Lambda the Ultimate

The more I read on his site the more I think it might all be a joke, because he sure makes a lot poorly informed statements. You might want to read his filesharing *essay.*

http://www.andkon.com/stuf/fileshare/index.html

And I fear I've spent far too much time on him.

allow beginners to do useful stuff
Doesn't that mean playing with bigger building blocks? Adding dll functionality to Excel or writing a spam filter addin for Outlook or...

Spoken language teaching very quickly places the learning experience within the context of normal conversational situations. Could cs teaching place learning within an industrial context as modification to an existing large code base? Teaching programming-in-the-small within the context of programming-in-the-large.

stop teaching languages based on market share
For some institutions, that matches the immediate goals of both the institution and the student!
Others have different ambitions: M206 Computing: An Object-Oriented Approach (it teaches programming skills using the object-oriented language Smalltalk).

But I decided that even if it is a hoax, it is still worth discussing.

Still Scheme gave me a hard time. Even I admit its elegance, but I don't believe that Scheme or Haskel is a good introduction lang. Think about how to write Scheme or Haskel code to get the input from keyboard :D I don't think the code is intuitive to beginners :) Those funtional langs have a very nice maths but don't give an intuitive way on how computers run the code. Keep in mind that the way of thinking about computers run step by step in code is the foundation for students to learn algorithms. ok funtional langs might be more declarative but it is not good for beginners.

Pascal is good for beginners but it is too old. recently I've tried out Python, pretty simple and easy to explain to beginners. However, it is script which does not have type.

Java which is a popular choice now is bad too :(

We don't need his input about C++, I guess we all know the language.

Personally, I prefer statically typed languages, but that's just me.

Anyway, the Haskell and Scheme thing was a bit of a troll. I am not sure I would really use either of them for CS1 these days.

I myself wouldn't use Haskell for CS 1 because it's too far from the mainstream. It's one thing not to encourage mutation, quite another to forbid it entirely. Disclaimer: I *like* Haskell a lot. It's good for more advanced courses.

Two reasons why this may not matter: You may have more than one intro course (two semester, perhaps?), and/or better students.

And laziness is what kills Haskell for CS1.

Yeah, CS1 students are often quite lazy ... :-)

I guess what you see as a negative I see as a positive. I like the fact that students can see a language with a deliberately unusual syntax, so that they can divorce themselves from the misconception that syntax is one of the more important features of programming languages (and that "all syntaxes are basically the same", which they clearly aren't). Also, the scheme syntax is so simple that you spend much more time dealing with concepts. That said, it does take most of our students 2-3 weeks before they get comfortable with s-expression syntax.

And laziness is what kills Haskell for CS1.

To me it's the absence of mutation and hence the need for advanced concepts like monads. To split hairs, laziness leads to both of these ;-)

Now, of course, since I was at Indiana University, I had access to amazing resources. Dan Friedman did more to instill the sheer joy of programming in me than anyone else in my life, period. And Douglas Hofstadter did more to stretch my mind than anyone else in my life, period. Not everyone who's exposed to Scheme as an undergrad has these same benefits, to put it mildly.

I wonder to what extent using Scheme is only a good idea if you're at MIT, Indiana, or Rice? That is, in environments with extremely strong Scheme cultures and professors?

BTW Dan Friedman is a heck of a nice guy. I emailed him once with a question about one of his books and his reply was incredibly friendly and helpful.

To say the least. Several years ago I posted something complimentary about "The Little Schemer" on comp.lang.scheme, and the next thing I knew, a copy of Scheme and the Art of Programming was on my doorstep, with a card: "Compliments of Dan Friedman." But that's Dr. Friedman: unusually generous. And his whole style is very much represented in "The Little Schemer:" the playfulness; the use of real-world examples that we can all relate to, e.g. FOOD!; the periodic reminders to take a break and try to digest the material, but not all in one gulp.

I think the thing I appreciate so much about Dr. Friedman is that he treats his students, whether face-to-face or not, like human beings. I can't recall ever having another professor that said anything remotely like "Hey, this is a lot of new stuff. Let's stop here. Maybe it'd be a good time to make a sandwich." That and his obvious sense of actual joy over CS. I mean real, in-your-face, jumping-up-and-down excitement. He really, really loves CS. And he really, really loves revealing it to 17-year-old minds. And his love affair with Scheme is, of course, legendary. As it should be.

Some people take this "nice guy" thing entirely too far ;-) Hey, *I* posted a glowing review of TLS on amazon.com, and where the hell is *my* free copy of SatAoP? Dan, are you listening? ;-)

Seriously, TLS is an insanely great book. Any book that can go from counting to the Y combinator in a few chapters is awesome.

Your stories match my experience. When I started working on EOPL, I made some comments about missing topics, I got an email telling me about the second edition, and yes the book also arrived on my door step well before our library managed to get its hands on a copy.

Hey! I had to buy BOTH editions of EOPL out of my own pocket! ;-)

Naturally, I was happy to: as I think you pointed out before, Ehud, the first edition is probably better-suited to self-study, but the revised material on CPS in the second edition is must-have.

As for the "student", (is this accurate? - He obviously didn't "study"), well I think this is evidence of a wider problem. I visited my old University tutor a little over a year ago because I heard he was retiring. He told me that modern students were nothing like those in my day. (Cue Monty Python sketch).

To begin with, the relentless pressure caused by the massive expansion of UK higher education has meant that the entry criteria had been relaxed from "Advanced" Level mathematics to just the basic GCSE qualification usually gained at 16. They were getting students who wanted to do computing because it was a well paid career, but who simply didn't have the basic ability to understand what was required.

He also complained that they now had to use Java, which he thought to be completely unsuitable as a teaching language.

The problems he was referring to are obviously pretty widespread.

I spent my first year of high school at Stuy, where the introductory CS elective was taught by Mike Zamansky in Logo and Scheme. It was a really neat introduction to CS that I think discouraged a lot less people from pursuing CS further than some syntax-thumping printf lesson. I really must give 'Z' credit for the excellent introduction to recursion (after that it was like riding a bicycle - I never forgot how). When I first read it, I was surprised to find that a lot of the examples and assignments used in the course are remarkably similar to parts of Flake's "Fish and Chips Book". That's probably the best way to describe the character of the class, the quality of instruction and the enjoyment I derived from it. Interestingly enough, he didn't stress Scheme's 'lispiness' at all, and it wasn't until years later (by which time I had pretty much given up on CS) when I found Structure and Interpretation that I realized how powerful the language really is. I think this is the key to teaching Lisp (or whatever language you're a rabid follower of) correctly - let the student have the pleasure of finding the pearl her/himself, and for this I am indebted to Z for his wonderful teaching style.