Crit has diminishing returns, power does not. You want to get your crit around 35% (including class buff), and then stack the rest in power. Power is not far and way inferior to main stat (willpower in this case), but I do agree that it is better for sorc dps (healing is probably still quite debatable). Power adds more to bonus damage than main stat, albeit very slight.
0.20 bonus damage from each point of main stat.
0.23 bonus damage from each point of power.
However, there are couple things to keep in mind. With the 5% sorc buff and 6% tree buff it puts bonus damage from main stat around 0.22 for each point. Also your main stat increases your crit %. And this is calculated separately from your actual crit rating so it does not run off the same diminishing returns. Sure you could squeeze out a little more bonus damage, but the benefit to crit outweighs it.
I think I read you would have to have main stat around 9000 to start seeing power outweigh it. Again though, the difference is very insignificant.
As for healing, it's 0.14 bonus healing from main stat and 0.17 bonus healing from power (I think these are the right numbers, someone correct me if I am wrong). Again, you have to equate the buffs and the bonus to crit from willpower. I see a lot of healers stack power over main stat though. (I don't heal on my sorc anymore, but on my Operative I tend to lean towards cunning for the extra crit for diagnostic scan)
You want your crit around 35%, surge around 75% (diminishing returns start hitting pretty hard around there). The rest into power and alacrity. Alacrity is a little more important for lightning and healing sorcs than madness sorcs. I had my surge up around 78%/79% for a while with madness. I've since brought it back down to increase alacrity for force lightning, which is >50% of your rotation even in madness.
Just wanted to point out that power does indeed have a diminishing return. It is a flat damage increase so it has a mathematically built in diminishing return. Other than that if you don't want to delve too much into the math involved (honestly i did it and the effort vs gain rate is really low) then what you said is spot on.