Maybe tangential to the point of the article, but I find that one of the most frustrating things about working with recent college grads is that they don't understand the ambiguity. It's not thier fault - they have spent the last 16+ years being told[1] there is a right answer. It is ingrained in them, and a lot of effort must be put forth to teach them that most of the work worth doing in our field is by definition not "solved". If there was a right answer or known best solution, we'd just be using that library and focusing on the ambiguous bits instead.

Yes there is a lot of boilerplate and plugging existing components together, but that stuff isn't the work that really brings in the bucks - it's the part where you are solving something previously unsolved, and making headway against ambiguous problem spaces that really differentiates companies/products/teams etc.

[1] It's more than being taught, it is being rewarded and punished on the notion of "one right answer" that happens in huge amounts of education. It gets mixed up with the limbic system and becomes a default assumption for everything. It has to be un-taught in a lot of contexts. Sure, there is a best or right answer for many things, and we need to show that too, but more education on how to deal with the ambiguous stuff is much needed.