste values. For relative weighting, you can therefore ignore them. The only coefficient you can't ignore are the one that modify only one or several of the weights. For holy, that's Empowered healing, that is not taken by everyone.

Second, you can't ignore crit and haste. Crit and haste increases the value of spellpower (and, same song as ever, haste and spellpower increase the value of crit and crit and spellpower increases the value of haste...). Since disc. has a better base crit, (via mental strength, focused will and renewed hope), a better crit coefficient (divine aegis) and a better base haste (enlightment, and sometimes borrowed time), it also has better relative return for spellpower than holy. That's easier to see if you consider that you need to "equilibrate" your stats. Holy has low base haste and crit, and high base spellpower (because of the spirit -> spellpower). Disc has no spellpower gain, but high base crit and haste. Since both need to equilibrate their stats, disc is more keen on spellpower than holy, and less interested by haste and crit. Note that I'm not saying that spellpower is bad for holy : just that the relative weights are shifted in some direction. The only counter-shift comes from Empowered healing : I'm too lazy to make a model and compute it to back my claim, but I'm pretty sure the Empowered healing alone is not enough to compensate the crit and haste differente.

Last, but that's not the main point, different haste effects stacks multiplicatively, not additively. So, the disc PW:S-spamm haste soft cap is more around 5% than 11%, as Constantius pointed out.

Once again, I need to disagree on your approach here. In addition to what has been said by other :
1/ I took the time to look at the log you gave : the disc. priest used 10 times penance for the whole fight, and 183 times PW;S. If you add POH and POM, then that's 86% of his healing that's done with raid-healing tools. If you look at healing received, both tanks got around 10% of their total healing from the disc. priest. I'm not familiar enough with WoL to look at which spells were responsible of tank healing. Anyway, that's what I call raid-healing usually, and I think we all agree with you : for raid healing and PW:S spamming, crit is not good. Neither is haste. You just want spellpower and regen. We already knew that...

2/ This point would put the numbers more on your side, and in this case, it would not change a lot, but it's more a methodological point :
the correct formula to compute the gain of 1% more crit for a disc priest is :


In details, you can't just take numbers, say that half of your healing are shields (mixing DA and PW:S), take the second half, and say that 1% more crit is 0.95% (or 0.45%) more healing on that second half. You first need to add DA to the crit side. Then you need to take into account that this "crit" side is already affected by crit. Going from 0% to 1% crit is not the same as going from 90 to 91% crit. In fact, you want to divide the total "crit side" healing by (1+x * 0.95), and multiply it by (1+ x * 0.95 + 0.01 * 0.95), where x is the effective crit ratio (between 0 and 1). Note that dividing by 1 + x*0.95 has as effect to remove both the crit and the DA effect, so that's why you need to count DA into the "crit side".

You are correct that this is bad math on my part. For a situation of some shields, some heals, a .6 return seems reasonable and seems to match the logs. I have trouble calling that number atrocious, but it's worse than that in a shield spam fight. Of course, haste is also bad in a shield spam fight, so perhaps it's better to theorycraft the right stat to take for the situation where they both might be useful.

Regarding your log analysis that estimated crit returns of around .7, did you base your numbers off the average heal size or average non-crit heal? If I have roughly 36% crit, then that average heal is going to be 18% larger than the base non-crit value, which would skew the estimates by 18%.

It sounds like your postulate is this:

- The evidence is wrong; haste is better than the meters show
- The theory is wrong; haste is better than the theory suggests

That's not necessarily false, but it's a rather bold suggestion, and I don't think most people here will be swayed by rhetoric alone. If you want to prove your point, you need to show numbers somehow. Either provide a new method for analyzing data that shows the additional value or provide a new theorycrafting framework that models this concept of "smoothness". Otherwise it's just another version of Russel's Teapot.

Now this is something measurable. It turns out that this is not true. More spell power does not translate into increased overhealing. Here's my healing on a pair of Algalon parses, one month apart. The first parse is from the first week of ToC and the second parse is from last week. Click the "Healing by Spell" tab and you'll see the total overhealing done by percent.

