Lambda the Ultimate

My opinion is that the reputation of Lisp has been tarnished by such a close association with neural nets and genetic algorithms. I don't know of any reasonable application of these algorithms to web technologies, contrary to the empty assertions of the article.

Well, I think the article was saying that neural nets and genetic algorithms were being applied to enterprise applications, not (merely) web applications. Which brings up another point: what the heck is an enterprise application exactly? (I really have no idea, other than it's an overused buzzword that makes me think of boring database applications written in Java). If that’s basically what they are (business applications) then it's possible that these approaches are used for data mining (finding patterns in data, etc).

Both neural networks and genetic algorithms (and the two together) have been used in real-world (finance, medical, chemical, and, yes, data mining) applications. I personally don't know of any web applications, but neural networks would be good for categorizing web pages (I started doing this myself once as a way to find pages on web stores which displayed products, but didn't have the time to finish.)

By the way, the preferred languages for neural network applications probably have been C and C++, not LISP.

There's a lot of truth in that. The LISP community has made the LISP language appear much more exotic and complex than it really is by surrounding it with such mystique. I used to use Tcl a lot, and do all sorts of down-and-dirty things to extend the language. When I decided to learn Common LISP and steeled myself for a learning experience akin to getting to grips with monads, I found that programming in Common LISP was little different to programming in Tcl or any high level, dynamic language.

One easy way to change the image of LISP would be to stop describing Common LISP as a big language. There is no clean dividing line between language and library in LISP, so it is more accurate to describe Common LISP as a small language with a large standard library. However, compared to Java or .NET, the standard library of Common LISP is very small.

I think they can stand to have a Lisp book or two out there

Well, they started publishing books on other historic languages recently. <gd&r>

at 618 Pages, that's some nutshell.

K&R (2nd edition) - 272 pages.

I know which one I'd rather have.

Since Lisp has been around since the beginning of time, the best cover would probably be an Ent ("the oldest living thing that still walks beneath the Sun upon this Middle-earth" - In the middle of reading LOTR to my daughter).

Current O'Reilly Animals.

Or a λamb

A crocodile.

It's been around practically forever, survived the extinction of the dinosaurs, and despite its long evolution, still pretty much looks like it always did. The design may look weird, but it works.

"The combination of AI capability with the ability to create entirely new programming languages make LISP uniquely situated for the near future of computation."

In an article about how Lisp deserves to be free of its rather dubious reputation, this seems like a strange quote to end with. I wonder if Lisp will ever be free of "AI"...

(By the way, you linked straight to the second page of the article. Here's the beginning.)

I agree. In fact, I think the claim that Lisp is "for" AI is one of the more absurd claims (or "programming language myths"), notwithstanding the history of the language.

A few years ago, I programmed an Eliza chatterbot for my students, in Prolog. Once I had finished doing it, I realized that roughly two thirds of the code was basically reimplementing a Lisp interpreter. Morality ? Whenever people ask me, I'll keep answering that Lisp is good for AI :)