As a graduate student of physics in the early 2000's, our particle physics classes started with quantum field theory, since QCD had long been established as a good model of the nuclear strong force. We sometimes talked about the S-matrix, but as a part of quantum field theory, not an alternative to it: we evaluated Feynman diagrams to compute S-matrix elements.
Recently, I've been reading a lot about the history of the field. In the 1960's, the S-matrix theory of Wheeler and Heisenberg was advocated as an alternative to quantum field theory, one that didn't make assumptions about space-time being a meaningful concept in the intermediate states of a quantum interaction. The troubles they had understanding renormalization and their lack of a field theory model for the strong force would have made it tempting for mid-century physicists to weaken their assumptions about space-time in nuclear interactions, but since the QCD model was discovered and verified (from the observation of gluon jets to the precision of today's lattice QCD calculations) and renormalization is much better understood today (RGEs, etc), those weakened assumptions no longer seem to be necessary. We can be bold in asserting that space-time is a classical manifold on nuclear scales (~10⁻¹⁵ m), just as it is on human scales (~1 m).
Agnosticism about what happens between the incoming free particles and the outgoing free particles may still be relevant at quantum gravity scales (~10⁻³⁵ m) and I hear that S-matrix theory is an important part of modern string/M theory. Frankly, it makes more sense for the notion of a classical space-time manifold to break down at the Plank scale than at the nuclear scale, anyway.
Am I correct in understanding S-matrix theory as a weakened form or subset of quantum field theory—that is, quantum field theory is S-matrix theory plus additional assumptions/claims? I'm asking if the "S-matrix theory" of the 1960's might have also made assumptions/claims that quantum field theory refutes, which would not make it a strict subset.
If it didn't, then S-matrix theory isn't really wrong, is it? It's just incomplete.