This seminar will demonstrate that the dominant source of uncertainty in long-run demographic projections is not data quality or parameter imprecision, but model choice itself.
In the late 1970s, Song Jian — a Chinese missile scientist trained in Moscow in control systems theory — encountered Dutch mathematicians who had adapted missile guidance equations to population dynamics. He took the framework back to China, applied it to demographic data, and produced projections showing China's population exploding past four billion by 2080. Those projections drove the one-child policy, adopted in 1980, with profound and lasting consequences for hundreds of millions of people. What is less known is that the model was structurally insensitive to its own inputs: an uncertainty analysis of Song's framework reveals that it converges to its target of approximately 700 million Chinese by 2080 along slightly different trajectories, but almost regardless of the actual values of its uncertain parameters. The 'answer' was not derived from data — it was embedded in the model architecture. A normative target had been laundered into a mathematical result.
This talk argues that the Song Jian episode is not an anomaly but a paradigm case. It shows that the dominant source of uncertainty in long-run demographic projections is not data quality or parameter imprecision, but model choice itself. Across a multimodel analysis, projected global populations for 2075 largely diverge, while fertility parameter assumptions account for only a tiny fraction of output variance. The number you get reveals more about the model than about its future.
The implications for the governance of knowledge are direct. When a single modelling framework is elevated to the status of official forecast — as in the Shared Socioeconomic Pathways underpinning IPCC climate scenarios — structural modelling choices are rendered invisible, and the "garden of forking paths" in demographic forecasting is closed off before the policy conversation begins. The speaker will discuss what sensitivity auditing and robust uncertainty quantification can offer as correctives: not to paralyse decision-making, but to make the assumptions embedded in demographic knowledge legible, contestable, and — where warranted — revisable.
Speaker bio:
Samuele Lo Piano works at the Department of Statistics and Econometrics at the Faculty of Management and Economics, Gdańsk University of Technology (Politechnika Gdańska), Poland. His research lies at the intersection of uncertainty quantification, sensitivity analysis, and the politics of modelling, with applications spanning energy policy, composite indicators, and demographic forecasting. He is a leading contributor to global sensitivity analysis, and has collaborated extensively with the European Commission's Joint Research Centre, the European Environment Agency, and DG REGIO. He teaches graduate courses on uncertainty and sensitivity analysis, forecasting, and statistics.
Register