Evidence Alignment and Transfer Boundaries in Physics-Editable World Models And Automated Program Repair

Abstract

Two distinct lines of inquiry—a large-scale dataset organized around physically editable world-model factors and execution-grounded reinforcement learning with sequence- and line-level reward models—converge on a practical question for physics-editable world models and automated program repair: what evidence is needed before a reported advantage becomes a defensible basis for explanation, comparison, or deployment? The review draws on two focal records and 12 established sources already present in the project evidence cache. Its comparative framework links dataset design, physical factors, and counterfactual editing to downstream questions of temporal consistency and evaluation. Across the evidence base, the decisive issue is alignment: dataset design shapes what is observed, physical factors shapes how it is compared, and evaluation governs whether the conclusion can be transferred. Uncertainty is most informative when reported as part of the result rather than treated as a postscript. The article concludes with a research agenda built around transparent comparators, targeted stress tests, and evidence records that can be reused without overstating causal or practical reach.