A New Kind of Science is on the iPad
blog.wolfram.com
blog.wolfram.com
I find that people's opinions of the book are quite polarized either they found it fascinating or were put-off by Wolfram's self-congratulating. So what does HN think?
Cosma Shalizi is a gem.
Aside from that, I think it's quite interesting and well worth a read, even if you're not into CA itself - that is, if you can navigate yourself around the bloat.
I've read the book from a data compression/information complexity interest standpoint. So, I can't comment on it's pretensions over it's usefulness in all area of sciences.
It has two parts, the main text and the supplement, which is actually longer. I've read most of the main text.
There's not much practically useful info in there that you couldn't pick up by reading good Wikipedia articles or other online articles, esp. now that the book has been our for years:
http://en.wikipedia.org/wiki/Cellular_automaton
I'd still consider buying it for the iPad if it's cheap (<$10), just so I can browse and read it when I'm bored on the subway..
Weinberg’s review in the New York Review was quite good: http://www.nybooks.com/articles/archives/2002/oct/24/is-the-...
The problem was not that he hadn't come up with something interesting. It was that what he had come up with was something everybody had come up with. He had just spent so much time holed up in his office writing and making diagrams that he hadn't poked his head up into the world enough to realize that he hadn't come up with anything unique or interesting.
This ego feeding manifested itself in all kinds of interesting ways. For example, his ego required that the things we were producing were unique based on what he knew. Even if we were building something that was essentially the same thing that everybody else in the world had built, he'd direct us to build it in some fashion that often ignored the practical results of good engineering practice (learned through thousands of iterations of trial and error over decades) and make it function in a way that was unique to our products, but not necessarily better, and because we were ignored so many hard lessons others had learned for us, often with significant downsides. Long after these things had been implemented, and the downsides of this approach were now readily apparent and in-your-face, we'd go back to him -- to which his reply would be, "I'll have a solution in the next version!"
Months later he'd roll out his new development roadmap that failed entirely to address those issues, pretending like they had never been brought up (no Adoubt since they bruised his ego) and sent us off in another random, but already well understood direction, but cock angled so we'd end up making another round of well documented mistakes.
A real simple example, instead of showing aggregate sums for discrete values in a data field as a simple bar chart, he insisted we use a different metaphor that had long since been abandoned by the field -- only he refused to acknowledge it. This fit the "novelty" requirement for his ego, but failed the "useful" part of building software.
What finally ended it all was a perfect storm of past customers all coming and complaining that they weren't getting the product they were paying for. In some cases we'd even been paid to fix long outstanding problems created by our failure to design well understood parts of the software around well understood metaphors. The result is that he very quickly went to being "no longer with the company".
My lessons from that experience is that innovation is important, but it has to be built on the solid foundation of past experience. Even Newton understood this, "If I have seen a little further it is by standing on the shoulders of Giants."
The problem with something like "A New Kind of Science" is that it's not new. It's a study of the applications of a well understood field onto other fields. If it was written without ego the title would have been "A Study of the Applications of Cellular Automata and State Machines" or "Recognizing Computation in Cross-Domain Studies Using Cellular Automata", or the lofty "An Exhautive Review of Cellular Automata with Applications as Turing Machines." In other words, NKS brings us back to square one with Turing's "What is computable?" Only looked at from another angle.
It's not that what's in the book isn't interesting, it is. It's that it's not revolutionary in the way Wolfram wishes it was. It's "merely" evolutionary and Wolfram refuses to understand that. And just like my old Boss always chided us with a "wait and see" and a twinkle in his eye whenever we pointed this out, Wolfram's response is "wait and see".
If you don't take the "this will change the world and is the most important thing since math was invented" tone too seriously, it's pretty entertaining. And who's to say, it might turn into a revolution. Certainly the ideas about complexity theory he puts forth are interesting and worth thinking about.
If I had to boil it down, it would be "Any system is either trivially simplistic, or complex enough to do anything." Think logic gates built with conway's game of life, or adders built with dwarf fortress pumps, that kind of thing. He's trying to unify all complex systems with cellular automata.
That's my take on it, at least. Read it a while ago. If you're interested in cellular automata or complexity, it's worth reading.
This would be really cool on the ipad if the illustrations were interactive, kind of like the NKS Explorer PC app he released a few years ago. That was a fun tool to play with this stuff.
It never did.
The book is filled with confused science, pretty graphics and massive ego stroking. Huge disappointment.
Setting aside Wolfram's book, I also wonder if Knuth's Art of Computer Programming, Abelson and Sussman's Structure and Interpretation of Computer Programs, or the Introduction to Algorithms books would benefit. We'll see soon I'm sure.
But getting back to the static version, certainly one benefit is to avoid carrying the dead tree versions, as Wolfram pointed out.
Wolfram mentions getting excited about zooming in on the diagrams, and about using the app to refer to parts of the book for himself or to show others. In fact, it seems like the tablet would be great for two people exploring a book or data together.
How many times do you really do the exercises in books, no matter how much the authors stress that it is important to do the exercises? I tend to skip and save them "for later".
Don't forget that whatever glorious visual interactive treats you envision will have to be, you know, implemented, and decent odds it's already either implemented or not really worth implementing for programming. Sure, it's nice to see a tree balancing or a linked list getting traversed but the virtues of pretty animations start fading as you get a hang of the material.
The value of digital education won't and doesn't come from putting a thin, shiny veneer of digital on top of analog textbooks, it will come from enabling entirely new learning modes based around instantaneous and copious feedback that humans can't give, because there aren't enough of them or because they simply can't feed back fast enough. Try to model more disciplines to look like how a modern self-taught hacker learns programming, don't pine as a programmer for flashy textbook-like "digital textbooks" for programming.
I liked the book a lot, but have a hard time not seeing this as hyperbole. Yes, it's the printing industry..but it's just just diagrams and text; I don't remember anything that a few weeks of learning LaTeX couldn't accomplish?
edit: typo.