The grid scoring code doesn't appear to pick the most interesting talker pairs consistently.
The grid scoring code doesn't appear to pick the most interesting talker pairs consistently.