Corrected ANKOS book cover: A 35-Year-Old Kind of Science

Origin of main ideas
in Wolfram's book

"A New Kind of Science"

By J. Schmidhuber,
based on a letter in CERN Courier
(Intl. Journal of High-Energy Physics, vol 43:5, June 2003)

Wolfram has repeatedly managed to convince unsuspecting journalists that his book "A new kind of science" (2002) caused a dramatic paradigm shift by claiming that the universe and everything is being computed by a simple program.

Those familiar with the subject, however, know that the paradigm shift (if any) dates back at least 35 years.

Konrad Zuse It was not Wolfram but Konrad Zuse who was the first to suggest that the physical universe is being computed on a discrete computer, such as a deterministic cellular automaton (CA). His first paper on this topic dates back to 1967: Konrad Zuse, Rechnender Raum, Elektronische Datenverarbeitung, vol. 8, pages 336-344, 1967. PDF.

Note that this is the same Konrad Zuse who built world's first working general purpose computers 1935-1941.

Zuse's book on CA-based universes came out 2 years later: Rechnender Raum, Schriften zur Datenverarbeitung, Band 1, Friedrich Vieweg & Sohn, Braunschweig 1969.

English translation: Calculating Space, MIT Technical Translation AZT-70-164-GEMIT, MIT (Proj. MAC), Cambridge, Mass. 02139, Feb. 1970. PDF.

More on Zuse's thesis.

Edward Fredkin (who initiated the MIT translation) later also published related ideas.

Wolfram grossly misrepresents Zuse's work, mentioning him in one single sentence (on page 1026), claiming that Zuse said that the universe "could be a continuous CA." He fails to state that Zuse suggested that reality is in fact computed by a discrete computer or CA, like those used by Wolfram. In doing so, he implies that something essential was missing in Zuse's work. But of course all the patterns in "A New Kind of Science" could be reproduced on a Zuse CA. In fact Wolfram's book adds nothing substantial to Zuse's ideas, besides the pretty illustrations. Nor does it offer any predictions that go beyond Zuse's. Compare both books and judge by yourself.

On page 486 Wolfram apparently tries to distance himself a bit from Zuse by writing "that the universe might not work like a CA ... but instead like a mobile automaton or Turing machine."

The first paper to suggest this was published five years earlier though: J. Schmidhuber. A Computer Scientist's View of Life, the Universe, and Everything (in C. Freksa, ed., Lecture Notes in Computer Science, pp. 201-208, Springer, 1997).

This essay focused on the simplest explanation of our universe, namely, the very short program that computes all possible universe histories with all possible computable laws. The approach is echoed in Wolfram's chapter 9, section on multiway systems.

The 1997 paper also talks about universes simulated within parent universes in nested fashion, and about universal complexity-based measures on possible universes. Pointers to follow-up publications (2000-2002, on universes sampled from computable probability distributions, and on the fastest way of computing all universes) can be found here.

Discussions of this and related work on the everything mailing list (since 1998) are archived here.

You will also find many critical and occasionally quite competent reviews of Wolfram's book in Edwin Clark's review page and among's customer reviews.

bitstring universesBack to intro to computable universes

Back to