Look, TFA is showing a model. Not the model. A model. I can apply a transactional lens just to see whether it helps me to "understand" (in fact: to predict) so

Look, TFA is showing a model. Not the model. A model. I can apply a transactional lens just to see whether it helps me to "understand" (in fact: to predict) something novel about the situation. If it checks out, none of your arguments matter. For a model, that's as good as it can get. It's important to have more than one tool in your toolbox and to practice their flexible use. Advertising any model as "true" counteracts that - the net effect is harmful, even when the model is otherwise decent.