I hate to run into the crossfire here, but I think I see some common ground with both approaches...

Or, more precisely, I agree with Stolidus in that each encounter has different threshold requirements, and I agree with galzohar in that there's a way to quantify gear valuations based on objective data...

And here's the common ground:

Stolidus, I think the answer to your question is to offer a possible clarification of your position: correct me if I'm wrong, but rather than saying "you need to be fine for each fight," would it not be more clear to say (this is synonymous!, just more precise, I think), "you need to have at least +XXXX healing for XXXX fight, and +YYY mp5 for YYYY fight." And as it applies to galzohar's model, are you not just saying "giving a simple value to +healing would favor someone stacking their healing beyond the base +XXXX healing threshold and leave their +YYY mp5 too far below the YYYY fight threshold?"

galzohar, I think the input you're looking for would be to say, rather than looking at it from the pure HPM from efficient HPS model, I would propose, value the stats based on their availability insofar as they allow you to reach the critical 'goals,' or 'thresholds' for fights. I don't think you disagree with Stolidus when he says you need a minimum +healing AND a minimum +mp5--and I don't think he disagrees with you when he says you need to be "fine" at least. In fact, I think you're both attacking the same problem: You need definite valuations to help decide what gems to socket, for example, and he's saying, those definite valuations have to be based on encounters--which favors definite values for stats based on reaching the requirements for each fight. This would result in a model based on the availability of given stats, as opposed to values that remain fixed regardless of your progress towards or beyond your goals.

In other words, for the fellow with +1800 healing and 0 mp5, mp5 is more valuable than +healing, whereas for the fellow with +0 healing, and 200 mp5, the opposite is true. Which is what you are both saying.

The thing is I don't agree with absolute +healing and mp5 requirement. While I agree with a minimum HPS requirement, since max rank HL is more than enough HPS for every fight that I see anybody talking about, there is no +healing requirement to have big enough heals to get through spikes. And if you're gearing by my model, you're definitely not going to have any less HPS than someone gearing "classically" (nightseyes etc) anyway. So it comes down to the mp5 requirement. The way I see it mp5 and +healing are interchangeable. The more +healing you have, the less mp5 you need. If you say you need 1800 +heal and 200 mp5 to do a fight, you'll probably be fine and even better with 1900 +healing 175 mp5, as by my model the extra 100 healing already give you 25 mp5 through HPS on top of making you more efficienct by doing more hp/mana.

Since we seem to agree this model works on fights where you heal by the style it assumes, I would like to look deeper into the examples where you say it doesn't work (if you have access to top guild WWS on those fights maybe we could work something with it, but can WWS really go to every single heal and see who it landed on, what was his HP at the time and what other heals landed at +/-2 seconds from it?).
The thing is if you're going oom it doesn't nescessarily mean you need to drop healing in favor of mp5. This is actually true for all classes, although it's much simpler to show for DPS classes. If you go to the appropriate thread, you'll see that no matter what length the fight is, there is no way mp5 is worth it over damage to maximize damage done in a fight, even if you run oom mid fight. Where this applies to healing is that the more +healing you have, the more healing you can do. The more mp5 you have, the more healing you can do. If ignoring HPS, mp5 and healing have the exact same purpose - to heal more, which is modeled well on the spreadsheet. In that case mp5 somewhat surpasses healing (but not by a lot! Even then 9 heal is clearly better than 1 mp5 for example). Now since HPS isn't just about "OMG my target is dying" (since we already eliminated that problem by having an overkilling max rank holy light spell), but also increases your efficiency by allowing you to use more efficient spells slightly more often (and yes, I realize 100 healing would stil be *slightly* more fol and *slightly* less HL, to the point where you might not even notice when testing, but it is there). That's why I say +healing has a bigger value than what's shown on the spreadsheet. Heck even the HPS from haste has efficiency equivalence, although with some napkin math it seems to be still way too weak (per itemization cost).

