(In my field, mathematics, mostly graduate-level, in the U.S., at R1 university) although this may be marginally legal, and does have some constructive features, it also has many failings. An even more entrenched situation is that slightly-more-advanced grad students (or, worse, those whose command of English is too poor to be in a classroom as Teaching Assistant!) are assigned as "graders" for homework and exams for graduate-level courses... indeed, to "spare" the instructor (whose raises and status are based on "research", not "teaching") the burden.
After some decades of wanting to believe that this system could work, I have eventually come to see that it does not. Thus, if nothing else to show that I put my effort where my rants are, I do give frequent feed-back, comment-on and return promptly homework, ditto exams. Yes, I do also discuss (an anonymized form of) foibles I see in homework and exams. In particular, this system (while much more effort) is nearly-infinitely more informative about the actual state of the grad students' minds than if I merely had numbers reported to me. (And surely vastly more useful to students than much-belated or cryptic grading schemes.)
Also, again based on some decades of observation, the critiques that students could/do give each others' work (even with the best of intentions) is ... duh... quite naive. At best, literal "logical correctness" (already perhaps challenging) is not usually the same as "conceptual optimality", which is really the high-end goal for people who want to be professional mathematicians. That is, it is possible to be "correct" but, nevertheless, so suboptimal as to be "wrong" by any utilitarian criterion.
So, yes, it is good to rhetorically ask students to appraise things their peers have written, but with me (e.g.) as intermediary, not only "blinding" the consideration, but also guiding the appraisal in a way that not only accomplishes the immediate task, but is robust and re-usable, and perhaps not completely-obviously-so to the novice.
A related issue is that "prelim problem solutions" are often put on-line by grad students with good intentions, but in many cases (and not easy to distinguish by beginners) these solutions are very awkward and unconvincing... but, ironically, intelligible to other beginners. Meanwhile, often, what we (=faculty) want people to learn is how to do a more sophisticated thing, as opposed to muddling-by with a crappy technique.
(And, another irony, often the most assiduous people are the ones who accidentally believe that they can or should resolve all issues with their bare hands, rather than benefitting from a few centuries of experience of many quite able people. And can often-enough muddle through, and feel exhausted, thus reasoning that this must have been a virtuous action.)
