The predecessor to this was running one hash function per cluster member and picking the one with the highest or lowest result.

Personally I still find that a lot easier to reason about. Especially when it’s time to resize the cluster.