Speaking for the average not-very-useful user, karma seems to be too big an issue
Benny was right. Some things in life cannot be correctly quantified. The usefulness of people is one of those things (some world leaders tried to do that; they are generally considered evil)
To be honest, i'm not sure how the council system works, so forgive me if i'm talking out of my backside. But from what i understand, all people with karma above a certain level (to leave out new-comers or mere plebeians) get to vote. The N people with the most votes get elected. Seems to work fine. Why should the device program use an automated system and bypass user choice? As a human, i can understand qole's work on easy debian is much more useful and hard than the mirror app (no offense to that guy, it's just not... useful). I can also value kathy's non-technical contribution, and give her posts more credit than, say, orangebox 's.
All in all, i say the issue is not about how to quantify karma, but rather what to do with it. If nokia isn't willing to call for votes when they want to give away devices (which might disclose their intentions and blah blah blah), hold double polls when choosing the council: one for the council, and one to make a list of the most useful 300-500-1M people, in order, to be chosen for fun stuff. That way it's objective, as fair as it gets (for all my distrust of humans, i still think they can choose better than computers) and easy to implement. And karma can have whatever algorithm you want, it will be just as relevant as the user's ability to do the electric boogaloo