(Just to anticipate, having looked at other closed threads about physics education, I don't think this question should be too opinion based or broad.)
Since being in a lab now for a couple years, I've noticed a problem with the classes we're required to take in physics grad school: Of our "core classes" we're required to take in our first year, 5 of them are theoretical and 1 of them is experimental, while when people end up joining labs after the first year, I'd say roughly 80% of people go into experiment and 20% into theory. From what I've seen at most other schools (barring outliers with theory powerhouses), this is a relatively common distribution. And of the available classes you can take, I'd guess ~95% of them are theory.
I think this is a problem because while I had a little lab experience before joining my lab, I still had to spend (and still do) a ton of time just learning how to use all the equipment and experimental techniques: SEM, TEM, AFM, XRD, photolithography, electron beam lithography, clean room techniques, metal deposition, FTIR spectroscopy, more general optics and electronics design, etc, just to name a very few. And there's no reason to assume I'm alone in this. Meanwhile, while I certainly enjoyed what I learned in my theory classes, I have used about 0.1% of what I've learned in them.
There are a few major counterarguments I can come up with:
- If we did the opposite and trained people almost exclusively in experiment, I suspect there's a fear that we'd just be trained lab monkeys rather than carefully planning experimentalists.
- Undebatably, experiment is more expensive to teach. Theory needs to pay for a professor, while experiment needs to also pay for very specialized equipment.
- Theory is "what's actually happening" or something -- meaning that at the most fundamental level, most experiment is trying to prove/verify some theory, so you need to know how that theory works.
- Taking all these different classes gives you a taste of lots of areas of physics.
- Ideally, theory allows you to know where to explore, even in experiment.
And there are definitely other big ones. But I still have to emphasize how little an impact the theory I've learned in classes has had on my experimental work. The way it worked for me (and what I've heard from friends in experiment), when you join the lab, you spend a little time reading the relevant parts of textbooks and a bunch of papers to learn the theory and history behind the experiment. It's possible I'm not appreciating the base of theory I learned in classes that allows me to read and understand this more specialized theory that's in the papers, but I don't think so -- I can tell you that I've essentially never used anything from Classical Mechanics, Statistical Mechanics, and most of the second Quantum course. Some of the first Quantum course, and a lot of Solid State has been very helpful.
I think a much better solution would be to have almost the inverse distribution of what there currently is: something like 4 experimental courses, and 2 theory ones, at least until you're specialized in an area. The courses would teach you general experimental techniques and design, which I think is better because it would give you an example of what you might actually do in your research. For example, if you're doing experimental research on quantum dots, you will need to know some QM, but what you might end up spending 95% of your time doing in that research is optics and fabrication.
Inevitably, the goal of the education needs to be defined, or this is all meaningless. This is of course also up for debate, but I doubt I'm too far off from what most people would say if I said the goal is something like "to produce effective scientists" (mostly because it's just so general). Again, anecdotal evidence, but a concept I've seen many times now: another guy in my lab clearly had a much bigger emphasis on research and experiment in his undergrad than I did, and while I like to think I see some "bigger picture" or have "deeper insights" into the research, he's just objectively the better researcher.
To put it all in a question format, is there a reason this theory/experiment class distribution seems to be the standard?