Some very good questions there. While I'm replying to your questions, I think this might answer other people's comments and questions as well.

1) Yup, I did not include Penance since, well.. I admitt that's my bad. I always had problems with pinpointing the exact measures of Penance. It just doesn't seem to follow any rules. In the end I cheated and ninja'd DrDamage's coefficient. I hope the author(s) forgive me and I hope this somewhat shameless plug to their addon compensates :)
http://wow.curse.com/downloads/wow-addons/details/dr-damage.aspx
I included Penance in the spreadsheet now and provided a link at the bottom of this comment. There are some other goodies added to answer your other questions.

2) I admitt I have somewhat trouble understanding your question. Let me rephrase some things more in detail. This is for the spells "entirely" (not just DA procs):
1 SP < 1 crit rating < 7/6 SP.
1 spellpower is weaker than 1 crit rating, but you get more spellpower per point of crit rating (7/6 in fact). This is a result of the lower stat cost of spellpower.
Exceptions to this are:
- PoM where 1 SP > 1 crit rating
- PWS where 1 SP > 1 crit rating
- PoH where 1 crit rating > 1 SP (a remnant of the coefficient nerf a while back)

For DA procs only, we see that all spells benefit more from crit rating than SP, including even PWS and PoM which are the big shots of SP.

Keep in mind this is valid for 45% crit and 3700 SP, with said overhealing. For other stats it might yield a different result, though it will be fairly reasonable to assume that for most circumstances:
- DA will still rely mostly on crit,
- PoH will mostly support crit,
- PWS and PoM wil mostly support SP.

The spreadsheet at the bottom can be downloaded and played with to heart's content. Feel free to experiment with some numbers and post any oddities you may find :)

3) The math was based foremost on 'just mathematical balance'. However, in order to obtain more concrete answers (read: understandable) we had to add concrete situations to it. For this we went as far as examining 1 SP, 1 crit rating, 7/6 SP (stat cost), 23 SP vs 20 crit rating. I take it your question is how hybrid gems weigh up against pure gemming?
Although the spreadsheet holds the basic information you need to answer that question, it does not answer this directly. A negligence I owe you my appologies for. I rephrased your question in 2 additional columns (K & L):
K - How much crit rating must a socket bonus yield before a 12SP+10crit gem outscores a 23 SP gem?
L - How much SP must a socket bonus yield before a 12SP+10crit gem outscores a 23 SP gem?

This yields low requirements for almost all spells except PoM and PWS. For some spells it even yielded a negative number. This would suggest that crit is more valuable but this isn't necessarily the case. What it means is that, while 23SP may beat 20 crit rating, the hybrid SP+Crit gem beats the pure SP gem for this spell. How can that happen? Because the 23 SP gem is slightly under budget (it should be 23.3 SP) and the hybrid gem is slightly over budget (it should be 11.7 SP). The difference is very small, but apparently enough for some spells to tip the balance.

Keep in mind however that these observations are only valid under the rigid situation we have examined:
- 3700 SP total
- 45% crit total (including raid buffs, target buffs, procs, talents etc)
- 40% overheal on average on normal heals.
- 50% overheal on average on crit heals.
For different situations, we can expect slightly different outcomes so it is well worth playing with the settings a bit.

The spreadsheet is available here:

The form ate my links.. booh! :-)


spreadsheets.google.com/ccc?key=0AndYOGtXAJobdDNKMmJxRDhzdDZFOTZLTXVFdV8zaEE&hl=en
(add HTTPS yadi yadi)

img186.imageshack.us/img186/3333/calcs.png
(add HTTP yadi yadi)

It worked!