Recent advances in machine learning, complex systems, and quantum-enhanced modeling have reported increasing success in long-term prediction and pattern discovery. Prediction is typically understood as the estimation of future states based on past observations, relying on causality, temporal evolution, and statistical regularity. This paper revisits these assumptions from a non-modal perspective. Within this framework, prediction is not treated as inference, estimation, or extrapolation. Instead, it is fixed as configuration fixation. Without introducing causality, temporality, or subject-dependent interpretation, prediction is considered without sequence, recurrence, or derivation. This reframing calls into question the conceptual basis of predictability. This work forms part of a broader non-modal structural framework in which scientific descriptions are examined without reliance on relation, representation, or explanatory structure.
Juza Minamikata (Fri,) studied this question.