I don't think the recurring failure is that we haven't found the right abstraction yet. It's that abstraction is often mistaken for understanding