Why our current obsession with high-stakes testing is wrong