Monday, August 20, 2012

On Transhumanism

Two years ago in 2010 for curiosity I immersed myself into a small computing project to study how evolution by natural selection works. As a result of this exercise the attached paper “Eight Queens with Evolutionary Computing” (aka Evocom) emerged. I wrote a program in Python language to solve Eight Queens problem using evolutionary computing. In due process I learned a lot about mechanics of natural selection. I could for instance simulate genetic drift by way of varying certain parameters; similarly I could study the effects of varying mutation probability and so on.

In this article I would like to demonstrate why Transhumanism would fail expectations if the lessons we learn from Nature is disregarded.  I am not going to analyse ethical implications, which is strictly out of my scope. I would like to propose a model in stead for being a precursor of further analysis in the area. I would like to draw our attention to population genetics.

Richard Dawkins makes a compelling argument in his debut, The Selfish Gene, stating we are mere vehicles of gene coalitions that use us as replicators. Genes themselves do not have consciousness but the net effect of them being cooperating in evolutionary scale certainly appears to be as such. So we are already robots, from an earthworm to humans.

Everyday more replicators are being created; the idea is to keep thriving in other bodies, until useful lifespan of the individual inevitably surrenders to entropy; the most fundamental property of our universe that is far grander and dominant than the natural selection.

Entropy (aka disorderedness of information) rises irreversibly and universally. Oddly enough life emerged as if to challenge and work against entropy. Each living being born is packed with densely ordered information. However almost instantly as the first cell division takes place, entropy starts rising until the death of the individual. In the end entropy wins.

So ultimately death is a necessity, or rather for the Selfish Gene a necessary compromise to cope with entropy. The Selfish Gene essentially buys time from irreversible entropy via replication. Consequently genes survive far longer than their carriers, albeit inevitable modifications due to copy errors or mutations of other origin.

Life is as much as about death as it is about life itself. There is a lot to be said in this area. For now it should suffice that extending the human lifespan without useful contribution to population would probably have an effect of pushing your limits against a proven model.

The most remarkable feature of evolution by natural selection comes from the fact that re-combination of genes and mutations occur in random fashion when a pair gives rise to an offspring. Randomness increases robustness of the algorithm and accelerates evolutionary process. However this is not sufficient for a given population to reach to a stable fitness peak. You also need diversity of genes in the gene pool, a sufficient initial population size where pairs are selected to breed, and number of generations participating a given evolutionary process spanned across time. Environmental conditions, distribution of and capabilities of predators and preys do not affect offspring selection, they merely affect survivor selection. This is explained in my attached paper to some extent.

For our purpose I would like you to focus on this simple picture, the model of evolution by natural selection.

I should confess I have not made an exhaustive analysis on theories surrounding Transhumanism, so my knowledge in the area is limited. However I have made some observations to convince me that they and we as humanity are not heading in the right direction.

Evolution by Natural Selection

Here are the fundamental problems I see in the light of what I learned about natural selection.  Please use the diagram I provided in relation to arguments given below.

1- Transhumanism focuses on architecture/charactersitics of a given individual. Whereas evolution acts upon populations not upon individuals. The individual is not selected by itself or for itself, it is selected from a population (see survivor selection branch in the diagram).

Note: strictly speaking for our species survivor selection has a broader meaning than life vs. death. It may mean for instance “become sufficiently successful in business despite GFC”.

2- Transhumanism invariable erases the randomness from recombination and mutation phase. This is problematic in terms of population genetics as it will create a major obstacle for stability. Random recombinations and mutations are essential ingredients for species’ long-term survival.  So far I have not seen anyone grasped this notion. Most Transhumanists seem to be fire dancing for the wrong reason.

3- Due to profit concerns Transhumanism will likely to create a heteregenous gene-scape at the very beginning. Islands of genetic pools will emerge, an island of “creative” genetic pool, and island of “martian” genetic pool (for manufacturing people who can withstand harsh environmental conditions) and so on. This sort of genetic fragmentation shall inevitably promote genetic drift, in other words it may push those islands to loose other genetically essential traits rapidly.  This will likely to happen because a genetic pool in its entirity has NP complexity. Regardless of your computational power there is no mathematical proof so far that NP-complex problems can be solved rapidly compared to the rate of problems emerging in such systems. For instance you may see a relatively sudden rise in Albino in “cerative” pool. Transhumanists may argue, those anomalies can easily be “patched up”.  But I doubt “patch management” will be a sustainable strategy for the long term success.

In summary I see Transhumanism as a failure of some sort if lessons learned from natural selection are not sufficiently factored in. I think the scientific community has an obligation to urge governments that policies must also include proper risk analysis of population genetics and its dire implications, should sufficent emphasis is not given into the matter, in addition to ethical concerns.

There is room for success if we control our indulgence for profit and focus on useful and sensibly limited areas, such as curing certain diseases and solving energy crisis.


Evocom (my paper)

NP Complexity

The Selfish Gene

No comments: