On national labs: I’ve worked with a few, and I have some friends who ultimately ended up working at some of them (personally I was not a fan but I see their appeal). You are correct - the ones I’m familiar with have something like 50% MS, 25% BS or lower, and 25% PhD as the makeup of their research staff. Some of them offer their members benefits that encourage them to take a leave of absence to get a higher degree. And yet, at the end of the day, they are all very strongly pedigree focused, in that the majority of the most politically successful employees earned PhDs. I’ve seen cases where the MS workers took around 20 years to receive any meaningful advancement, with the lack of pedigree being the cause of them having their career put on “slow mode.” I can’t speak for LANL because that happens to be one I am not particularly familiar with, but for the ones I am familiar with, that seems to be the general pattern. An MS gets you employment, a PhD gives you a chance to advance. I would personally recommend a PhD to anyone who would choose to go that route, even if an MS seems like it would suffice, because in the long run it would be more effective.
Everything boils down to money if you really think about it, of course. Research wouldn’t happen either if there were no money. However, I don’t think most people associate marriage or stability with a “purely monetary outlook” on education.
I don’t think an MS thesis is useless or that there’s always a better option, but the MS thesis does represent a sort of awkward middle ground - it’s just barely enough for traditional R&D, but the MS non-thesis is more than enough for most industrial positions, a lot of which could be R&D. In that light, I’d question the wisdom of 2+ years when 1 is enough, or of choosing not to finish the entire PhD. I know a few good reasons to pursue one - to add depth to a resume, to complete a trial run of the PhD, if it’s externally funded with a salary (some national labs do this), or if it doesn’t take 2 years because the Bachelors degree makes it faster to complete.
The fact that MS degrees may take longer as a result of bad advising is a real issue. It’s not always possible to predict this ahead of time, and it does depend on a fair number of factors including quality of advising, how effective the student is at researching the best advisors, the tendency of that field of study to introduce new unexpected challenges, politics, unrelated personal factors, etc. You can no more expect to finish an MS thesis in 2 years than you are to finish a PhD in 5 years - that is, it’s feasible if everything turns out well, but it could easily take longer.
UG research is indeed pretty variable. But if an undergraduate did develop solid research skills in a useful field but chooses not to go for the PhD, there is much less to be gained from an MS thesis.
You seem to be saying that a PhD is not only about learning a specific topic, but also about learning a wide range of broadly applicable skills that apply to not obviously related fields. That’s true, but that’s also very far from the characterization of a PhD as “learning how to learn.” Or perhaps the phrase “learning how to learn” is just so broad that it could mean anything, and it’s just a buzzword that should be dropped.
And the skills you actually learn do matter a lot. Quantum computing is really not all that far removed from fluid mechanics - all the PDEs, physics, computation, linear algebra, etc. are transferable skills that can be applied there. Same deal for the many PhDs who decide to work in quant finance. Going from CFD to something like pharmaceutical research or psychology research, which has very few transferable skills, would be a more interesting transition to consider. Besides, national labs are often unique in that they want lifetime employment, while a fair number of more profit-driven employers would prefer someone to actually have a thesis based on something that they could directly use because a few years of training is expensive.
I agree with what you say that a coursework-only degree does not provide. However, these practical skills are rather readily acquired by having “work experience.” I really haven’t found research-style work in an academic setting to be particularly different from work in an industrial setting. Work experience is as transferable to research as research experience is to work, and the skills are very similar. And in both environments, the coursework provides a base of theoretical knowledge to competently understand the work/research to be done, and the know-how to look for further knowledge. When it comes to pedigree-independent practical knowledge, the difference between non-thesis, thesis, and PhD, is neither trivial nor essential.