it appears to me that in some parts of mathematics, errors in proofs often are fixable (number theory seems that way?), in others, a bug is often fatal (computational complexity theory, it seems)

why the difference? or am i just imagining things