Crit is something I'm still trying to figure out. While the mana restored from crits is very much straight-forward, the HPS and healing efficiency (on top of the spell being cheaper, of course) gained from critting is very hard to figure out. Can be anywhere between 0 (if every crit will always do 100% overhealing) to full effect (if the portion of crits' overhealing is equal to the portion of non-crits' overhealing). I'm trying to figure out a way how to take a WWS and figure out where that % roughly is in general, but I'm kinda clueless as how to do it. One guy posted already that he got a certain % of his critical healing being overheal, but you have to consider you're already overhealing without critting, so you have to somehow take into account your non-crit overhealing when figuring out the crit overhealing (as in, how much your crit overheals MORE than your non-crits, rather than just how much your crits overheal). That's why I would take his "10 crit is worth 1 healing in terms of healing done" as the minimum HPS/HPM value of crit (added to its regen value).
The thing is I'm pretty confident that the extra healing done by crits is far lower from "crits heal for 1.5X", just not sure how much lower. I think the spreadsheet is modeling crit heals for being counted as 1.5X healing, any way the spreadsheet maker can adjust it and allow you to enter your "own" crit value? While this doesn't make a big difference when you consider more healing done per heal as more efficiency, when you consider the HPS factor of it as additional efficiency it becomes very important to at least roughly figure out the crit value. Crits on average account for 1.1X size heals? 1.2X? And again the srpeadsheet should allow you to put the value that one should figure out and go by that rather than assume crits heal for 1.5x.
Remember if you consider the HPS from crit as 0, but count healing done the way the spreadsheet does and the mp5 gained in the same way, crit is rather weaker than all other stats, but not a lot weaker. If you take into account its HPS value, it becomes incredibly good (don't remember if better than +healing or not, but this is obviously not a realistic scenario anyway). If you gimp down the crit coefficient you lose a lot of value for crit again... What I really want to figure out is how much.

As for finding where my model doesn't work, I'm still watiing to see if someone can put up a real example where my model fails and show mathematically how different gearing would be better, which wasn't done with either examples. Both examples say the model fails because "you can't FoL at all" but then again say "you do use FoL sometimes".
For the people who "run out of mana so need more rege" I think by that point it should be obvious that the extra efficiency and HPS from +healing more than makes up for that and you will run oom even more if you stack mp5 and cast more HLs and less FoLs. This contradicts intuition (which says "hmm I'm oom I need more mana") obviuosly, but I think we all know intuition is very weak at evaluating small differences such as this.

To be fair, if there is any sort of mana-burning going on, it does. Though I don't think you meant that sentence to be absolute--I was just mentioning for those who might have misread you.

In regards to the crit question, my view has been that since you can't count on critting on any given heal (Divine Favor excluded, of course), you can't heal with the presumption of a crit--meaning you should, theoretically, be overhealing for the extra 50% of the heal everytime. Thereby, the only real advantage you can take into account for critting is the mana regen. Something on the order of:

(Total mana pool) / 2) * crit% = the amount of mana it will give you assuming you cast until out of mana.

Taking into account the 1:22.1 ratio of crit%:"spell crit strike rating" (SCSR):

(Total mana pool) / 2) * (SCSR / 22.1) = the amount of mana an item will return to you assuming you cast until out of mana.

This doesn't factor in regen, of course. Rather than fill this post with equations that everyone knows (unless you want that?), I'll leave it at the above.

The only time I would really factor in the extra healing benefit from crits is if you know you won't be healing to full, at which point the crit gives a full advantage. I really don't see that as our role as healers, however--it appears that in today's raiding, keeping everyone topped off is the norm with few exceptions.

That's exactly why I was saying I would guess crits are much less than full value when it comes to HPS. However this value is NOT 0, as your heals don't ALWAYS heal the target to full. Yeah you're trying to do that most of the time, but in reality it doesn't happen all the time. What I want to evaluate is how much of the time it's actually happening when the crits actually do healing, and how much healing they do in those situations on average, and from this to figure out the effective healing out of crit.
I agree with you it would probably be a fraction of the actual "crits heal for 1.5X", however this fraction is definitely not 0.

