I'm not sure if the more generous interpretation of duration cost in thaumaturgy (i.e. you pay once for an entire spell) actually makes thaumaturgy significantly more powerful than it already is. Even assuming that you have to pay for each aspect's duration independently, you could just split the ritual into multiple smaller rituals, thus establishing 6 taggable aspects (or whatever) on your character as the result of 6 small rituals instead of one big one. Each of these small rituals has a complexity of 8, needing about 2 declarations or other means to make up the deficit for your standard wizard, which in most cases is trivially easy to get. Depending on your group's interpretation of how long rituals take, 6 of these may even take up less time than one big ritual with a complexity of 23 (the complexity you get if you only apply the shifts-for-duration once)!
In other words, I think the core problem is that thaumaturgy opens up some abusive options regarding long duration maneuvers for aspects, not that one or the other method of calculating the shifts needed to enhance the duration is to blame.
I have given some thought to that, and I have some ideas to mitigate the damage done. First, you could rule that setting up aspects in a prior scene means that they are not available to tag in a subsequent scene; because you waited until a later scene to make use of them, they have to be invoked with a fate point as normal. If you specifically want to have your aspects be taggable, you can essentially use your ritual to place aspects in a sort of storage, to be released later at your command. This option does not require any shifts beyond the normal duration shifts, and allows you to release them at the moment of your choice. The trade-off is that you must spend an action (supplementary, maybe?) to release the stored aspect, and you may only release one at a time. This prevents the worst abuse of the thaumaturgy system while still allowing it to be very useful in the right circumstances.
Thoughts?