I agree that visual programming can be great - my argument would be that starting from a model that we know works and making incremental steps towards something more visual is more likely to succeed, than attempting to jump all the way to something fully visual based