On being "turing complete"

The Caltech Project Von Neumann defines "Turing complete" this way:

A language is considered Turing Complete if it is able to compute any function computable. It is still counted even if it takes the language a huge amount of time and memory to complete this function. It is somewhat important to try to make the Alife Engine turing complete so that we can try to maximize the possible behaviors evolvable in the creatures.

While Caltech is exploring simulations with artificial life ("a-life"), or self-regulating processes that in my terms are MAWBA ("may as well be alive"), my previous post here used the term Turing Complete regarding the actual, hierarchical, fractal, multi-level ladder thingie on which  humans occupy some middle rung.

As a prior post noted, if we assume, one way or another, that the world around us incorporates some very powerful design patterns, then we would expect recursion and a Turing Complete multi-level emergent computational ability to be one of them, as the Caltech project does for a-life. 

In other words,  we don't want to repeat the mistake Marvin Minsky made regarding perceptrons and single-level neural networks as being incomplete,which set Artificial Intelligence back 15 years.  Minsky, of course, later retracted his error, recognizing that 3 level neural nets can be complete, if they also include feed-forward.

Similarly, it seems silly to continue to explore single or two level versions of  "multilevel models" for human beings in social context (which is all that really matters for public health.) There is solid reason to believe, on general principles, that the emergent, evolving beastie is 3-levels, with feed-forward, or more complex than than.

Because each level can succeed in breaking loose from the levels below it, and taking on a mostly independent life of its own,  there is no way this can be "determined" upwards from "selfish genes."   Only by including vertical feedback and feedforward pathways that span 3 levels, that smear out "causality" via phase-lock loops,  can we achieve Turing completeness, and we should assume "Life" has figured that much out by now.

Especially, if even artificial life models at Caltech have figured that much out.

Conclusion - it is not possible to find a solid "causal pathway" from Dawkins and Darwinian evolution with "selfish genes" until you look upwards and include at least two higher levels of hierarchical organizational groupings, that are co-evolving independently. That whole gemisch is perfectly capable then, of being Turing complete, and figuring out what even most CEO's of US corporations can't -- that global altruism is a higher-level and more fit, longer-range better solution than local selfishness.

In the long run, in the large scale picture, we are all collaborating, not competing, and our co-evolution is mutually beneficial.   Coherent solutions to the US Army problems require coherence and integrity;  coherent solutions to nuclear power plant operations or aircraft cockpits require coherence and integrity across multiple levels of management, not just a single level.    Then, innovative solutions are visible.  Then, we are all Turing complete.

So, in that view, there is no surprise that humans are hard-wired to have emotions that drive them to altruism or larger group survival pathways, away from local dog-eat-dog competition models. The savage battle for local fitness misunderstanding is based on not seeing the higher levels that are operating in parallel defining the rules for the locally "selfish" interactions.

Also, in this view, it is not at all surprising to find massive depression and even the human version of apoptosis (cell suicide) resulting from philosophies (such as materialism) or behaviors that cause "individual" humans to break "free" of the "bonds" to other humans and the larger computational engine and be "rugged individuals."

Computer networking people figured out last century that networks of collaborating smaller computers have massively more power than trying to make a single "supercomputer".  LIfe doesn't try to make an "amoeba" with a single-cell that spans the earth - a model that doesn't scale up.   Life transitions to "multicelluar" forms at or around this point.  That's what humans need to see as the design pattern that works.

We need tightly knit teams that work as one, and there is where humans are hard-wired to find joy, happiness, fulfilment, and the abilty to crank out products that fill the bottom line.

technorati tags:positivepsychology, multilevel, hierarchy, alife, artificiallife, caltech, turing, completeness, turingcomplete