This content was published by Andrew Tomazos and written by several hundred members of the former Internet Knowledge Base project.

Immunities Passed & Auto-malware Detection

Evolutionary theory, despite the conflicts and quibbles between its proponents is a model that has great explanatory power. But that power is couched in the language of reduction. Reductionistic explanations are good because they "reduce" complexity to a comprehendible set of questions and answers. The problem with reductions is that they have to throw away or ignore data that is as yet uncomprehended or poorly comprehended by the theory.

Part of the difficulty of using the analogy of evolution to describe computed behavior has been in the recent past the belief that computers produce predictable results based on a straightforward input::process::output algorithm where the process part is clearly understood. I just read in one of the recent daily journals that no one actually knows what is going on any more inside the processing chip. Yes, we understand what it does, but not the decisions it makes while processing that produce the result. The reason for this is simple. We have designed our chips to manage a level of complexity that we cannot any longer do manually. Heuristic algorithms make judgments about the behavior of chips that we can't predict. We plot the behavior to discover the relative success of the algorithm statistically.

So it may be that computers are now complex enough to be describable by the evolutionary metaphor when their complexities can be reduced to parallel each other. The web certainly is complex enough. Our exasperation over poor search tools is telling about the level of complexity and the organic nature of the web. There are other elements in the web that parallel organic nature. Disease and disease vector research certainly take up a good deal of our effort.

Lamarck roughly gives us a path for the development of organisms as the inheritance of acquired characteristics. (This is clearly a reduction of his view, like "survival of the fittest" is a reduction of Darwin's.) But modernize this a bit and we see how organisms acquire defense characteristics by building in an autonomic response to certain diseases or disease types. To avoid the reductionistic determinism of genetic inheritance, we also receive from our parents a plethora of immunities as well as bacteria and other organisms that have shaped our bodies and minds. We are in many respects a superorganism that no longer functions correctly without all of our symbiotes. We inherit more than our genome describes for us.

In computer terms, we build systems that include in them defenses against disease. We learn to defend our systems from generation to generation. Each new system has these acquired characteristics or dies quickly. At first the defenses and tweaks are "utilities" then they become system level tools or processes, invisible to the user, not part of the genetic "kernel" but symbiotic with it. These utilities may even force change in the kernel because of perpetual attack. View now the microprocessors that have built-in malware protection. Deeper than the kernel, it is now part of the physical substratum.

For biology, this is not much of a problem. The genetic code that already has this tweak by mutation or ordinary genetic variation will survive with or without the "bacterial" utility, though it will not conflict with the utility. Its chances of reaching maturity to reproduce are better than the individual without this tweak. If the disease is virulent enough, then the individuals that successfully pass on the bacteria and those who have the genetic tweak will be the only survivors, eventually blending their gene pools ensuring the redundant utility as well as the genetic tweak.

However this redundancy entails a certain "overhead" that requires better processing power, or all the anti-viral protection will eventually bog the system down until it can not function as a reproducing member of the population.

Back to Index