I could (and arguably should) have added "gatekeeping" as another function of the higher-education system, a role that manifests in multiple ways: admissions, graduation, faculty, curricula, publications. Whilst fixing a few typos I thought of adding that but enough time had already lapsed that it felt like a stealth change of excess magnitude. Consider it added here.
What you call a racketeering scam ... is an argument with merits, but it also reflects another persistent pragmatic informational reality: assessment costs of complex informational goods, such as highly-trained higher-education graduates, are high. Universities effectively play two simultaneous roles: both shaping and measuring (and certifying) educational accomplishment. The flipside of this which you note, is a perennial HN favourite topic: how broken the tech hiring process is from both sides. For the employee and for the employer. At another level, it's the problem faced by both VC and founders in startup launches, and for users and marketers in product. Branding, reviews, resumes, certifications, and other shibboleths serve as information-simplifying mechanisms. See Akerlof's "Market for Lemons". Discussed on HN here https://news.ycombinator.com/item?id=18895171 and in context of hiring here https://news.ycombinator.com/item?id=12671290
The resulting process remains ... highly frustrating and unsatisfying. For me as much as anyone else. But it's another of those persistent behaviours that is too enduring to be dismissed lightly. A classic case of Chesterton's fence.
On low-student:instrctor ratio in-person vs. large-scale, mediated, remote, independent, or other instruction: I wish I did have a good citation, I really don't, other than repeatedly encountering the claim as a simple statement of fact in my reading on educational methods: that one-on-one tutoring remains the gold standard to which all other instructional methods are compared. Hopefully someone with better familiarity with the field than I can substantiate this. Example citation in the opening of "Automated facial affect analysis for one-on-one tutoring ..." (https://inc.ucsd.edu/mplab/wordpress/wp-content/uploads/2011...) expressed as accepted fact, and in numerous Google Scholar cites: https://scholar.google.com/scholar?q="one%20on%20one%20tutor...
Reasons are various, but I think the real key is that educational systems aren't merely informational delivery systems, but tune what's being presented, closely and constantly, to how it's being received. This is a point lost on advocates of technologically-mediated educational methods dating to textbooks, audio-recorded lectures, filmstrips, instructional videos, educational software, and remote learning. Efficiency-minded public education administrators implementing 300--1,000 seat lecture halls with overflow televised satellite sections have foundered on this issue. My own experience has shown the distinct qualitative difference between large (100+ seat) lectures and more intimate sections of ~30 students, or more rarely, seminars of 8-15. The amount of focus and attention instructors can offer smaller groups, the personal relations formed, and interactions among students, are far stronger in smaller groups. Size isn't the only factor, but it's a major one.
The matter of what feedback is or is not perceived in online discussion is also huge, addressed in another current HN thread on "The Silence is Deafening": https://devonzuegel.com/post/the-silence-is-deafening and https://news.ycombinator.com/item?id=23728212
Thinking of universities as nothing more than skill factories is another long-standing flaw, and neglects the role of employers themselves in mentoring, training, and development. See the classic divide between the liberal and servile arts: https://news.ycombinator.com/item?id=23038380
I've addressed ancillary elements of my own uni experience some months back: https://news.ycombinator.com/item?id=20959811