3 ms·
"at the earliest it'll be the next day." One of the interesting hard mathematical results to come out of generalized learning theory, as used by machine learni
by jerf 11y ago
"at the earliest it'll be the next day."
One of the interesting hard mathematical results to come out of generalized learning theory, as used by machine learning, is that there is a fundamental relationship between the speed of feedback and the ability to learn. It is not merely a matter of "willpower" to learn from temporally distant feedback, it is fundamentally, irreducibly harder, and there's no way to "correct" it because it's not even wrong.
Homework grading (note the emphasis) as a learning mechanism has such a delay that its learning rate is effectively zero. And what feedback you get is often not even very much beyond WRONG!, which is not exactly rich with information itself.
Homework is only useful for the feedback and learning you get while doing it, and for the ability to assess the student because we live & die on assessments, give us a direct choice between assessment and learning and the system will generally prefer the former. So, I won't say "homework is useless" but we could save everybody a lot of time if we only checked that it was done and spot-graded it for assessment, rather than making the teacher laboriously, yet nearly entirely uselessly, grade stacks of homework.