Programming by poking: why MIT stopped teaching SICP | Hacker News Reader

Programming by poking: why MIT stopped teaching SICP

lambda-the-ultimate.org

33 pointstosh9 comments

Previous discussion: https://news.ycombinator.com/item?id=11628080. Unfortunately we have to have a base case for this recursion, which is to treat this as a duplicate.

> The same is true at the software level, since programming environments consist of gigantic libraries with enormous functionality. According to Sussman, his students spend most of their time reading manuals for these libraries to figure out how to stitch them together to get a job done.

You don't need an MIT education to do that kind of programming, and do it well.

There was an interesting lecture Theodore Roosevelt gave in 1911 at a small vocational school in Pasadena, California called Throop Polytechnic Institute. He said:

ROOSEVELT> I believe that the most important work for us in America today, in the way of education, is to train men so that they can do the most efficient possible work in the vocations they undertake; and I also regard it as of the utmost importance to train the exceptional man on a purely cultural basis. In other words, I want to see institutions like Throop turn out perhaps ninety-nine of every hundred students as men who are to do any given pieces of industrial work better than any one else can do them; I want to see those men do the kind of work that is now being done on the Panama Canal and on the great irrigation projects in the interior of this country---and the one hundredth man I want to see with the kind of cultural scientific training that will make him and his fellows the matrix out of which you can occasionally develop a man like your great astronomer, George Ellery Hale."

A few years later, that school changed its name to California Institute of Technology, and made it its mission to concentrate on Roosevelt's "hundredth man".

I think SICP is the kind of preparation that a hundredth man needs for programming.

SICP changed the way I think about computing. Despite being a book intended for freshmen, there's lots of interesting things to learn about even for experienced programmers, and the footnotes are terrific.

It may have lost relevance at MIT but all around London, companies are still using examples from SICP in their programming interviews and tests. Does this mean these selection schemes are stupid and irrelevant or does SICP still possess the power to sway and influence? Just not in the way the authors expected?

What are the examples being used in interviews/tests, and for what jobs? (I took one of the last instances of 6.001 / SICP right as the curriculum was changing; I don't regret it, but I already knew programming pretty well coming in.)

If you're testing someone's ability to write some code from scratch, as opposed to how to modify a large system that they can't fully understand by the time the interview is over, you are either conducting a good interview for a very small number of jobs, or a bad interview for a much larger number of jobs. A lot of MIT classes (e.g., the OS class, 6.828, where you're given most of an OS but with the important pieces left out for you to write) are much better at teaching the latter skill.

Also, to be fair, most programming interviews and tests are stupid and irrelevant, regardless of inspiration. :)

The ones I remember doing myself were the Tree recursion count-change problem, Simpson's Rule, and counting pairs in Lists. These were for general programming jobs as I recall on Javascript (not my speciality) or Ruby systems.

I went to university 8 years ago and learned from SICP. It has decades worth of influence left even if everyone in the world stopped teaching it today. Not to mention the second order effects.

In my experience, SICP's main effect was developing "analysis" muscle. I remember reading first 3 chapters and realizing that we are starting to write fairly complex stuff but still our language doesn't even have assignments let alone for loops or other bread and butter of programming. It's that moment, it dawns on you what computer program really means. It's not prescription of step-by-step instructions, but set of ideas that come together, compose with each other and make a whole that is far bigger than sum of small pieces. In SICP, you actually start at "atomic" level to build real world objects. I think this exercise is still relevant at least in that way and chances are that SICP readers are better at grasping and using abstractions.

Yes, I agree - I think this is why so much was also used in the Coursera Scala course to build up the student's understanding of the functional aspects in composing software. Must revise SICP sometime soon.

Reply on news.ycombinator.com