I'd argue that a lot of those subjects are actually relatively cheap to run, and therefore often subsidize the more expensive ones. There's no requirement in tuition fees to link the price the student pays to the actual spending on education, so many universities used this extra funding to pay for expensive improvements and research in their science education. It's one thing to say that humanities students should fund their own education, but they're quite often subsidizing the education of students studying more expensive courses, which is what the government should be doing. If a degree is, as you put it, less valuable, then it should be cheaper, but we don't see that, because student loans create a market where universities will almost always simply charge the maximum allowed.
Also just to point out, it's not this government that writes off half of student loans. That cost has been passed on to a future government.
So what courses in particular would you identify as needing subsidizing? I mean there's the obvious like medicine, nursing and other public sector jobs (interesting you point to sociology as an unworthy course when we have 122,000 unfilled vacancies in social care in the NHS, and that's often the first step to getting a professional licence). But there are plenty of others where it's far more difficult to predict, because you're looking five years into the future to decide what is worth studying, but ignoring the fact these individuals are choosing something that's hopefully going to provide them with an income for the next 40-odd years of their life. And the reality is that none of us know what that will look like.
It's also worth mentioning that often you don't have a shortage because you don't have an industry. In Sweden, for example, all youngsters are entitled to 230 hours of publicly-funded music education and out-of-school education is subsidized too. And nowadays, American singers and bands are falling over themselves to record in Stockholm and are increasingly getting produced by Swedish producers. They're only behind the US and UK in the size of their music industry, despite a population of just 10 million. But then you combine this with high levels of education in technical fields and you get things like Spotify, that are able to work with Swedish musicians to further grow the industry. It's only through this collaboration of a range of highly educated people from a range of disciplines that you can build successful industries. They didn't fund music education because they identified shortages in a music industry and needed to fill them, they have a music industry because they decided to fund music education.
When I was choosing subjects for university in the early 2000s, no-one would have suggested linguistics as a degree choice. It was all computing, technology and engineering (when isn't it?). And yet now technology companies are falling over themselves to hire trained linguists because it turns out that a background in programming and engineering doesn't help when you're trying to create technology that understand's people voices and can communicate effectively with people. It wasn't IT engineering that turned Apple into the behemoth it is today, it was industrial design and business acumen combined with good engineering. It's at that intersection between the disciplines that you often see the best innovation.