We show that (a) the patterns of the climate history as different as reconstructions of Lisiecki and Raymo (2005) and of Clark et al (2025) may both be produced by the same long-memory climate system with different initial conditions or under different millennial forcing, and (b) a single brief perturbation of the forcing, consistent with a possible human impact, may change the dynamics of the global ice-climate system for million years, making the time series we did not observe in the past.

We present a method to separate uncertainties from global and regional climate processes within the Hawkins & Sutton (2009) framework, partitioning sources of uncertainty in future climate projections. This helps interpret sources of uncertainty in future climate projections and informs adaptation planning. Focusing on regional projections reduces total uncertainty, highlighting the value of isolating regional uncertainties when developing regional climate information.