Forecasting Newsletter: May 2021 — LessWrong