You learn to solve hard problems just like you learn to do anything else: by working on them. The problem is that there isn't a "List of Hard Problems" posted anywhere. Aside from a few well-publicized ones (e.g. cure for cancer), being able to recognize a Hard Problem requires years if not decades of domain expertise, which most programmers don't have (since software engineering is a relatively young profession).

It's harder than it looks. To make an average CRUD app the current state of the art, you need to know a very large number of tools, each of which has its own ecosystem of alternatives, each of which has a different philosophy. This makes it all but impossible to offer a course on the subject that meets any standard of exhaustiveness. So the would-be web developer has to wade through a morass of conflicting ideas on best-practices and will inevitably screw something important up, like authorization.

Then you have the constantly-shifting nature of the field, forcing you to continuously study to stay on top. It's much more difficult than it looks to an insider who already put in the hours learning.

You're right, but making CRUD apps pays a lot better (and more quickly) than complex math/physics.

I don't want to downplay learning complex math and physics, but there are other "hard problems" that programmers (and hence products) aren't solving that don't invoke math or physics. One of the things I harp on in product development is "if we only solve the easy problems, what's to keep someone else from coming along and solving the easy problems better than we do?"

Just a different perspective on the same topic ;)

so very, very little of day-to-day coding is what I bucket as CS problems. It's mostly turn the crank part assembly that's needed.

Not that CS problems aren't out there, but your architects should be decomposing those for more junior people.

Why junior people?

I guess we are still at a stage where there are many problems that can be solved by building 'simple' CRUD apps.

Once these problems are considered as solved and as frameworks/tools evolve further to get the CRUD out of the way really quickly, we'll see more people learning to solve complex problems.

Sure, it's not hard, but there are some things that are challenging, like controlling customer expectations without upsetting them, making deadlines, or moving deadlines.