You're quite correct. My bad. How is taking a 2 stress mental hit per spell for a theoretical +1 to hit unbalancing again?
It's not. What is inherently unbalancing is the following scenario:
Barry Fresden, hypothetical wizard, has, after all is said and done, a +6 for channeling power and a +3 for controlling it. He designs a rote that is a 7-shift fire attack, call it Weapon:7 against a lone target. He automatically rolls an effective +6 to control the power, but let's say he flubs the attack roll and ends up with a +1. By your "backlash raises the attack roll to the control roll" reading, one extra mental stress hit gives him a +6 to his attack roll.
Under your scenario, it's actually more beneficial to slightly overpower your rotes, as a single mental stress hit could give you a potentially massive boost of attack accuracy.
Hmmm, that's the one reference you've found that would seem to support your point. The other references still seem to bring some support for mine, though. I still say we need the developer's answer to have anything definitive.
I'm really not seeing anything that supports your interpretation at all. If there were a mixed example of fallout and backlash where the attack roll was treated as the difference between the backlash and the fallout , maybe, but all you've cited is an attack spell being treated as an attack with a number of shifts equal to the attack roll, and that really doesn't support any kind of attack roll boosting with backlash.
Put another way, backlash keeps the power of your spell from going haywire. It doesn't do jack to help you
actually hit. That's why shifts of power give evocations a Weapon:X rating, not a flat boost to attack rolls.