Home » forums » LtU Forum

Stephen Wolfram forecasts the future

From this article:

A dominant theme of the past 50 years has been the growing understanding of how to create computer programs for specific tasks. A central theme of the next 50 will be the systematic exploration of the "computational universe" of all possible programs.

In recent years we have discovered that very small programs can produce immensely rich and complex behaviour. As we explore the computational universe this fact will become crucial to science, art, technology and our whole way of thinking. Today when we create things, we expect to do it by explicit human effort. Fifty years from now, it will more often be done by automatically mining the computational universe. Perhaps we will even find the ultimate model of our universe that way. No doubt many industries will be created from technology discovered in the computational universe - whether for programmable nanostructures, algorithmic drugs, mass-customised art or generative microeconomics.

Just as scientific concepts such as momentum and natural selection are commonplace in our thinking today, so in 50 years will be concepts like computational irreducibility and the principle of computational equivalence. I expect the children of 50 years from now will learn cellular automata before they learn algebra.

By Curt Sampson at 2006-12-01 06:21 | LtU Forum | previous forum topic | next forum topic | other blogs | 6785 reads

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

urg

Ya know, I read his book (all of it). When you tout a computational method for the fact that it doesn't handle constraints, doesn't that invalidate the usefulness of it? Sure, there're examples of directly modelling fluid-flow and the like with CA, but in each case, the CA model was no different than a naive matrix-based approach would've been. Maybe what he says is true, but I'm very skeptical.

By jimdesu at Fri, 2006-12-01 07:13 | login or register to post comments

I read all of it, too. His

I read all of it, too. His approach is not new, infact there is a question about what in it is *new*. It is, however, an excellent synthesis of information on CAs, and it provides the insight into someone who is regarded by many, a genius.

However, his fluid flow was a basic example of how to apply a rules based approached to natural phenomena. He reasoned that if you can model each particle bouncing one another using what is understood to be happening then you should get something that resembled fluid flow.

The microscopic view showed the chaos one would expect; the macroscopic view showed eddies. Wow, he was right. My point is that this approach is very useful for modeling natural phenomena where we know the rules.

What is even more critical about what he shows is that simple rules can produce four different classes of results. They range from mundane to downright random and chaotic. The big result was that there was a transition that exhibited characteristics of both ends. He then searched all rules (encoded in binary) up to some number - I forget. He gave some reason for not going past this number of bits, but again I forget.

One of the things that he did, which I thought was really cool, is that he showed that one of his rules (like GoL) was universal. From this he created a turing machine using his rule "110" - http://www.wolframscience.com/nksonline/page-707 using 2 states and 5 possible colors.

In fact, the most significant chapter for me was chapters 11 and 12. I regarded the previous chapters as background information, but it is in these 2 chapters that I believe he presents the best stuff.

It has been my experience that people who criticize the book have either never ready any significant portion of the book, or they have misunderstood several key points of the book. If anything, the book is an excellent book on CA in general and its implications in the computer science realm.

Now, back the the main point. Searching for rules or sets of rules may yield a significant set of applicable "rules" (not formulas) that describe some phenomena. With the power of computers today, this is a worthwhile effort. I also believe that CAs offer a way describe systems that lack traditional mathematical formulations.

If you have the patience/interest make your way through the whole book, I highly encourage it. The material is not so high level that it would stump anyone who felt the urge to even comment on the subject matter - this includes high school students.

By estrabd at Sat, 2006-12-02 21:15 | login or register to post comments

Here are some little less

Here are some little less friendly remarks about Wolframs impostures.

By Kay Schluehr at Fri, 2006-12-01 07:47 | login or register to post comments

explanation?

Perhaps someone can offer an explanation, for those of us who don't understand what he means by "automatically mining the computational universe." Is it anything like using genetic programming to generate a bunch of random (but directed) programs and picking out the best one?

By shahbaz at Fri, 2006-12-01 15:53 | login or register to post comments

Maybe it means that people

Maybe it means that people will just let the google codesearch bot automatically assemble code snippets into working artifacts. It seems to be a common claim that programmers will eventually be replaced by computers once we get a sufficient amount of code and constraints together.

By Fred LeMaster at Sat, 2006-12-02 05:59 | login or register to post comments

Re: explanation?

I recommend reading the link provided by Kay. It explains pretty well where Wolfram is coming from. But the short answer to your question is "yes, but there's nothing practical right now, and there may never be".

By Jeff Nowakowski at Sat, 2006-12-02 17:29 | login or register to post comments

I learned cellular automata

I learned cellular automata before I learned algebra. I learned CAs by 6th grade (that's the first time I remember writing one) and algebra not until 7th grade. Is that unusual?

I haven't finished reading the book myself, but beware of giving Wolfram credit for any of it. The Shalizi Review clearly describes some of the strong-arm intimidation tactics Wolfram has used to capture credit for various inventions. As demonstrated here in the comments, estrabd got hoodwinked into thinking that Wolfram invented lattice-gas automata and proved that rule 110 was universal, but neither of those is his work.

The International Journal of Unconventional Computing seems to be where the honest researchers in the field are publishing.

By Kragen Sitaker at Sat, 2006-12-02 22:49 | login or register to post comments

Thanks for that.......

thanks for the information...I guess I've been duped as well :)

By estrabd at Sun, 2006-12-03 00:18 | login or register to post comments

For the paranoid, the

For the paranoid, the copyright page of the book (and, in particular, the considerable quantity of discussion about how concepts presented in the book are property of Stephen Wolfram, LLC) makes good reading, too. :)

By Brooks Moses at Sun, 2006-12-03 04:23 | login or register to post comments