One tiny piece of criticism though: when calculating the percentage of unproductive keystrokes, divide by the number of ALL keystrokes, not just the correct ones.