School improvements like smaller classes, vouchers, new standards or more spending disappoint when they leave untouched what actually happens between teacher and students. Cutting a class from thirty to twenty makes teachers happier, but if they keep the same methods, students' chances to learn barely change.
Achievement tests show how well students are learning but can never reveal what to change to raise the results. For that you need direct evidence of classroom processes, yet decisions are often made without even basic information about what goes on in classrooms.
Teaching is the one part of schooling designed specifically to produce learning, and unlike students' backgrounds it lies within teachers' control. That combination of being central and changeable is what makes it the most promising place to direct improvement efforts.
Poor results come less from incompetent teachers than from the limited methods a whole culture passes down as 'how teaching is done.' Because teaching is a cultural routine, swapping or blaming individuals changes little; it is the shared practice itself that has to be improved.
A profession gets better over time only if it has a routine for getting better; without one, even capable practitioners stay stuck. Giving teachers time within the working day to jointly plan, study and refine lessons turns classrooms into laboratories and makes learning to teach a career-long process.