The mana restored from critting is simple to calculate and is modeled by the spreadsheet already, so that's not really a topic for discussion ;p

Of all the characteristics of gear evaluation that are subjective, overhealing from crits is probably the most dependent on play-style.

Any average about "this is how often your crit heals will be most or somewhat more effective" is so generalized as to be useless.

Think about it this way: two classrooms with ten students each may have an average of a "C," but one class has 5 students with A's, and 5 students with F's, while the other class has 10 students with "C's."

Any "average" crit healing is going to be misleading since sometimes you're always "wasting" the extra crit% because you're keeping your target topped off, and other times you'll be behind the eight-ball--and evaluating gear based on an average of both will make you worse off for either.

I would advise that you only take into account the known and consistent result, the mana regen, and treat the bonus healing as "gravy:" that way you can maximize your effectiveness, rather than just being "ok in everything, but the best at nothing."

My point is best crystallized as: You can never cast a spell and be assured that it will crit*, thereby, casting assuming a crit can be potentially fatal to your raid. Hence, in the best of all possible worlds, and the worst of all possible worlds, the extra healing potential from crits is 0, due to it being an overheal. The only remaining scenario of benefit from crit is in the "middle of the road" and precisely what objective gear evaluation hopes to avoid; some vague "sometimes" dependent on playstyle.

I should clarify here, I'm a huge proponent of your HPS = HPM because of being able to "downrank" from HL to FoL, and I use and love gear/stat evaluation tools and ratios...

But they must be useful for those of us who enjoy the high-end--and that means avoiding things like averages with huge variation, and focusing on specific data with little to no deviations.

I happen to have built, as I went along, a mp5 set, a +healing set, and a +crit set of gear, because I'm a pack-rat and ridiculously "OCD," and I test each as much as possible. To be frank? In most 5-mans, the mp5 is king; on raid trash, the +crit is the best; and on the bosses, +healing is ideal; generally. The gear is not all equal, but it's still a good "vague indicator" of what our different needs are.

You put it best when you said:

I dunno about the crit. In optimal situation, you'll adjust your heal size to what the target needs, resulting in 0 overhealing and 100% crit overhealing, in which case the healing from crit is useless. However in reality, while you're trying to avoid it, you're still getting times when you heal someone for less than full for a few reasons, one of them can be "he's not gonna die so I'm using the more efficient heal even though it's not bringing him to 100%", in which case a crit will do more. Or when you heal a tank and he got hit right as your heal landed and you ended up healing for less than topping him off... I still think we should find some way to evaluate the extra healing from crit, however small it is, as it is not absolutely zero, although until I find something better I suppose we can treat the extra healing from crit as "near zero". I suppose even if you do count the extra healing that was done in a fight, it's probably just causing someone else to overheal assuming perfect play, reducing the effective healing from crit even more (maybe that's why you feel crit is better on trash? Slackier play leading to more situations where a crit heal won't overheal?).
Anyone know how to adjust the spreadsheet to take it into account, as I assume it takes into account a crit heal as 1.5X healing? Would be nice if you can set the crit heals to only heal 100% isntead of 150% and see how it changes stat values.

As for actual in-game testing, maybe a video (with good addons that'll show what's going on) be more useful. You can analize it and count how many times in a fight you could've flashed instead of HL if you had more +healing, and even though it would be a very small number of times, it's the kind of difference you should expect by a small change of gear and means my theory is far from irrelevent. That is assuming that number of times is more than 0, and it seems to not be 0 even in the examples that were supposed to be "the scenario where the theory fails".

Bottom line is of course you'll have hard time seeing those differences in WWSs. Even DPS where it's a straight-forward calculation of damage ends up with variance on WWS, not just because of the RNG (critted more during heroism? Got less AOEs targeted at you so moved less?) but because of mistakes and assignments different from purely DPSing. However looking at a WWS and saying "if you had this and that you would've done better becuase of X and Y" should be a more realistic thing to try.

