Don't forget randomness is still just a hypothesis

Anton Zeilinger's bold essay "The message of the quantum" (Nature 438, 743; 2005 ( claims "the discovery that individual events are irreducibly random is probably one of the most significant findings of the twentieth century." But we should not forget that the claim of true randomness has not yet been backed by evidence. Neither Heisenberg's uncertainty principle nor Bell's inequality exclude the possibility, however small, that the Universe, including all observers inhabiting it, is in principle computable by a completely deterministic computer program, as first suggested by computer pioneer Konrad Zuse in 1967 (Elektron. Datenverarb. 8, 336344; 1967).

The principle of Occam's razor, which is fundamental to theory-building, favours simple explanations (describable by few bits of information) over complex ones. But if the Universe's history really included many truly random events, an enormous amount of information would be necessary to describe all the random observations inexplicable by the known, simple, elegant, compactly describable laws of physics.

A few previous attempts at discovering a pseudo-random generator behind seemingly random physical events have failed (see T. Erber and S. Putterman Nature 318, 4143; 1985 ( But as long as the randomness hypothesis has not been verified, physicists should keep trying to falsify it and search not only for statistical laws but also for deterministic rules explaining any type of hitherto unexplained apparent randomness.

The correspondence to the left appeared in Nature 439, 392 (26 January 2006); doi:10.1038/439392d; copyright © Macmillan Publishers Ltd.

Jürgen Schmidhuber
IDSIA, Galleria 2, 6928 Manno-Lugano, Switzerland & Robotics and Embedded Systems, Tech. Univ. München, Computer Science, Boltzmannstr. 3, 85748 Garching, Germany

Related links:

1. Algorithmic theory of everything

2. Zuse's hypothesis and Zuse himself

3. Occam's razor and generalized algorithmic information

4. Speed Prior. On the fastest way of computing any computable universe, plus optimal predictions of the future

5. Universal learning algorithms plus consequences for optimal inference of laws of computable universes