I beg to differ about Python. My understand is that Python, along with Matlab, are the most common languages used to implement the "deep learning" algorithms (really nothing more than multi-layered neuro-nets) that are purported to be the next advance in robotics and machine vision.

It depends on the industry you work in, to be honest. I work in banking, so my experiences colour my judgement on which languages are most important. In order to become a quant or developer at most banks, you need to be proficient in AT LEAST one of: C/C++, C#, JAVA, and possibly Ocaml. (If you know C/C++, the next two are pretty easy to pick up.) If however you're working in biomedical/pharmaceutical research, or robotics, or machine-learning, then the unforgiving nature of C++ combined with its need to compile first means that you'd be better off with something like MATLAB, R, Python, or SAS (for really big datasets).

As you say, the greatest benefit of learning a really hard language like C++ (or APL) is that, once you know how to program using those beasts, most other programming languages become a doddle to use. My experience has been primarily in banking and consulting, and that's where skill with C++ is a huge benefit. Female programmers who don't know C++, or who study a dumbed-down version of a CS major, simply won't hack it in the industry.

Matlab is a good language to use as it is the best for scientific programming.

I like it a lot, especially the later versions which are fully object-oriented. Personally I prefer using R for most applications, even financial ones, largely because I really like its syntax and its flexibility. But then, I'm a FLOSS nerd; MATLAB has better help files and a much slicker interface. If it were up to me I'd use R for everything- some commodity hedge funds actually do.

I beg to differ about Python. My understand is that Python, along with Matlab, are the most common languages used to implement the "deep learning" algorithms (really nothing more than multi-layered neuro-nets) that are purported to be the next advance in robotics and machine vision. Nevertheless, you are correct that one should learn structured programming to begin with (C/C++) in order to get the basic concepts down. Matlab is a good language to use as it is the best for scientific programming. The google search engine is implemented in Python.

I agree that replacing "introduction to programming in" with "creating approaches to problem-solving" is just fluff. We have far too much fluff in this country.