HPS[HL]/MC[HL] = HPS[FoL]/MC[FoL]

HPS[FoL] = (BaseHeal[FoL] + (+healing*0.42))/1.5

HPS[HL] : Healing Per Second for Holy Light
MC[HL] : Mana cost of Holy Light
HPS[FoL] : Healing Per Second of Flash of Light This quantity is unknown
MC[FoL] : Mana cost of Flash of Light
BaseHeal[FoL] : Base heal amount on FoL
+healing*0.42 : +healing with modifier This quantity is unknown

This would tell us how much +healing is needed to reach a threshold amount of HPS as provided by HL.

Correct?

Ok, so, I did some math here. I was attempting to discern at what +healing can FoL match the HPS of HL.

Gosh, I hope this comes out ok:

+healing||HPS of FoL|HPS of HL
1||316.9467|928.628
100||344.6667|951.2
1000||596.6667|1156.4
2000||876.6667|1384.4
3000||1156.667|1612.4
4000||1436.667|1840.4
5000||1716.667|2068.4
6000||1996.667|2296.4
7000||2276.667|2524.4
10000||3116.667|3208.4
12000||3676.667|3664.4

After the initial 1,000, I started putting arbitrarily large values just to find the intersection. It appears as though as you increase the +healing, FoL gets a disproportionate benefit (which we knew), but only at a minuscule rate of increase. It isn't until you reach +12,000healing (that's twelve-thousand, not 1,200) that FoL finally matches the HPS of HL.

With such a small rate of increase, it doesn't appear that +healing can translate to non-negligible amounts of mp5 except in the aggregate.

The point is that:
1. If you spam flashes the whole fight you will still go oom with any realistic gear setup. You *will* do the most effective healing in a fight spamming flashes if the fight is a basic tank&spank with no damage spikes (unrealistic, which is why we have HL).
2. At some points in the fight you end up not really doing any healing because you simply don't need to (and/or can afford to let HoTs/other healers cover it to save your mana for later), while at other points a lot of healing is needed, and when that healing needed is too much, you use holy light instead, which supposedly always heals for more than enough for your target to not die (or else, you should stack +healing/haste regardless of anything calculated here).
3. To increase FoL enough so it provides the same HPS as HL would in your current gear, you need something like 800 additional +healing (I think my basis was 1800 +healing). For 100% HL spam to give you the same effective healing done in a given fight (again I don't remember exactly which stats I took but they were pretty standard) as flash of light would, you would need additional ~200 mp5 on your gear. Therefore to maintain the same HPS we could increase our healing by 800 or increase our mp5 by 200 and have the exact same healing done in the fight. That's why the HPS portion of 4 healing is interchangeable with 1 mp5.
4. The mp5 from HPS is not taking into account the fact that all your heals are now also more efficient on top of having additional HPS. That is already represented on the spreadsheet. So if the spreadsheet tells you, for example, that 10 healing is equal to 2 mp5, then I say 10 healing is equal 2 + 10/4 = 4.5 mp5.

Of course the exact 4:1 ratio I got is gear dependant, as your gear can affect both the mp5 needed and healing needed, but that's nothing you don't know how to calculate for your gear (can we have burst HPS of heals added to the spreadsheet?). What does remain the same for every gear setup is that +healing gains a huge benefit over the spreadsheet's value due to allowing you to reduce the amount of HLs and increase the amount of FoLs.

Note that there is a "second order" correction to this as your spreadsheet's HL:FoL ratios changes as you upgrade gear, however the effect of this difference on your stat evaluation will be extremely small (just like for a DPS caster increasing 10 spell damage has extremely small effects on his damage:crit value). Remember with my gear example I checked ratios for you needed 800 +heal to go from 60:40 FoL:HL (in mana used, not spells casted, as that's what the spreadsheet goes by) to 100:0, so some item upgrade with like 20 +healing would change this by something like 1% (probably a little closer to 2% since the spreadsheet % is mana usage while this calculations is for time spent casting these spells - this is still small, however), which should have very little effect on how the spreadsheet values the stats against eachother.

