With the way spell power, crit and haste are dependent on each other, increasing the amount of one stat (e.g. crit) makes it relatively less valuable to add yet more of that stat, compared to adding more of the other stats (haste or spellpower). More details in Scaling F.A.Q.

This fact makes constructing Best Raiding Gear Available threads to be seemingly useless tasks since static scalars hardcoded for the specific BRGA will change as soon as you swap around several pieces of gear which will invalidate gear values. However, BRGA lists are very guick and concise way of making gear decisions, so it's a pity to get rid of them.

Here's the solution suggested by SimCraft's development team and manifested in http://code.google.com/p/simulationcraft/wiki/LatestReport:

To match comprehensiveness of SimCraft and acceptable performance, we'll be providing scalars for specific gear points, multiple for the expansion, i.e. Naxx starter kit, Naxx-25 geared and so on. I sincerely believe it's a very fine compromise, that makes calculating scaling values for every single set of gear unnecessary, and still yields quite precise results.

The point of this research was to illustrate that scalars do not change enormously due to, say, wapping 200 haste with 200 crit. This (if proved) provides rock solid basis for the approach declared above. Here are the results of a similar experiment (data mined by Battlemaid, comments by Karnor):

priest=Shadow_Baseline
gear_spell_power=1300
gear_crit_rating=275
gear_hit_rating=135
gear_haste_rating=100

priest=Shadow_Rehashed
gear_spell_power=1200
gear_crit_rating=475
gear_hit_rating=135
gear_haste_rating=0

Shadow_Baseline  : avg=1899.3, spread=126, conf=0.7, scaling=0.005
Shadow_Rehashed  : avg=1864.5, spread=138, conf=0.7, scaling=0.005

Going from test 1 (baseline) to test 2 (rehashed) we lose a bit of spellpower and haste but gain crit. So following the logic above (the very first paragraph of the page) I would expect to see the scalars for spellpower and haste to increase a bit while the scalar for crit decreases a bit. And thats exactly what we see.

The value of spell hit (up to the cap) depends on how much dps you already have. The more dps you have, the more expensive a miss. And indeed test 1 has slightly higher dps and a correspondingly slightly higher hit scalar.

Let's say we have two necklaces that are quite relevant to the gear level described in the above experiment (let's ignore stats like spirit or intellect and focus directly on main caster stats comparison, also let's assume that inspired by new crit changes we socketed PoS with a +10 spellcrit gem):

Pendant of Sunfire
Yellow Socket (Gemmed with Gleaming Lionseye)
Equip: Improves critical strike rating by 25+10(gem)=35.
Equip: Improves haste rating by 25.
Equip: Increases spell power by 34+2(socket bonus)=35.

Brooch of Nature's Mercy
Equip: Improves haste rating by 33 (2.09% @ L70).
Equip: Increases spell power by 40.

Below we see that with equipment rehashed difference between necklaces got smaller because crit which is a major stat of PoS got discounted a bit, while strong points of BoNM both got more useful.

Now let's compare stuff that is drastically different but still appropriate to the analyzed gear level (let's ignore stats like spirit or intellect and focus directly on main caster stats comparison, also let's assume that inspired by new crit changes we socketed SoS with two shiney +10 spellcrit gems):

Footpads of Madness
Equip: Improves haste rating by 25.
Equip: Increases spell power by 50.

Slippers of the Seacaller
Yellow Socket (Gemmed with Gleaming Lionseye)
Blue Socket (Gemmed with Gleaming Lionseye)
Equip: Improves critical strike rating by 29+20(gems)=49.
Equip: Increases spell power by 44.

Now we see that difference gets changed in much more drastic way since these pieces of gear are literally antipodes of each other. Tho the magnitude of this change remains the same.