Sign up ×
Physics Stack Exchange is a question and answer site for active researchers, academics and students of physics. It's 100% free.

Evolution is a principle in biology, whereby organisms evolve their ability to replicate and multiply in numbers over successive generations. From a computational point of view, the organisms employ a common programming (DNA, RNA, etc.) and runtime environment (cell biology). This evolution has eventually evolved a more powerful ((in a restricted sense) computational environment, the hierarchy of biological brains, with the currently most powerful class instance being the human brain. A collection of human brains is now working on evolving a quantum computer. Perhaps one day the the quantum computers will produce a....

From an anthropic point of view, we are here to think about these things, because our universe creates environments such as our planet earth, which support the evolution of life.

So my question is whether there is a general physical and computational principle at work here, demonstrated by the following incomplete, in parts almost surely incorrect, and highly speculative hierarchical chain of evolution:

  1. Within the Multiverse, universes are instantiated following particular physical laws originated and adapted from their embedding universe

  2. Each instantiated embedded universe follows computational rules (the particular physical laws of this universe) and create a finite number of embedded universes (the only candidate seems to be for singularities of black holes to correspond with a white-hole, inflationary, embedded universe), each of which evolve in the same general manner

  3. Each universe produces a finite number of computational schemes to build organisms that follow certain classes of program patters. These organism evolve according to the theory of biological evolution.

  4. Biological organisms evolve biological brains, a higher class computational scheme

  5. biological brains build computational devices, which at some point attain the property of replication

  6. The replicating computational devices build higher classes of computational devices, following the general principle of evolution...

share|cite|improve this question
You may enjoy this (purely fictional):… – Manishearth Mar 24 '12 at 16:20
Evolution works because less fit specimens are killed off without what do you suggest is "killing" universes? – dmckee Mar 24 '12 at 21:27
@dmckee, the universe can, depending on the particular physical laws it is instantiated with, for example (a) die fast due to implosion, radiating as a white-hole to the embedding universe (b) die slowly by hawking evaporation , (c) evolve so slowly that it is relatively or completely frozen in its infancy, or (d) just in general evolve so slowly that it is relatively ineffective at spawning embedded universes or generating higher level computational structures. – Halfdan Faber Mar 25 '12 at 5:39
The fittest universes are those that evolve relatively fast both horizontally (spawning universes) and vertically (expanding evolutionary/computational levels). Those universes will dominate the multiverse. Satisfying an anthropic principle, any biological or higher order life found in the multiverse will almost surely originate from a dominating region. If anyone thinks it is interesting, I will amend the original question with brief fitness examples for each level in the evolutionary hierarchy. – Halfdan Faber Mar 25 '12 at 5:40
@Manishearth, that's a hilarious and interesting read. A very talented writer... Seems to have been written back in 2003? I can only find a couple references online. – Halfdan Faber Mar 25 '12 at 5:56

1 Answer 1

This is the fecund universe idea, due to Smolin. The original form assumed that a new universe formed every time a black hole appeared (as a sink for the information loss that relativists believed in back then), and then the universe is tuned to maximize the number of black holes formed, constrained by the condition that life is possible.

These types of ideas are anthropic, and they are hard to make testable. Even if the universe is replicating itself and changing, it is not life. Life is not about replication. Fire replicates itself, and tries to maximize cumbustible consumption. Fire isn't life.

Life is when you have a computer in nature. That's not the case for universe-forming processes, because the universe is just not that complicated on the elementary scale. You can see the universe wasn't designed, and evolution and design are synonyms. All design is a process of evolution in your head, and all evolution in a complex system can be equally well called a process of design in a disembodied computational entity formed by all the evolving creatures.

Since evolution is a property of complex systems, and there is no complex system here, just black hole formation from galaxies (this doesn't allow a universal computer), you don't have evolution as I see it. You just have, at best, something replicating, like fire.

The theory is also incorrect because black holes don't lose information, and don't make new universes. The current universe we are in is also not particularly tuned for black hole formation. Further, the measure which tells you how to maximize is not at all clear: should you maximize the total number of black holes the universe will ever form? Does it matter if they form early or late? What's the weight? These questions are, to my mind, an abuse of language in the sense of Carnap--- they are positivistically meaningless.

share|cite|improve this answer
"All design is a process of evolution in your head" - awesome! – Slaviks Mar 26 '12 at 20:17
I saw you talking about computers several times, but I can't follow your perspective. If I think of a computer, then I see something which is able to perform different computations, something I interact with. If there are natural computers then noone is deciding what to compute. So if there is something that's being computed, then it's just an autonom evolution process. This means it's just a step by step action and I don't see why one would call that computer. But I also don't see how rule 110 is turing complete as I don't understand where one would put the input? A starting strings I choose? – NikolajK Mar 26 '12 at 22:06
@NickKidman: The way you see a natural system is a computer is if you can encode a computer into the initial conditions with a finitely computable encoding, and run an arbitrarily program on the coded information just by allowing the system dynamics to go forward in time. In rule 110, you take an arbitrary memory/instruction-set and encode it into a very long initial condition, and it will compute the result for you. Similarly, for random proteins, you can encode the informations in their conformations and bindings so that their evolution will reproduce the computation. – Ron Maimon Mar 27 '12 at 17:57
You have to be careful that the encoding is not too complex, or takes too long to find, because otherwise you aren't doing the computation in the system itself, but in the encoding algorithm. For example, in a randomizing automaton, you can always encode an arbitrary computation by making the map encode the initial conditions in the initial string, and some random string at some long time will be said to "encode" the result. This is an abuse of the word "encoding", you aren't encoding, you are computing the answer. It is usually easy to find encodings which prove systems are Turing complete. – Ron Maimon Mar 27 '12 at 17:59
In real physical systems, there is also always randomness, so the computation has access to a random oracle. This type of computation is stochastic computation, and it is strictly more powerful than Turing (or rule 110) computation. But it is not studied so well, perhaps because it is hard to define a random oracle precisely as a real number in standard mathematics. – Ron Maimon Mar 27 '12 at 18:02

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.