I do think there is some potential in the idea of people proving they are at face-to-face events (signing each others' zero-knowledge tokens) and then timing/placing those events such that someone can't be in two places at once.
However, doing graph-based Sybil detection is already a hard problem[0], and trying to create an infallible algorithm that also works using homomorphic encryption is maybe pushing beyond the boundaries of known technology, unfortunately.
[0] https://dl.acm.org/doi/10.1145/2492517.2492568
