 When in doubt, choose the course that has the most experimental work

You can always learn pure theory later on for free or very cheap from books.

And above all, you can always learn software engineering later on for free, because the programming community is so much more open than any other so far, notably e.g. with Stack Overflow and GitHub, see also: Section "Ciro Santilli's Open Source Enlightenment". Ciro Santilli is trying to change that with OurBigBook.com, but don't hold your breath. But it is increasingly hard to understand why there isn't an university that forces teachers to publish all their notes and lecture videos (which should be mandatorily recorded) with a Creative Commons License, and then let anyone take whichever exams they want for a small fee or for free.

Actually, there is a good chance you will learn to program, like it or not, because chances are that you won't be able to find as decent a job doing anything else.

But there is one thing you cannot learn for free: laboratory work. Laboratory work is just too expensive to carry out outside of an institution.

Basically, if you don't do laboratory work in undergrad, you will very likely never be able to do so in your entire life.

Because laboratories are so rare and expensive, it is laboratories that put you in the best most unfair position at creating world changing deep tech startups, which is why when in doubt, choose the course that has the most experimental work. Yes, you won't be able to achieve those insanely concentrated equities of the early-Internet, as you will need more venture capital to run your company, but those days are over now, deal with it.