I'm not a fan of GURPS Magic. I didn't like it in 3E, and it wasn't really changed for 4E. In its defense though, it is serviceable, and as a skill-based system, it fits GURPS pretty well. My primary complaint is that each spell is essentially a 1-point purchase, with little incentive to improve that. I'm also not a fan of the vast chain of prerequisites needed for all of the cool spells. I feel there is a place for characters that are one-trick magic ponies. I have other complaints, and I may or may not address them as I work through this.
First up, I don't want IQ+Magery to be the end-all, be-all factor in how powerful a spell caster is. My initial plan is to create a new advantage to replace Magery that will cap magical skill levels, but also represent how powerful a caster is. For now, I'll call the new Advantage "Magical Aptitude." Magical Aptitude will unlock the ability to cast magical spells (so, like Magery, the initial level will probably be more expensive than subsequent levels). I'm not certain yet if I want skill to be based off of this score, or perhaps off of a second Advantage (or possibly attribute). Maybe a "Magical Control" score will suffice. This opens up the possibility to have wildly powerful wizards with wildly unpredictable results, or to have wizards with adequate power, but perfect in their application. Perhaps I'll do both, or a combination (Magical Control attribute, modified by either a fraction, or a root of Magical Aptitude).
In Tyranny, each spell is made up of two or more "sigils." Each spell starts with a Core Sigil (basically, the college of magic), and an Expression Sigil (the range or AoE of the spell). Spells can be further modified by Accent and Enhancement Sigils. Some of the Accents increase the spell's power or range, while Enhancements may provide things like armor penetration. I'll be using Core to represent the element or college; Expression to determine the range; Accent to determine are of effect; and Enhancement to provide modified, or additional affects to the spell. This sounds like a complex re-write, but I think it will trim down the spell list enough to offset itself.
I should probably add something here.