The whole point is that if you have people die becuase your HL is too weak, you need more healing/haste and NOTHING else (as if you get anything else over healing/haste, your target will die...). If you're already keeping everyone up with FoL and still run oom spamming it on everyone that need healing (which according to the spreadsheet will happen on the theoretical fight in which you can actually have enough HPS with FoL) then you can compeltely ignore me and go by the spreadsheet putting 100% FoL there. If FoL does enough HPS and you have enough mana to spam it non stop for the whole fight, then you're way overgearing the fight anyway and should worry about more advanced content. These extremes, however, never happen afaik. In reality you will occasionally need the more HPS of HL but will never have the mana to sustain casting it forever and thus want to FoL as much as possible, which makes additional HPS allow you to save more mana by casting more FoL and less HL, resulting in more healing done overall.

A more concrete question: given chain-chugging super mana pots and spam healing FoL at max rank, and assuming no crit heals, what is the mp5 required to not go OOM?

Either your math or my math is wrong.

Given the equation:

HPS = (BaseHealofSpell + (+healing*(CastTime/3.5)))/CastTime

FoL reaches the same HPS as HL at +12,000 healing.

I used a (another) spreadsheet to figure out the HPS differences. I don't know what went wrong there and don't want to look it up (especially if it has errors), so I'll just re-do the math:

I'll ignore the +12% healing from talents as it just multiplies the HPS values and then you divide by it to figure out the +healing needed, so I just won't multiply by it from the start. This is also assuming BoL is affected by that 12%, which might not be true which can slightly alter results.

HL base: 2196 - 2446, or 2321 average.
With 1800 +healing HL and +580 BoL average would be 5686. Over 2s that's 2843 HPS.

FoL base is 475 at lvl66, though sould be more at lvl70 due to spells scaling with level when there is no higher rank available. Due ot lack of data (my original numbers came from playing with another spreadsheet someone posted which I don't want to find ATM) I'll work with the 475 though.
With 1800 +healing and 180 BoL FoL average would be 1426, or 950 HPS.

HPS difference would be 1722 so you would need 1893*3.5/1.5=4417 +healing to cover the gap... Defnitely a lot more than 800 (makes me want to find that spreadsheet again but I'll refrain ATM) but not the 12000.

Just looking at the spreadsheet again, I would need with the gear I put there, for HL spam to do the same total healing done over 10 minutes as FoL spam, to increase my MP5 on gear by 206. Note that FoL spamming would end up casting 75% of the time, with HL much less (obviously, due to the much higher HPS, even after adding the mp5).

Bottom line is while my math was way off, the basic theory remains. Except the HPS->efficiency conversion is more like 4417/206=21.44 healing needed for an additional mp5 benefit rather than 8. Note that since the spreadsheet is counting crit heals as effective healing, and HL has 6% more crit, it will actually require some more mp5 to actually reach and thus less +healing needed to be equivalent to 1 mp5 from the HPS viewpoint. However this is just an example - the spredsheet needs to be fixed and everyone need to do this math for their own gear anyway (and if you're too lazy, TBH I don't care if you use inaccurate numbers as a result, as you purposely chose to do so).

