Whereas science needs both hypothesis generation and experimental validation/repudiation of hypotheses.
Hypothesis generation sometimes has to go out there, and invent new concepts that have so far only been thought about, not yet tested.
So to summarize, science needs both creative conceptualization (ontology formation) and experimentation (validation or repudiation of the ontology and/or hypothesis).
These need to go on in circular reinforcement. (Spiral development model).
Experimentation without re-conceptualization will eventually run dry, because it will get stuck in a local-maximum paradigm, and people won't know what new things/aspects to test any more.
Remember, relativity was discovered in a thought experiment by Einstein. Is a thought experiment a real experiment in the article author's view? I doubt it.
Einstein, from the outside, was doing "magic". Speculating about the larger truth.
Relativity was an example of theory creating a completely new set of concepts that were way ahead of the ability to carry out experiments that could validate or repudiate them. It was a well-formed theory, in that it clearly suggested new kinds of experiments that could test it, but it was pure non-experimental theorizing nonetheless.
Darwin also, most likely, happened on his key theoretical insight about natural selection (the simple core of it), by thinking about the generalization of many observations, and having a theoretical insight.
Experimentation has its essential place in science, no doubt (keeps the theorists honest and humble), but it is only half of the game. The other half is innovative philosophy, carefully practiced, in the mind.