As you may or may not know (or care), I am a longtime deep frost mage who recently abandoned frost to go fire. I have some pretty fond memories of running around pew-pewing things with my frostbolt and racking up some relatively serious dps in the process. In fact, there was no way of second guessing the potency of deep frost once upon a time. If you didn't think frost could pump out some big damage, I like to think I could have shown you otherwise. Once upon a time.
Things changed though. I believe I've posted at some time about my feelings of distress and disappointment upon reaching level 70 and boggling over how little my damage output had improved over the course of the 10 levels from 60. I never really did any hard math to figure it out... I didn't need to. I think deep down I had a sense of what was going on, so I simply left it behind and moved on.
If I had to look back upon it now and figure out what exactly went wrong with the spec, it might include the following points, what I'd regard as the "truth" regarding what happened to frost:
1. Frostbolt Rank 11 was "Uber"
This was by design. Rank 11 was the rank of Frostbolt that pre-2.0 could only be learned from a tome drop off bosses in Ahn'Qiraj 20. I had the fortune, as a raiding mage, to obtain this about the second week after the AQ gates were opened on the server. I got pretty used to it too. The fact was, it was meant to seem overpowered by design and it is easy to forget that if you'd used it for a long time.
When the 2.0 patch came along however, it became trainable to all mages, so everyone should have had it in the short interim before the expansion came out. Upon reaching level 63 when rank 12 became trainable, the scaling of rank 12 seemed comparatively vanilla since it was in-line with the scaling of ranks 1-10, not rank 11. Again, this was by design and not a mistake, even though it did make rank 12 appear weak.
An important thing to keep in mind is that the spells at a mage's disposal today are normally scaled ones, not the "overpowered" ones available to mages in post-AQ level 60 Azeroth.
2. Spell Damage Coefficients dictate how much damage a spell does far more than base damage does
This is a tricky statement. But I believe it's true. Afterall, if you look at the base damage of Frostbolt rank 13 and think that it actually does 597-643 damage then you're sorely mistaken. My rank 1 Frostbolt also does a heck of a lot more than 18-20 points of damage, make no mistake. This would seem to become increasingly true as the amount of +spell damage that can be acquired via gear gets higher and higher. The amount of damage any spell does is more dictated by its spell damage coefficient than it is the base damage of the spell or anything else.
Where does this leave Frostbolt as spell damage scales upwards and upwards? Not in a good place, really. At least not as a direct damage spell, since unlike Fireball, it is penalized for being a binary spell. In effect, it benefits less from spell damage for no other reason than because it has a slowing effect. As spell damage increases and raid mobs remain largely immune to slowing effects, it will NOT scale properly as a primary raid nuke when compared to Fireball. The differences were negligible at level 60, but this is increasingly not the case.
The only solution is for Blizzard to buff the coefficient of Frostbolt, if they are interested in seeing it remain viable as a high end raid nuke, though there is no indication at this time that they see this as necessary.
A normal 3 second cast time spell has a spell damage coefficient of 85.71%, but the penalized coefficient of Frostbolt at 3 second cast time is 81.4%. At less than +300 spell damage, this is not a big deal. At +600-1000 it starts to show. If spell damage bonuses get to reach amounts of 2-3k, this will be a big, big deal. The price of having a slowing effect on the spell gets more and more expensive as gear scales upwards. It'll be something Blizzard will probably have to look into at some point in time.
3. The -10% Spell Coefficient Nerf hurt Frostbolt more than Fireball
This is silly, because Frostbolt was the weaker spell in the first place. The actual penalty is less than -10% straight off the coefficient though, because it never has a 100% coefficient. -10% off less than 100% is less than -10%, afterall.
The primary reason for this as far as I can tell is within the "Empowered" talents, which improve the spell damage coefficients for Fireball and Frostbolt when taken respectively. At 5 ranks, the Empowered Fireball talent improves the coefficient for Fireball by 15% while the similar talent for Frostbolt only improves the coefficient for Frostbolt by 10% with an additional 5% chance to crit. This allows Fireball to actually overcome the nerf (and pick up +crit elsewhere), while the 5 talent points in Empowered Frostbolt just barely removes the nerf (it actually adds +0.9% net benefit to the spell after overcoming the nerf) and only adds some crit to it after that.
But 5% to crit is pretty nice too, right?? Not really, since it's scaling we're concerned with here and crit is a notoriously overvalued stat. Percental additions or reductions to the effects of gear on your spells are of paramount importance, since they ensure that gear upgrades are actually upgrading your spells... While yes, Frostbolt does benefit from gear upgrades just like everything else, relative to other spells such as Fireball, or even more notably Shadow Bolt (due to the poorly inconsistent application of the mage nuke nerf), your Frostbolts simply aren't upgrading as well...
Of course, all of this is relative to what your goals are and what your gear level is. Obviously, PVP mages typically have no complaints about frost, since the binary effect of Frostbolt is actually... worth the penalty to damage. I'd say that if Frostbolt is performing well for you, that's great. I'm talking primarily about high performace raid environments. But if you stick around long enough and go through enough gear upgrades, I believe you'll begin to see what I'm saying.