In a way it is a meta-science because you need to find a methodology to evaluate which theory, when proven or disproven has the most impact.
I would imagine that the current working method would be to look at how often theoretical physicists cite a specific theoretical result/paper and hence "popularity" is the prioritization mechanism.
If a large cadre of highly educated people find a particular theoretical mechanism sufficiently compelling to dedicate their careers and time to it, then surely that's a good basis on which to develop experiments to try and confirm whether it's right? Because the benefit of proving it wrong means they'll reallocate their resources.
Presumably the economical cost of the experiments are higher at this point then the resource-allocation of that cadre of highly educated people.
It's worth thinking about, but not sure if there is a viable alternative methodology to be found
Experimentally, people continue to hammer away at the Standard Model (and gravity, my specialty), and the paradigm continues to hold. With few exceptions, most ideas from 40 years ago have been put to significant tests and turned out not to be how Nature operates.
From a theory standpoint, without new guidance from us, theorists are forced to attempt to out-think Nature, which is extremely hard to do. Smolin may be bummed, but theorists can take solace in the fact that the problem is extremely difficult.
In the particular case of quantum-mechanics, the day that theorists divine a compelling way that one interpretation of QM makes a prediction that differs from that of another interpretation is the day that an experimentalist starts building a test to find out which one is true.