I don't think it's a question of design. Some ideas in stats are just fundamentally unintuitive. Pedagogy can help mitigate this, but it would take a revolution to fit it into a neat and intuitive framework, like calculus.