The main problem is scaling the teachers.

Individual 1:1 tutoring with largely Socratic-style dialog with a world-class expert is definitely the most effective and efficient way to learn just about anything. e.g. see http://web.mit.edu/5.95/readings/bloom-two-sigma.pdf

The expert can figure out precisely what the student’s weak points are and can nudge the conversation in a productive direction while allowing the student to feel like he/she is making the conceptual breakthroughs, and the expert will know what misconceptions are common and how to dispel them with pointed counterexamples or logical contradictions, which again the student can be guided through. The expert will know when to offer new topics, when to offer practice problems, when to call back to previous material from a few months before, when to drop the question–response and just give a quick explanation or show a bit of tricky logic/proof that the student is unlikely to ever come up with independently, or when to spend time on hands-on projects.

Student-centered individually-tailored instruction works well if you have <5 students per teacher (and 1:1 or 1:2 is best). When you get to 30 students per teacher, it starts to fall apart, and you need to fall back on lectures, textbooks, handouts, pre-recorded videos, pre-made computer explorations, or the like, and on individual homework and paper quizzes and exams for assessment, because there’s just no way for the teacher to be simultaneously working directly with every student.

The students in a large class can spend some time reasonably productively in small group discussion and the teacher can then visit groups in turn helping dispel misconceptions or nudging the discussion around. Then back in a whole class discussion, the teacher can try to discuss the most common points of confusion. This can’t possibly be perfectly directed at every student at once, but given motivated students and a good teacher with enough out-of-class preparation and grading time, it’s usually the best we can do as a society.

Contrived anecdote: imagine a doctor going through med school being taught to socratically think about how and why various drugs have various effects. It will end up sidetracking you into the depths of microbiology to the extent that you'll never get out of that rabbit hole. At some point, a doctor needs to learn things like "X drug is recommended for Y condition" and just use that rote knowledge, trusting the appropriate authorities. Sure, they can and should be skeptical and critical-thinkers and learning to think socratically is super valuable, but you just can't do it all the time. No doctor can possibly understand all the actual medical details of everything they prescribe, and if they do, there's near-zero chance they still have time to practice all the social skills needed to be an effective practitioner and so on.

Socratic doesn't mean depth first.