We don't need the complexity of supporting independent scoring setups for general peer behaviour and per-topic behaviour.