Tuesday, 3 January 2012

On David Abel's "The First Gene"

I would like to draw the attention of the interested readers to this brilliant book: 
In particular, I would want to highlight some things which were key in improving my understanding of the evolution controversy. The book is clearly written and is full of insights.

1. Constraints Vs Controls. This is what the author calls the cybernetic cut. On the left of this divide, there are inert physical/chemical constraints determining conditions for spontaneous generation of low-informational redundant regularity. On the right, there is a world of active controls, i.e. of arbitrarily set rules to achieve a useful function. Empirical data available on mass suggests that while constraints are present in the physical world and are determined by physico-chemical interactions, the insertion of controls can only be done by an intelligent purposeful actor (or actors). In reality, various controls will necessarily be "instantiated into physicality" (Abel's expression). However, physicality only provides the substrate to the functional information conveyed by it, whereas the functional information itself is never reducible to physicality.

I find it useful to think about this in terms of complexity classes of decision problems. In the graph below (Fig.1), each lower class is included in its parent class. In other words, there are fast transformations of a problem in a child class into a problem in its parent class. However, no inverse inclusion or transformation have been found to date. They say that a parent class is irreducible to its child.

E.g. take two classes, NP and P. NP is decidable in polynomial time by a non-deterministic Turing machine which includes an oracle. In human language, it means that while no easy algorithm has yet been found for it, it is still possible to quickly check if a proposed solution is genuine. P is decidable in polynomial time by a deterministic Turing machine. So there are known fast algorithms to solve a problem in P. Clearly, we can represent a P problem as an NP problem. To achieve this, it is sufficient to replace the NP oracle with a polynomial time algorithm for the problem in P. However, no known inverse representations exist. So they say that NP is thought to be irreducible to P. 

Fig.1. Complexity Classes of Decision Problems. Source: Wikipedia, Complexity Class.

2. Formalism vs Physicality. While non-living nature operates within the confines of physicality and constraints, it is only formal things that operate with rules. Rules are arbitrarily defined irrespective of ("on top of") physicality. That is very important to understand. For example, if I want to assign some specific meaning to a random sequence of symbols, I can always do so in the context of a particular system. There can be different rules in different contexts. For example, using the Boolean alphabet {0,1} I am free to define 0 to represent TRUE and 1 to represent FALSE. Rules can, of course, be instantiated into physicality. This is done by means of what Abel calls configurable switches i.e. logic gates that bear 1 bit of information, represent purposeful choices of intelligent actors and prescribe/program the development/deployment/ontogenesis of a given system. The switches collectively perform a mapping from the far (control) side of the cybernetic divide to the near (physicality) side. Metaphorically speaking, the switches build a one-way bridge from the far to the near side of the cut. Based on overwhelming evidence, it is important to realise that it is impossible to move across the bridge in the opposite direction in reality because physicality alone has never been observed to give birth to control, purpose, utility or any kind of formalism. Models similar to Stuart Kauffman's edge of chaos which operate with hypothetical spontaneous self-organisation of control at the level of physicality are not empirically warranted.

3. DNA Is Prescriptive Information. Fundamental Problems of Macroevolution. It is known that DNA stores ontogenetic information i.e. information about how an organism will be developing. Importantly, this information is functional and its recordation and reading are formal processes. In other words, there is a certain alphabet, syntax and semantics in the language used to record and interpret the information. In other words, DNA is code in the strict sense. However, the only empirically observed source of formalisms in nature is intelligence. So to assume that high enough taxa (such as classes or phyla) can be generated step by Darwinian step is equivalent to assuming that prescriptive information can be generated spontaneously. There is no empirical data to warrant this hypothesis.

I can extend the above analogy of polynomial time irreducibility of complexity classes to show how I understand Intelligent Design interprets relationships between physics/chemistry, life, information, intelligence and consciousness (Fig.2). Again, this is my perception of the big picture of the said relationships.

Fig.2. The Intelligent Design view of hierarchical relations of physicality, functional information, life, intelligence and consciousness.

As with complexity classes, life is not reducible to physicality alone (i.e. to physico-chemical interactions and their constraints) but also requires input of functional information to control the ontogenetic processes (metabolism, replication, adaptation, reaction to stimuli, etc.). Likewise, intelligence is not reducible to biological processes alone and, furthermore, consciousness is not reducible to either biological processes or intelligence. It is a gray area and no generally agreed understanding of how intelligence or consciousness operate is available as yet. Arguably, supposing intelligence and consciousness can be formalised, they may require some sort of integrating information on top of their respective Shannon channels, similar to functionally specified information in digital form (dFSCI) that defines biological functions.

No comments:

Post a Comment