That was my point. The OP states that quantum mechanics "allows us to construct", not only describe.

Incorrect. Lets say I grant you that Shockley and Bardeen didn't know anything about quantum mechanics and constructed the first transistor purely due to "observable experimentation" (whatever that means), that idea is completely false today.

I use and write atomistic simulators and semi-classical (finite-element analysis based) simulators to simulate modern FET devices every day as part of my research. I literally use quantum mechanics to write the algorithms behind the simulator (DFT, NEGF in my case). Moreover, if you look at the manual for Synopsys Sentaurus (which is the semi-classical simulator), they have hundreds of pages describing the derivation of their mobility models and structural models from basic quantum theory.

In summary, modern transistors are most definitely constructed taking quantum mechanics into account because without it you wouldn't have transistors functioning at 90nm much less at 14nm. You wouldn't take into account gate-leakage, tunneling, stress, bandstructures, and a host of other properties of modern transistors.

Read the early experiments with crystals (not wikipedia, which most deals with stuff post-patent). They didn't start with physics textbooks. They started with odd electrical behavior observed in practical experiments.

I already granted you that.