What is an instance of "democracy in general" that you think Arrow's theorem doesn't apply to? It will apply any time "society" (i.e. more than one person) makes a choice among more than two options.
It won't apply if you can assign cardinal numbers to the options, but I doubt that's what you have in mind.
That's actually exactly what I had in mind. Range voting[0], for example, doesn't suffer from Arrow's impossibility theorem, Gibbard–Satterthwaite theorem, nor the Condorcet's paradox.
You know, I had a long comment here pointing out many problems with range voting. Instead, I'd like to observe that it really takes balls to defend range voting as "not suffering from the Gibbard-Satterthwaite theorem" when it's easy to show that range voting exhibits one of the failures that the Gibbard-Satterthwaite theorem guarantees in pure-ranking voting systems. Sure, the premises don't hold, but so what? If Gibbard-Satterthwaite did apply to range voting, that would guarantee no other problems than already occur.
Theorem: Hitting your thumb with a steel hammer, instead of hitting the nail, hurts like crazy!
Problem: The pain of a smashed thumb is bad.
Solution: Use an iron hammer. The requirements of the earlier theorem don't apply.
It won't apply if you can assign cardinal numbers to the options, but I doubt that's what you have in mind.