I posted this on the WoW forums about a week ago, in hopes of getting a little developer attention, but I'm going to re-post it here for the EJ community to go over my math and make sure I'm not missing anything:
Every source of mana can be boiled down to a set amount of mp5. If a fight lasts a certain amount of time, and you gain a certain amount of mana, you can derive how much mp5 you earned from that source. With critical strike rating it is a little more difficult due to the inherent RNG -- I've got 45% crit fully raid buffed, but I've seen fights where I pulled both 30% effective crit and 60% effective crit. So, for the purpose of these calculations, I will be calculating the mana that you should gain from crit, not the mana you always will.
I am going to be calculating the amount of mana gained by 1 Critical Strike Rating, not illumination in general. Even if these changes go through, illumination will continue to be an amazing talent due to all our baseline crit and the crit we gain from raid buffs. The real question is, should we continue to itemize for crit, and this what I intend to answer.
The value of crit is dependent on your cast rate. If you cast no spells, you gain no mana from crit. If you cast a lot of spells you gain a lot of mana from crit. Therefore, I am going to be using the stats HLPM (holy lights cast per minute), HSPM (holy shocks cast per minute) and FoLPM (flashes of light cast per minute) to calculate the returns you're receiving from crit rating with certain rates of spellcasting.
The basic formula is (3.1):
HLPM * 764.4 (mana gained from a holy light crit) * 0.0002178 (additional crit added from 1 crit rating expressed as a decimal) * 1/12 (amount of mp5 ticks in one minute)
+
HSPM * 474.6 (mana gained from a holy shock crit) * 0.0002178 (additional crit added from 1 crit rating expressed as a decimal) * 1/12 (amount of mp5 ticks in one minute)
+
FoLPM * 184.2 (mana gained from a flash of light crit) * 0.0002178 (additional crit added from 1 crit rating expressed as a decimal) * 1/12 (amount of mp5 ticks in one minute)
equals the amount of MP5 one critical strike rating is worth on this particular fight.
This breaks down to this nice and easy formula:
(0.01387386 * HLPM) + (0.00861399 * HSPM) + (0.00334323 * FoLPM) = mana return from 1 crit rating stated in mp5.
In 3.1, 10 critical strike rating is worth as many item points as 4 mp5. This means when plugging numbers into this formula, if you end up with 0.4 or more, critical strike rating is as good as or better than mp5 on a per item point basis. If you end up with less than 0.4, mp5 is outperforming crit on a per item point basis.
Now, to put some numbers in those holes. As you can guess, the value of crit rating is going to change based on which fight you're on. The more you spam, the more it's worth and vice versa. I'm going to start out using some personal parses, and then I am going to take a couple parses from WMO's top HPS list to illustrate both the normal scenarios and the extreme scenarios.
1a. Thorim Hard Mode, personal WoW Meter Online - Combatlog Replay
21.03 HLPM, 5.52 FoLPM, 2.24 HSPM, total value of 1 crit rating is 0.3295 MP5, or 82.38% as effective as MP5.
2a. Hodir Hard Mode, personal WoW Meter Online - Combatlog Replay
22.92 HLPM, 1.04 FoLPM, 2.78 HSPM, total value of 1 crit rating is 0.3454 MP5, or 86.35% as effective as MP5.
3a. Assembly of Iron Hard Mode, personal WoW Meter Online - Combatlog Replay
21.28 HLPM, 2.94 FoLPM, 2.08 HSPM, total value of 1 crit rating is 0.3230 MP5, or 80.7% as effective as MP5.
1b. Thorim Hard Mode, top raw HPS parse on WMO WoW Meter Online - Combatlog Replay
28.30 HLPM, 4.82 FoLPM, 1.09 HSPM, total value of 1 crit rating is 0.4181 MP5, or 104.53% as effective as MP5.
2b. Hodir Hard Mode, top raw HPS parse on WMO that isn't skewed by JoL WoW Meter Online - Combatlog Replay
27.3 HLPM, 0.00 FoLPM, 5.32 HSPM, total value of 1 crit rating is 0.4246 MP5, or 106.15% as effective as MP5.
3b. Assembly of Iron Hard Mode, top raw HPS parse on WMO that isn't skewed by JoL WoW Meter Online - Combatlog Replay
32.7 HLPM, 0.00 FoLPM, 0.38 HSPM, total value of 1 crit rating is 0.4549 MP5, or 113.74% as effective as MP5.
In short: Depending on how much you spam and whether or not you're trying to cheese a WMO top parse, crit rating can vary between 80% as effective as MP5 and 115% as effective as MP5 per item point with 3.1 numbers.
What happens when you apply the 3.2 changes to these numbers? To do that, we halve the effective MP5 of crit, and the value of one item point of MP5 increases to 0.5. If you do that...
Example 1a becomes 0.16475 MP5, or 32.95% as effective as MP5.
Example 2a becomes 0.1727 MP5, or 34.54% as effective as MP5.
Example 3a becomes 0.1615 MP5, or 32.3% as effective as MP5.
Example 1b becomes 0.2091 MP5, or 41.81% as effective as MP5.
Example 2b becomes 0.2123 MP5, or 42.46% as effective as MP5.
Example 3b becomes 0.2275 MP5, or 45.49% as effective as MP5.
In short: At best, post 3.2, crit will be worth half the amount of mana as mp5 is per item point spent. At worst, it is worth a third the amount. Also keep in mind that with less mana, our ability to spam decreases, and with less spam we get less return from crit -- bringing this numbers further. In conclusion, I plan on avoiding crit like the plague.