Anyway, I think the T-schema is useful enough understood as a formal device to get model theory off the ground but that’s about the extent of it’s relevance. As far as a philosophical observation it says nothing.

As far as true goes I do think there is some interesting mathematics about theories that contain their own truth predicate (it’s possible if you admit that some sentences are neither true nor false, e.g., they are meaningless) but this is really a very different issue than the fact that theories are evaluated or refuted as a whole. One is dealing with the possibility for a theory to coherently talk about itself the other is telling you that the terms in a theory are defined by the entirety of the theory itself.

I too am underimpressed with the T-schema.

I don’t think that ‘true’ has any significance in a formal system modeling language. Scientists agree, asserting that theories (rather than sentences) are “evaluated as a whole” (rather than “true or false”).

But, what some have called “emprically minded” philosophers, like Putnam, protest at length about the methods of the scientist. I assume it’s because scientists are trained to actually solve problems whereas philosophers are trained to perpetuate them so that their departments can be maintained.