Population projections are being revised more often and by more, and migration is the assumption doing most of the work. The ONS's 2022-based projections raised the long-run net migration assumption substantially over the previous round; the UN lowered its long-run fertility path and its peak population; Eurostat, Japan and Korea have all revised toward older, smaller populations. Migration is the least stable assumption because it is the least predictable: it responds to policy, conflict and labour demand within a year, where fertility and mortality move over decades. Statistical offices know this and are responding differently — some publishing variants, some moving toward probabilistic ranges, some holding a single central assumption because users want one number. The evidence that the assumptions are being rewritten is clear; what has not been done systematically is the comparison of each vintage with what followed, which would show which assumptions have historically been wrong by most.