To those of you who hate R, consider investing 1/2 hour to learn the dplyr package. Dplyr is, in my view, Hadley Wickham's real masterpiece and is why I use R for most data analysis nowadays.
As for ggplot, the 'grammar of graphics' approach makes it intuitive to get started with but I often run into trouble with both the inheritance hierarchy and with getting graphics 'the last mile' to presentation-quality.
My favorite ggplot2 graphic? The London Cycle Hires Map:
These calculations seem fishy, for reasons stated above and others (e.g. why 20-year ROI? why exclude graduate degree holders?).
However, it is true that rising education costs are eating much of the ROI of attendance and lack of transparency has made it harder to see where that crossover point is.
But in a world in which the equity and college wage premiums are headed in the directions that they're headed, I would not give this advice to an intelligent 18-year-old.
Writing a full reply since I don't agree with much of the advice given.
I've worked around/in data science teams at a large BigCo and I think that you're far overestimating the bar here. There aren't enough people to who can write data pipeline code (SQL/Shell/etc.), much less implement and intelligently explain statistical/ML models. Also, the average decision maker here does not understand the difference between 'created model in Pandas' and 'created model with Amazon's ML API'.
The modal background of data scientists in industry is closer to 'Econ BA + knows Python' than 'Artificial Intelligence PhD'. Moreover, the former will still enjoy a remunerative career if (s)he's sufficiently savvy about identifying problems and showing off how they can be solved with technology.
There may be a point in time when companies can't get a return by throwing math-savvy programmers at a problem, but that will be long after you and I have passed from the scene.
People who see something wrong with your question/project/link tend to be more motivated to comment, which I think explains some of what you're talking about.
The other way may be the way you use HN. There's a connection between distraction and anxiety/depression; many of us, of course, use HN as a distraction from something else we ought to be doing.
I recommend John Guttag's Intro to Computation and Programming Using Python to people interested in starting to code, as the accompanying video lectures are available on MIT OCW and the book itself is enjoyable to read. That book (+ the Django tutorial) might be a good way to get started before doing a boot camp.
As others have pointed out, there are other ways to learn the material, but it may be that the 'career day' activities, etc. are worth the price of tuition.
My favorite Grand Central easter egg is the 'Whispering Arch' by the Oyster Bar. Standing in one corner, you can speak into the wall and be heard by someone standing in the opposite corner.
As for ggplot, the 'grammar of graphics' approach makes it intuitive to get started with but I often run into trouble with both the inheritance hierarchy and with getting graphics 'the last mile' to presentation-quality.
My favorite ggplot2 graphic? The London Cycle Hires Map:
http://spatial.ly/2012/02/great-maps-ggplot2/