#1: People say using shields against spells is in? Sounds good. I thought Oblivion relied a little too much on manual dodging and aiming that had little to do with your character's abilities (there was dodging with acrobatics, but just strafing seemed as good if not better). Giving warriors an alternative to this with blocking is nice.

#2: I'm not fond of adding cooldowns. Games that use this usually have a bar with clocks that show when the spell will become available again. It works in a more RTS-y game like Dragon Age but in this one I'd rather be able to keep my eyes on the world instead of on meters.

#3: Possibly. It's used in other games but unlike those you can't hide your wizards behind the fighters so everyone needs to be a front line fighter to some extent. Maybe limit this to power attacks that take more time charge up so a fighter can't just overwhelm any mage by slashing repeatedly.

#4: I can think of one change: if a spell's duration doesn't matter, then you shouldn't be able to reduce the spell cost by making it a low duration. For example, fortify skill if the skill uses a seprate menu where time doesn't pass.

#5: I liked this idea. Both for the balancing aspect and what it would add to the appearance of combat, seeing someone use an item instead of just making some sparkly lights. And I figured it could add a fun tactic if hitting someone at the right time could make them drop their potion and you could take it and use it yourself, similar to disarming attacks in Oblivion.

Here are some items from my ideal magic system:

1. Spells will have up to four modifiable variables (not including Self, Touch, and Target):

A. Duration - Length of time the spell will be active
B. Range - Area affected by the spell while it is active
C. Potency - Spell's power shown as either X or X to Y depending on spell type
D. Casting Time - Length of time the spell takes to cast before it becomes active
There will be a hard cap of "spell points" that no spell can exceed (not counting pre-made spells or ones added via the CS). Self, Touch, and Target spells will have different caps available. Duration, Range, and Potency will add spell points until limited by the cap and casting time will reduce spell points (allowing more duration, range and/or potency). You can also add negative self effects which will increase the hard cap of the spell points.

2. Spells can be Cast or Channelled

Cast spells must wait for the Casting Time to complete before becoming active. Channelled spells always begin instantly and will run for as long as the cast button is held or until the player runs out of magicka. Channelled spells cannot be used while in motion and damaged received while channelling them will also decrease the user's magicka and/or fatigue.

3. Fatigue and skill level will determine spell failures.

A spell failure will not be an all or nothing ordeal, rather if you have low fatigue or are not skilled then you have a chance to have the spell output a decreased value. Your fatigue affects the degree of failure and your skill affects the rate of failure. Examples:
A. A mage with full fatigue and 100 skill would fail about 1% of the time and one modifier (ABCD above) would be reduced (or increased if D) by 5%
B. A mage with full fatigue and 25 skill would fail about 50% of the time and one modifier (ABCD above) would be reduced (or increased if D) by 5%
C. A mage with no fatigue and 100 skill would fail about 1% of the time and one modifier (ABCD above) would be reduced (or increased if D) by 75%
D. A mage with no fatigue and 25 skill would fail about 50% of the time and one modifier (ABCD above) would be reduced (or increased if D) by 75%

4. Magicka would have a nearly non-existant recharge rate, but all mages (or all players) receive a scaling channelled spell that recharges magicka.

Something like "While being channelled, 2% of your total Magicka is restored every second". This would be great for post-combat recharges and could be used during combat in emergency situations for early level players (before they can use or can afford a real battle-regen spell).

5. Stacking spells and enchantments will suffer from diminishing returns and in no event will they be able to achieve a 100% rating.

Example: Equipping a ring of 50% Reflect Spell will give you a 50% chance to reflect spells. Equipping a second ring of 50% Reflect Spell will increase your current spell reflect rate by the amount of the enchantment (50% of 50% is 25% for a total spell reflect of 75%). Equipping a necklace of 50% Spell Reflect would grant 50% of 75% (37.5%) limited by the hard cap for Spell Reflect (which should be in the 90% range).

6. A new spell called Armor should be added which increases the user's armor by a percentage (Shield spells will still increase the user's armor by a specific amount).