I think the reason visual programming hasn't taken off is there is a very real difference between what you can express with language and what you can express visually. Images don't really contain determinate, precise statements or propositions the same way that written or spoken word does. In contrast language can express any thought, no matter how recursive or abstract, effortlessly. For the narrow domain of programming - not typesetting, not rendering and manipulating images - the text editor happens to be an extremely very powerful tool that allows us to tell the computer exactly what we want it to do. It hasn't been replaced because it hasn't needed to be - it is simply the most powerful tool in which to write language.
The other reason visual programming hasn't taken off is because programmers aren't visually oriented, right-brained, intuitive people. Exceptions to this rule abound, but I am certainly not one of them, and neither are many programmers I've known. They've tended to be left brained, linear, linguistic, rational-over-intuitive sort of people - which is why we pick up computer languages so quickly, and it's also why we like typing code into a text editor instead of drawing our program on screen. I know that functionality isn't lateralized to easily and this is an oversimplification - but there does seem to be a dispositional difference between those who are attracted to programming and those who aren't.
But this'll probably change, as everything does. As the author notes FP and lambda calculus have been around since the beginning of computer science as a field and they are just now starting to get mainstream traction. So, maybe, in about 20 years visual programming tools will start to be mature enough to applied to any arbitrary problem., and a powerful visual language will develop that attracts enough visually-oriented people to change or revolutionize the field. But we can't put the chicken before the egg - the tooling needs to mature enough to cause this network effect to occur.