I don't understand where is the problem in accessing students who learn from open courses (I am not saying there is no problem, but I just don't understand it). Most of the grading is done based on quizes, mid terms, terms, term papers etc. All these can be given by these students and you get the assessment.
Testing in post-secondary is usually very idiosyncratic. At best they teach to a textbook and test that textbook evenly. More likely, a professor teaches their angle on a subset of the topic, and then grades based on that.
Because of that, if you understood the 3rd year Algorithms courseware from MIT, you would likely fail the midterm for 3rd year Algorithms at your local university.
Fixing this is very difficult, since it would mean you need standardized tests to measure post-secondary topics. Implementing that (and keeping it up to date) would be extremely difficult and contentious.
I believe it's perfectly OK that you can succeed in Algorithms at one school and not do perfectly at another with the same knowledge.
The area where I believe the problem is in specifically teaching to a test, rather than general deep understanding (of algorithms in this example). Tests in a lot of cases don't test for deep understanding, but rather memorization and understanding of very specific algorithms. The focus is completely off. It's the "meta-algorithmic structure" that matters for the future of students (when they themselves have to go learn/create algorithms to solve future problems), not just recall the small subset of algorithms they learned in school.
The problem is, universities are verticals, selling both knowledge and assessment. Like IBM, Sun, and HP selling the OS and hardware. The knowledge is already a commodity, but the unis still want to bundle it with their exams.
