Easy: Lecture contains an algorithm for diagonalising 3×3 matrices and an example. Tutorial/exam has students do a very similar example (e.g. both matrices have lots of zeroes in the same cells).

Harder: Lecture contains an algorithm for diagonalising 3×3 matrices and an example. Tutorial/exam has students do a matrix that is structurally very different.

Hard: Lecture contains an algorithm for diagonalising 3×3 matrices and an example. Tutorial/exam has students extend the ideas of the algorithm to a specific 4x4 matrix.

My own undergraduate experience was largely in the "hard" category, whilst as a lecturer at a different university there was pressure to move increasingly towards the "easy". It was even fairly explicit: student feedback surveys would be aggregated and you'd be informed that the students are unhappy because the tutorial exercises cover things they "haven't seen" in lectures.

The reweighting is not a good solution, but I would ask questions to the lecturer why a good chunk of her students are failing. It's her responsibility to teach them after all, and not to filter out the elite.