This puts kinda brings my trust back into blizzard's item balancing as +healing is no longer hands down the best stat, however it may still be quite better than what most value it for. Let's illustrate with my gear:

For 40% mana to HL and 60% mana to FoL (or 177.3 FoLs and 26.5 HLs, or 83% of the time spent casting used on FoL), the basic spreadsheet AEP system would be (again those values need to be fixed via removing the heals critting for 1.5X on the spreadsheet somehow):
1 +healing = 0.091 mp5
1 int = 0.256 mp5
1 crit = 0.198 mp5
However since +healing also increases HPS and allows us to add more FoL instead of HLs, 1 +healing is actually worth 0.091 + 1/21.44 =~ 0.138 mp5.
So basically 9 healing would be equivalent to 1.24 mp5 making a nigheseye the optimal gem without a shadowpriest. With a shadow priest 9 healing is worth 1.252 mp5 by the spreadsheet so 1.252+9/21.44=1.67 mp5. Maybe if you add a shaman with mana spring and tide (which I didn't include) you might actually reach a point where 9 healing is worth MORE than 2 mp5, however I think that would be unlikely and you can safely say (after the spreadsheet is fixed and you use your own stats, of course!) that nightseye is the best gem for all situations as 9 healing is very unlikely to ever beat 2 mp5, while 1 mp5 is very unlikely to beat +9 healing. This also means that for a yellow socket where the socket bonus is helping you should put a 4 int 2 mp5 gem rather than 9 heal 4 int.

So while my numbers don't change anything about standard gem choices (the 2 mp5 for hybrid gems' stat rounding is just too good apperantely), they do help to get a better picture of which item is bette when comparing different items, or to help you decide when a socket bonus is worth going for or not (as this is a non-trivial decision for any item with yellow sockets and a socket bonus that has any relevance to healing aka int/crit/+healing).


Please verify my calculations are now correct, please
Any chance anyone has a clue how to fix the spreadsheet to allow you to modify the crit value? I'm not very good at messing with other people's excel work, I barely get my own straight.

Also if you're replacing HL9s with FoLs instead of HL11s, assuming HL9 gets full coefficient (taking untaleneted base HL9 from untalented character screen rather than wowhead for average healing) the HPS difference is much lower and you only need 3304 +healing to make up for it, making the HPS from +healing much better if you're mostly using HL9 to make up for lack of HPS situations rather than HL11, but still far from the previously calculated value. And again wether BoL gets +12% talent healing or not and how much healing my FoL missing (using wowhead's lvl66 rank7 foL instead of lvl70) can change results a bit.

Gal, you make a mistake by saying that 4417healing will cover the gap. Since adding this healing you increase the effectiveness of HL as well. If you would check how much healing you will need to make FoL and HL equal it will be around +12k.

Thats was also major flaw in your previous posts. Even if 800 healing will cover the HPS difference by moving to pure FoL casting, it will cover the HPS difference between OLD healing with HL/FoL and NEW FoL. If you simply add 800 healing and move FoL you would have the same HPS as without adding 800healing using old tactics. But if you add 800 healing your HPS using old tactics will increase and will still be better than pure FoL, despite the fact that difference will be less now.

You got the whole idea wrong. I'm not trying to find a point where FoL and HL do the same HPS. I'm trying to find how much efficiency you gain from +healing due to HPS by allowing you to cast less HL and more FoL while keeping the same HPS as before. With 4400 +healing to do the same HPS you can spam FoL instead of HL which is equal to the benefit of 206 mp5 (again, recalculate for your own gear of course and accurate formulas). So if the spreadsheet says 4400 +healing is equivalent to X MP5 for you, it is in fact worth X+206.
When upgrading gear you're comparing your new state to your old state. Since the HPS you're trying to maintain is %HLXHL HPS + %FoLXFoL HPS, having additional X HPS allows you to have less HL% and more FoL% to keep the same HPS, thus saving mana.
That is assuming you only HL when you *have to* and prefer to FoL spam when you don't, as we all know that if there is always someone to heal with FoL and you never reach a point where "he'll die if I don't HL now", FoL spamming will be significantly more healing done in the fight than HL. Of course dead people can't be healaed and thus sometimes you use HL to save them. The amount of times you have to HL in a fight will go down as your HPS increases, thus mana is saved and more total healing can be done.

