Forecasting newsletter #3/2025: Long march through the institutions — LessWrong