Aug 11 (45.4%): World of Logs - Real Time Raid Analysis
Sep 10 (47.8%): World of Logs - Real Time Raid Analysis

Sadly I can only show parses 4 weeks month apart, as I played shadow on the fights previous that. However, I do have parses for a Paladin on that fight over a 6 week spread:

Jul 30 (66.5%): World of Logs - Real Time Raid Analysis
Aug 9 (64.6%): World of Logs - Real Time Raid Analysis
Aug 11 (64.6%): World of Logs - Real Time Raid Analysis
Sep 10 (62.8%): World of Logs - Real Time Raid Analysis

You are looking at variations of around 2%. This is sufficient to prove that if there is any effect, it is minimal. Plus at least in the case of the Paladin, the effect seems to be that more spell power decreases the amount of overhealing rather than increasing it.

More spell power does not result in more overhealing.

Re read what I wrote, I know how the multipliers work. Claiming that talents such as spiritual healing favors spell power over haste is absurd.

That's not what I was claiming though. The original post was dealing with calculations regarding the effective returns on spell power a given spec gets. My point was that an extra X points of spell power will provide greater returns for a spec with +20% multipliers than one that only has +10%. I'm talking about the relative value of spell power between discipline and holy, not the value of spell power relative to crit and haste.

At any rate, Power Word: Shield is the only Discipline spell that has particularly good returns on spell power, so if you aren't casting it a lot, spell power will benefit holy more. All this really means is that holy is the better spec for that combat situation though.

Again, how valuable spell power is to a class/spec has NOTHING what so ever to do with the modifiers you are claiming do. That's the problem with your post and especially the conclusion you draw. The value of spell power is primarily determined by the spell power coefficient, stating that holy gets more benefit from spell power than haste compared to disc because of higher multipliers that apply to both the base heal and spell power coefficient is pure and simply wrong.

When it comes to direct heals, however, we the number that we are going to cast between when a person is hurt and the next time they are at full is very small. From one of my logs I get about 2.56 heals that I can each time I took a person to full (counting penance as one heal rather than 3). This is composed of a vast majority of times where I cast 1 or 2 heals, having the first or second one overheal partially (or fully because someone else got there first) tanks needing 3 with some regularity and a few times where someone stays below max health for quite some time. Because of the distribution, it seems like a plateaus for reducing overhealing are going to be minimial or non-existant until we get to the casting 2 things instead of 3 stage. But for Flash Heal, if you've currently got about 3k spell power, that would require getting about another 2760 spell power to increase your effect by 50%. This is very quick a dirty, but I think these plateaus are unattainable. Even if they come with noticably lower numbers, if we have to wait until IC then the hits will be larger and the amount of spell power we need to realize those returns will go up again. Of course this is a log that happens to go through regular ToC and up to the end of Faction Champs. If I analyzed my log of Val'kyrs/Anub the number of heals between someone being full and them being full again would probably be much, much higher.

That begin said, the spell power really doesn't increase overhealing much. On the same log I simulated adding 299 spell power (the amount I would current get if I swapped from gemming for int and matching red sockets to gemming for spell power and matching yellow sockets). This increased my total overheal from 23.42% to 23.87%. If I wanted to factor this effect in to my valuation of spell power, I would still be rating spell power at over 90% of what I am rating it now. Of course if I did that I'd also have to factor a similar thing into crit and spell power would actually end up gaining on crit for throughput by a small margin (and who knows what I would have to do to simulate the didn't-make-a-difference component of haste).

I checked and I can't find any glaring mistakes in my calculations, though there may be some inobvious ones since the whole procedure is getting incredibly complicated at this point. But the most likely explanation for a large difference between what I get and what you get is just different casting patterns. If I parse different bosses I can get sizable differences in stat valuation. Since I raid 10's the numbers I produce are probably pretty different than the ones that most of the people on this board would produce.