Has anyone successfully used the spell complexity rules (or a variation thereof) from Chainmail in their OD&D campaigns? It's something I've long pondered and am considering testing in the next session of my Dwimmermount game this weekend. The gist of it is that, when a spell is cast, a roll is made to determine if the spell is cast immediately, is delayed, or fails entirely. In addition, there's a chance the spell may or may not fade from the character's memory upon being cast, introducing the possibility that a spellcaster could cast, say, charm person successfully and then still have access to that same spell for another go rather than having to re-memorize it.
That may sound like a serious buff to the effectiveness of spellcasters, especially magic-users, but there's no guarantee of this. Moreover, the possibility of a spell's not fading from memory is counterbalanced by the chance that a spell could be delayed in casting or fail entirely, outcomes not possible under straight OD&D rules. I am intrigued by spell complexity because I think it'd go a small way toward making magic more unpredictable and less exacting. That's a good thing to my mind. I've never been particularly fond of D&D's "scientific" approach to magic, even if it does make the game run more smoothly than other options. The notion that even a veteran spellcaster can't be certain that a spell will function as planned -- or at all -- is something I think adds great flavor to the game, but does that flavor come at the cost of useless complexity? That's what I'd like to know.
If anyone's used Chainmail's magic rules in their OD&D games, I'd love to hear about your experiences.