Do random graphs with directed cycles generically produce edge currents?

I have not thought about this much, but I liked Evelyn’s work a lot and have recently been wondering under which circumstances one can have a detailed balanced system that still has topological signal processing