EDIT: I still don't get how you got to 12k. Even with healing->infinity, HL HPS = a + healing*2.5/3.5/2=a + healing*5/14, while FoL HPS = a + healing/3.5
HL HPS / FoL HPS with healing->infinity is then lim(healing->infinity)HLHPS/FoLHPS = 5/14 / 1/3.5 = 1.25. This means that with infinite +healing HL will still have 1.25X the HPS of FoL. Since with 0 +healing HL is way more than 1.25X the HPS of FoL and both increase linearly in HPS with +healing, it is safe to assume that as you increase your +healing the HLHPS/FoLHPS goes closer and closer to 1.25, but never less. Therefore I can tell that with 12k +healing HL will also be more HPS than FoL, as well as any other amount of +healing.
Again the assumption is you use HL only when FoL is insufficient HPS, so if you make FoL sufficient HPS you won't have to HL as much. As you can see with how it changes itemization values that this difference is kinda small, but it's there and can affect certain gear choices (though doesn't seem big enough to make or break the difference between rare gems at least with the overpowered roudning of 3 mp5 / ~2 -> 2 mp5).

At the end I'll probably keep working on the model where I reach a point where I see 2 items I can choose between that are just too close in value, thus improvements will be needed to increase its accuracy and choose the right one if one item is way better than another by an approximate model, there is no reason to make an accurate one. However the basic spreadsheet model overvalues crit and undervalues +healing, and that's more than just inaccuracy, and can possibly make the difference between certain items and maybe even the epic gems at which I hadn't looked yet.

I thought I should point out--my result was completely unexpected, and supports Gal's proposition.

If someone else can do the same math that I did so as to verify, that would be great.

Does anyone know if the gear sheet has been updated yet?

Or a link to an updated one I can check out?

The reason you got such different values is that you took 0 healing 0 regen paladin, I did it on a "standardly geared" paladin, which makes a lot more sense.
All my approximations I ever do around gear upgrade values I do "around my current stats". As the value of all stats change as you change your gear, the most accurate results would be when working around your gear level, and rather accurate when working around someone with gear not far from your gear level. However taking a naked paladin is very far from anyone's gear level, and thus those numbers are extremely unrealistic. Once you stacked your 1800 +healing the values you calculated will change dramatically and be much closer to mine, and what we're really looking for is "I have this much healing and this much mp5/SP rege/totem support etc etc. How much healing do I need to increase for it to be worth to drop how much mp5/crit/int?" That would be answered by looking at the current state of his spells, not the naked paladin's spells.

Just a few things I may (or may not) need to fix:
-BoL affected by talents?
-HL9 gets full bonus of +healing?
-Do the same math but assume HL9 is more than enough HPS most of the time rather than HL11.
-Find a model that would actualyl allow you to say FoL is enough HPS X% of the time, HL9 is enough HPS Y% of the time and HL11 is enough HPS Z% of the time and then tell you how much mp5 you would gain from increasing HPS in that case. Again assuming no other ranks are being used but I think that's a much more reasonable assumption that assuming one uses only max ranks. Note that adding HL9 into the mix will probably increase the value of HPS, although I can't be sure as HL9 is not only closer to FoL's HPS, it is also closer to FoLs mana efficiency, so the end result really neeeds to be calculated... And modeling 3 spells is going to be even more complicated.

I don't think this is true, it seems to use the cast time after Light's Grace to calculate the hasted speed. It might be new in 2.3, not really sure since I didn't really have haste gear pre 2.3. But with 60 spell haste without lights grace up HL cast time is 2.41. After casting a HL and getting light's grace the new cast time is 1.93, not 1.91 which is what was expected.

Which is consistent with calculating the base cast time after Light's Grace. 60 Spell haste is 3.8% haste.

2.5 / 1.038 = 2.41
2 / 1.038 = 1.93

It is relatively minor, but it makes the weak stat even weaker.

if what you say is true, its new in 2.3, sadly. This was a fact pre-2.3, I tested it extensively myself and a few others did as well.

I can confirm this, haste applies on the buffed speed since 2.3 .