- the main result is demonstrated in a minimal model with two policies;
- the definition of
wrong gameis relative, not universal; - the current optionality dynamics is compact and useful, but still simplified;
- the model does not yet capture endogenous exit menus or multiple competing regimes.
- the relation between reward and permanence is modeled by a simple logistic form;
- local competence is treated as a narrow proxy, not as a general theory of intelligence;
- sunk cost, reputation, identity, and coordination do not yet fully enter the demonstrated core.
- the current grid is bidimensional;
- the slices concern regime structure, not empirical calibration;
- the outputs are reproducible, but not yet adjusted to real domains;
- the numerical interpretation confirms coherence with the model, not general empirical truth.
- the intellectual differentiation is already reasonably well intuited, but still needs a finer comparative matrix for stronger shielding;
- the project still does not have a literature review extensive enough to support an aggressive priority claim.
These limitations do not invalidate the work. They define its status honestly: a strong theoretical core, publishable as a preprint or working paper, but still expanding before a more ambitious final version.