Brauer characters are defined only on p-regular elements of G. Why can't ordinary trace be used as a character in characteristic p?
Think about your answer, then reveal below.
Model answer: The trace of a matrix over a field of characteristic p takes values in that field, where distinct eigenvalue configurations can give the same trace (e.g., a matrix with p identical eigenvalues λ has trace pλ = 0). Brauer's solution is to lift eigenvalues to characteristic 0 via a ring of p-adic integers and take the trace there, but this only works for elements whose eigenvalues are roots of unity of order coprime to p — the p-regular elements.
In characteristic 0, the trace determines the multiset of eigenvalues (via Newton's identities). In characteristic p, information is lost: tr([[1,1],[0,1]]) = 2 = tr(I₂) over 𝔽₂, even though these matrices are not similar. Brauer characters bypass this by lifting to characteristic 0, recovering enough information to classify irreducible modules. The resulting theory is powerful but more intricate than ordinary character theory.