As most of you almost certainly know already, spells in Chainmail are rated according to their complexity. There are six levels of complexity, which not at all coincidentally maps on to the six levels of magic-user spells in LBB-only OD&D. An optional rule pertaining to complexity compares the complexity rating to to power of the magic-user casting it (there are, in descending order of power, wizards, sorcerers, warlocks, magicians, and seers). The more powerful the magic-user, the more likely his spells will take immediate effect, while less potent casters have a greater chance for their spells to be delayed until the next turn or to be negated entirely.
This is a rule I find intriguing and one I've often contemplated adding into my OD&D game, but I've never done so, both because I'm not sure how I'd implement it and I'm uncertain the effect it'd have on gameplay. I'm familiar with the original Chainmail rules, as well as adaptations of the spell complexity system in Spellcraft & Swordplay and Brendan Falconer's article from issue 2 of Knockspell, so I do have resources to draw upon should I ever go ahead with the idea. However, I do worry about the impact it'd have on spellcasting in an OD&D campaign. Clearly, it'd make things more unpredictable, which I like, but it might also make magic-users more ineffective as well, unless one adopts the interpretation that only negated spells vanish from the magic-user's memory. In that case, it might be a fair trade-off, I don't know.
Does anyone have experience with using spell complexity on a long-term basis?