On the success of ‘natural language programming’