This is a pretty neat idea. But let me ask you this. Would this fatigue mechanism interact with other types of characters or just vancian wizards? Does the fighter suffer fatigue as well for fighting all day? What if you have a 4e style wizard in the group with at-will powers? how would this fatigue system interact with a toon designed using the AEDU framework?
Sure, you could do something similar for the fighter--assuming that the fighter has some analog to 4E powers and/or 3E barbarian rage, that are special things he can only do so often in the fiction and the mechanics. Though part of the point of such a mechanic is to encourage the characters to act in the niche they are intended for. So, if for example, you
want the fighter to be the guy that just keeps on an even keel all day long, then you'd avoid such a mechanic for him, or use a different one.
But as I said before, I don't think my example is a particularly good one other than for explaining what I had in mind. It's just too crude and obviously tied to "make the wizard act this way" to work well. There is a reason why you don't see negative feedback loops done (at least on purpose) in games very often, and the difficulty of making them I think is the main reason why.
As far as interaction with other characters by the wizard mechanic itself, you could use several means there. I chose group fiat--i.e. consciously change the range and threshold if the wizard is not performing the way you want. But for groups that didn't mind it, you could make this more mechanical than fiat if you wanted. And that kind of solution might work across classes.
For example, let's say that you want average fights to last about 5 rounds. That's enough time to be interesting, for buffs to matter, to try things. It's not over in 2 with save or die, or dragging out to 10 in mop up. So you build in a effectiveness adjustment mechanic (similar to the previous wizard example) that is based strictly on the number of rounds. Every fight, you start at -5 levels effectiveness. Each round, you go up by 1.
That means, among other things, that you can have save and die kind of effects, because they are much more likely to get results when the fight is dragging out than when it starts. But it is more likely to get to the middle ground so the wizard can use those "finishers" because everyone else is a bit hampered early too. Since when to "pull the trigger" is on the players, it can still be interesting. Of course, a big objection to such a system is that now you've built in a way that all the characters can be mechanically different but dynamically balanced--at the cost of forcing your fights into the same pattern. (Which brings up another one of my pet bugaboos--there has to be fixed structure somewhere, to support differences. Pick your poison; there is no free lunch.)
A lot of games try to do this halfway by mixing up a bunch of elements. 4E, for example, uses several means to make sure you don't have 2 round fights, but doesn't really do anything on the long fight end. (The group can use some sense with their daily powers and handle that side, especially if the DM isn't picking max encounters all the time. But this is not the mechanics pushing you that way, but simply common sense and experience with games.)
Whereas, Iron Heroes, if I understand the mechanic correctly, goes the other way with its "tokens" which are accumulated by small actions, and then enable bigger actions. It biases the fight towards a small, small, small, BOOM! sequence. If the DM and players time the BOOM! part right, they'll get something very similar to what I proposed.
The only direct feedback loops I know of are by reputation--"The River" in Weapons of the Gods (maybe?) and merit/flaw systems where you only get benefits from the flaws when they directly affect you. Those are both narrative techniques.