Going back to the twinned ray of frost:
It is very simply stated in the twinned spell feat description that one spell is all that is necessary to counter the twinned spell, not two.
However, what I disagree with is the extrapolation that this would mean that the spell is only cast once. It is perfectly rational to explain that the counterspelling option which technically takes place before you start casting as it is a readied action affects the spell before it gets off (before it gets to be metamagicked). Before this is shot down (as I just divined such an explaination a few moments ago), consider my point, which is this:
To have to cast two spells to counter such a spell would completely alter game balance. Regardless of how one ruled on the issue of twinned spell, I think everyone would universally agree that to require a twinned version or two spells to counter the twinned ray of frost would be too unbalancing. I take the excerpt in the feat description to be a clarification and restatement of the original rule that a counterspell can affect any version of the spell in question (regardless of how it is metamagicked).
Moreover, I don't think it is necessarily inconsistent to say that the metamagicked version casts the spell twice and to counterspell the spell you only need one spell to counter it. Afterall it is the
that causes the spell to take affect twice (something that is beyond argument), and the counter spell affects the spell only, irrelevant to any metamagicks.
Along the lines of reasoning for a twinned spell having sneak attack damage apply to both effects of the spell, consider expert tactician.
A character with this feat gets an extra melee attack when an opponent loses their dex bonus. A character with SA dmg applies their SA dmg to each melee attack. So in the case of a standard action, the character in question gets a free partial action where extra SA dmg is applied.
Now take the arcane trickster. If you look among the feats for an equivalent feat for ray of frost, there isn't one, except for twinned spell. This effectively balances the two abilities. Now, that extra melee attack for the mundane rogue translates to a doubling of a spell for the Arcane Trickster.
Let's portray this point with numbers
Take 14th level rogue vs. 4th level rogue, 5th level wizard, 5th level Arcane Trickster.
14th Level rogue
7d6 SA dmg
feats: expert tactician, etc.
During surprise round gets 2 attacks (all else being equal)
Take the A.T.
4d6 SA dmg
feats: twinned spell, etc.
During surprise round gets one spell (all else being equal) - if that spell is twinned, it acts just like the exp. tac. feat, but for magic.
This I think is a good comparison, especially in light of showing how much more powerful the standard rogue is considering only this type of attack. Even with the twinned spell counting SA for each effect of the spell (trying to steer clear of using "second casting of the spell" language so as not to get bogged down in semantics), the extra damage is only 4d6 relative to the extra 7d6 the rogue gets with exp. tac.
Also, I would like to make several preemptive counterarguments:
Haste is irrelevant in the sense that they both get the extra partial action. It is not irrelevant in the sense that whereas the rogue gets one extra attack, the A.T. effectively gets 2. However, this is balanced in the sense that damage output, as shown above is so much less.
Quickened spell is countered by all the ways rogues can gain extra attacks as with 2wf and i2wf. Again, the arguement for haste applies here, that for the quickened spell, you would get 2 effects whereas additional attacks only provide one additional attack.
So initially I would say that exp. tac. is balanced with twinned spell for the purpose of getting the second S.A. damage, especially since the S.A. dmg is so much less for the A.T.
Now let's maximize both and see what happens:
Hasted 13th Level Rogue, 1st Level Ranger
Feats: ambidexterity, two-weapon fighting, combat reflexes, expert tactician, improved two weapon fighting, point blank shot, rapid shot, quick draw
Suprise Round: 3 Attacks all with S.A.
Round 1: 8 Attacks all with S.A. (assuming initiative), taking a -4 total penalty. I checked in core rules only, rapid shot would stack with two weapon fighting (so long as you could draw the weapons as a free action). I do not wish to spend time arguing this point, so if you disagree, just consider it 5 attacks.
That totals (assuming all hit) 77d6 S.A. dmg + weapon base damage + mod's for str, magic, etc.
A.T. from above:
Feats: twinned spell, quickened spell, point blank shot, weapon focus ray, spell specialization - ray, energy sub (so can affect things immune to cold)
Surprise round: quickened ray of frost, twinned ray of frost, twinned ray of frost: for 5 total ray of frost effects
Round 1: quickened ray of frost, twinned ray of frost, twinned ray of frost: again 5 total ray of frost effects
This gives 40d6 S.A. dmg + 10d3 (certainly less than the weapon base damage), +30 (even if the rogue only had a +4 avg dmg. mod, which is extremely low for a 14th level character focusing on melee damage, he would still have higher than +30).
Now consider each character with 2 additional levels (done so we can quicken a twinned version of the spell).
That example, for the rogue, yeilds an extra 11d6 S.A. giving 88d6 damage all else being equal. The A.T. now can twin the quickened versions giving an extra two S.A. applications yeilding an extra 12d6 S.A. for those attacks + 8d6 S.A. for all of the others. This grants a total of 60d6 S.A. dmg compared with the 72d6 of the rogue. Now factor in that ray of frost is a touch attack:
Ray of frost from maximized A.T. has BAB of +7 (or +8 with the second case). Lets say a Dex of 26 (not too unreasonable) with PBS yeilds +17 (or+18) to hit (touch).
Rogue's attacks are at +11 BAB. Let's say Str of 26, Dex 20 (again not too unreasonable) with PBS, +4 magic weapons We have +23,+23,+23,+19,+19,+19,+19,+17,+14,+14,+9
Now for the opponent - let's say they have a flat footed AC of 20 and a flat-footed--touch AC of 12 (again, seems pretty reasonable). So all of the A.T. attacks hit: 40d6+10d3+30 - 190 dmg. If the A.T. were two levels higher, this increases to 260 dmg.
For the Rogue: First 6 are auto hits (I am ingnoring in my calculations the auto miss for rolling 1's). The other attacks yeild .9, .75x2, .5 of total damage respectively. Total damage: 6x(8d6+d6+12)+(8d6+d6+8)+.9x(8d6+d6+13)+.75(8d6+d6+12)+.75(8d6+d6+8)+.5(8d6+d6+12) = 315.375+40.05+109.175=464 dmg If the Rogue were two levels higher, the damage increases by approximately 40 dmg (not going to go through and calculate again).
One can see, that no matter how you cut, it, the rogue still outperforms the A.T. - moreover, if the rogue has B.E. weapons, weapons with energy enhancements, or uses a more normal weapon (not something that does d6 with crit =20x2) his damage also greatly increases. Damage for the rogue is all that much greater when you account for crits as the A.T. does an extra d3+3 dmg on a crit and the rogue can do up to 3d6+36+burst+4d10 for a burst heavy pick. Finally, I did not include any spell enhancements for either character - Because the rogue gets more attacks, anything that increases attack bonus/damage, like divine might, will benefit him more than the A.T. because his dmg yeild will be greater (he hits more often with more damage) where as the A.T. will always hit so the issue is moot in terms of the to hit-dmg ratio. If the rogue got a hold of a potion of divine might and scroll of divine power his attack bonus suddenly increases by 5 meaning that all attacks but his lowest iterive are auto hits - the damage then becomes simply ridiculous.
So after, this extremely thorough analysis, I hope that I have shown that it would not be unbalancing to allow the twin spell feat to apply S.A. dmg twice. I don't think that the rules specifically preclude this anyway. However, for all of the naysayers, this analysis shows why they ought to allow it anyway. In fact, not allowing the twin spell feat to allow S.A. dmg on each effect is unbalancing!