[TimPC]

I think STEM is suited to academia in multiple ways: first, the chances of success at learning it seem to be higher in a structured, rigorous program compared to independently. One of the problems with online universities is that they don't apply enough deadline pressure so in many cases only 20% of students who sign up for a course even finish it. If you want to put four years into something and reliably get somewhere then an academic program still seems like your best for that. The other piece of the puzzle is that a STEM degree is a useful credential that can help substantially in the job market compared to being self-taught. Furthermore, most STEM fields have job markets where salaries that can realistically make getting a degree a worthwhile investment are obtainable. I think it's this combination of good way to learn, valuable credential and decent cost/benefit that make STEM degrees make more sense.

[SketchySeaBeast]

> first, the chances of success at learning it seem to be higher in a structured, rigorous program compared to independently

That seems an argument for all education to take place inside academia. Keep in mind that liking to read history books is not equivalent to a university degree in history.

> Furthermore, most STEM fields have job markets where salaries that can realistically make getting a degree a worthwhile investment are obtainable.

Yeah, that's ultimately it, it's seen as more valuable because you can make good money in tech. Though I'd argue that much of that may not be the case - are those with science and math education making money hand over fist?

[TimPC]

Math education is the second highest paid undergrad in arts and science and was number one ahead of CS for a long-time. Many people with math degrees end up in high paying technology or statistics jobs. Math is both an incredibly marketable skill and the program often indirectly selects for high IQ which is correlated with success to a certain degree. You generally aren't going to do abstract algebra or topology for a career unless you are a math professor but being good at abstract algebra or topology generally requires a lot of skills that are useful in the labour market.

As for the main point, I think the best form of education for something does depend on what the outcomes are. If you're getting a degree that doesn't qualify you pay back the costs of that education maybe alternative methods have better outcomes (we'd accept a decrease in educational attainment for cost). If being a small amount better at something is lucrative we should optimize for performance more than cost.