Excellent stuff from Ruth.
“Possibility models” are far more important than “role models”.
Original link and Wayback Machine link