[Disclaimer]Obviously, the optimal dps rotation is to snipe your arcane missiles perfectly; however, the wackiness of the T10 bonus and my low latency gave me a reason to do some number crunching.[/Disclaimer]
I've noticed that the T10 bonus doesn't appear to affect the arcane blast immediately following the MBAM if you somehow clip the cast (stopcasting doesn't work). I did the math to see at what point it might actually be a dps increase to add in a [nochanneling:arcane missiles] modifier to your arcane blasts.
I used a nominal haste rating of 700 to do the cast time calculations, assuming 6% Netherwind Presence, 5% sham-haste, and 3% ret-haste buff.
| w/o T10 | 1.7971 sec |
| w/ T10 | 1.6046 sec |
| delta | 0.1925 sec |
Having a latency of less than half of the delta (correct me if I'm wrong) should be enough to offset the delay in client server communication. The dps time value of this isn't just the difference in cast times between the arcane blasts - how badly you're clipping the missiles also plays a factor. The worse you're clipping your last missile, the better it is, since you're spending less time casting a missile that never fires. As an additional bonus, using a nochanneling macro would make this a nonissue.
Conclusions about whether or not to use the macro are quite nuanced: If you find yourself habitually clipping the last missile in a MBAM *AND* as long as you have less than half the latency of the difference in cast time between 2PT10 hasted and non-2PT10 hasted arcane blasts, you would be better off using the macro. If you don't have a problem clipping arcane missiles, it will be a dps loss to use the macro. If your latency is higher than half the difference in cast times, it's inconclusive.
I should mention that as you stack on additional haste modifiers, the gain in cast time is going to shrink, and the latency required to ensure a dps increase is going to go down as well. If you use it, I'd keep a modifier key bound to cancel the nochanneling condition for when you have additional haste buffs.