> If an instructor deliberately gives a student a problem that they know the student _cannot_ solve, then it rightfully destroys trust.
This does not destroy trust, but gives the student an important lesson: we only have the techniques to solve, say, 0.0000001% of the problems. So you have to learn brutally hard for the next many years (or rather decades) to have the minimal qualifications to be able to invent whole new techniques that no person has ever come up in history before to increase this ratio from, say,to 0.000000100000000001% (even this would trigger a whole new aera in the history of science).
This does not destroy trust, but gives the student an important lesson: we only have the techniques to solve, say, 0.0000001% of the problems. So you have to learn brutally hard for the next many years (or rather decades) to have the minimal qualifications to be able to invent whole new techniques that no person has ever come up in history before to increase this ratio from, say,to 0.000000100000000001% (even this would trigger a whole new aera in the history of science).