STS 129: Paper 1
Word Count: 2200
Contacts for Nearsighted Watchmakers
“The last time a piece of mutable self-replicating code got out of control – when the selfish gene chain-reaction took off about 3.5 billion years ago – the harvest was very remarkable indeed” (Dawkins, 1992).
1991: the year of Tequila. Consumption of the fierce liquid had doubled since 1985, making it the fastest-growing distilled spirit in the United States. In cyberspace, Tequila highjacked the information superhighway – the world’s first non-research polymorphic computer virus ravaged PCs across the globe in April 1991. Tequila polluted local executable files such that when users ran the infected program(s), the virus appended itself to the hard disk’s file storage area, altered Partition data, and modified the Master Boot Record to “point” to itself. Moreover, and perhaps most troubling, the liquor-turned-virus differentially mutated its own code with each infection, making detection extremely difficult for anti-virus software, including Symantec’s Norton AntiVirus launched in December 1990. A computer infected with Tequila suffered myriad File Allocation Table errors, and terminal data loss.
In January 1992, Bulgarian hacker “Dark Avenger” released a mutating engine (MtE) – a toolkit for virus creation. Equipped with an OBJ file and the source code of a simple virus, the MtE endowed the latter with full polymorphic capabilities. Early in 1993, Masouf Khafir wrote the Trident Polymorphic Engine (TPE), a more sophisticated descendant of MtE. Both engines enabled worry-free, hands-off polymorphic virus writing, and duly frightened companies that specialized in anti-virus measures. That year viruses Girafe, Trigger and Bosnia – all products of MtE or TPE – infected their targets with a modified or encrypted version of themselves; and, “by varying the code sequences written to the file (but still functionally equivalent to the original), or by generating a different, random encryption key, the virus in the altered file [was not] identifiable through the use of simple byte matching”(Spafford, 1997). “With no fixed signature to scan for, and no fixed decryption routine, no two infections [looked] alike. The result [was]”, and still is, “a formidable adversary” (Symantec, 1996).
Today network administrators, businessmen, and home users describe computer viruses as security hazards, and detriments to the virtual fabric of everyday life. They traverse worldwide networks thanks to global communication technologies, causing computer scientists, industry executives, and college students to cringe in “cyber fear”. Consider, for example, internet-enabled NewLove, a polymorphic worm that wreaked havoc on companies in September of last year. Clogging mail servers and erasing files, officials estimated that NewLove cost businesses, on the average, $120,000 based on lost productivity and other measures (CNN.com). Damages wrought by the “I love you” virus in April 2000 totaled nearly $100 million, affecting an estimated 60 to 80 percent of US companies (E-Commerce News). Briefly, viruses are bad. They bring about destruction. But a more probing question – indeed, the primary subject of this document – remains: do these self-disseminating, self-replicating pieces of computer code constitute a form of life? In particular, do polymorphic viruses warrant the phrase “living”?
In an attempt to muddy the waters of life and non-life, this document presupposes the following: (1) “It’s just an incidental fact that in real living things the entities that happen to be organized are made of organic, soft, squishy stuff” (Gleick, 1987); that is, according to universal biology, carbon-based organic structures are not a prerequisite for, but rather a subset of life. (2) “Artificial” means initially produced by humans; therefore, in light of various reproductive technologies, Dolly the sheep, and genetically engineered foodstuffs, that something is artificial does not preclude its living. This document proceeds, then, to observe how viruses satisfy properties associated with life. Specifically, it (a) examines the virus’s genotypic and phenotypic properties along with its innate propensity for self-reproduction; (b) juxtaposes its differential self-mutation and replication with “genetic selection”, a process complementary to Darwin’s theory of stepwise evolution by natural selection; and (c) considers the aggregate behavior of polymorphic computer viruses as characteristic of genuine life.
In organic life forms, as in computer viruses, genetic constitution (the genotype) serves as a contributory causal factor in the development of individual conduct and or configuration (the phenotype) within some specific environment. That is, the genotype is any “set of low-level rules”, and the phenotype is the “behaviors and or structures that emerge out of the interactions among these low-level rules” (Langton, 1987). To put it yet another way, the genotype is the architectural blueprint, whereas the phenotype – realized by way of surveyors, housing committees, and construction workers – is the resultant two-story house. It is patently clear that both humans and computer viruses possess distinct behavior-guiding genotypes: DNA controls the former while “code” – lines of particularly situated ASCII text – dominates the organization of the latter. However, a computer virus’s phenotype can be difficult to interpret, much less visualize.
The phenotypic effect of a gene is usually conceived as the structure that carries the gene. Thus the human body, including its reproductive apparatus, is the most obvious phenotypic expression of human genes. In this way living organisms are the product of their low-level genetic rules. But in The Extended Phenotype, biologist Richard Dawkins argues, “Why stop there”? He proposes that the phenotype is anything that results from the gene’s expression. This includes, for example, a beaver’s dam, or – in talking about computer viruses – vigorous deletion and inadvertent transmission of infected program files by a PC user. It follows then, that organic life forms and computer viruses share this discous, extended phenotype; bandwidth bottlenecks and recurring system crashes, like boroughs and beach condos, indicate manifestations of specific genetic instructions. (It is no coincidence that anti-virus companies underline the extended phenotypes of polymorphics when pitching their software to prospective buyers.) But how do viruses affect entire communities comprised of millions of personal computers? Why do six billion human beings inhabit the Earth? In short, life reproduces. Cyber Tequila included.
Richard Dawkins maintains that replicators – mechanisms that store and copy information – produce all living things (Woolley, 1990). The replicator for natural life is the gene, made up of DNA; equally, Eugene Spafford of Purdue University notes, “the code that defines the virus is a template that is used by the virus to replicate itself” (Spafford, 1997). Hence it comes as no surprise that polymorphic viruses spread within a local machine or across networks by altering other programs to include copies of themselves. The infected program, or programs, interacts with other applications, local or remote, and transmits the virus ad infinitum. Likewise, human instructions for cell reproduction are executed when transcription processes utilize DNA sequences as a template for protein synthesis. Replicators – the instigators of all life – account for human and virus duplication.
In sum, human germ cells and polymorphic Tequila exhibit a genotype/phenotype distinction, as well a capacity for self-reproduction. But that’s not all. Remember that polymorphics change with every infection – a process analogous to naturally occurring “crossing over” and or mutation during cell division. This results in genotypic and (extended) phenotypic variation that in turn, engenders disparate survival rates among offspring.
The process that Darwin called natural selection, “the preservation of favorable variations and the rejection of injurious variations” (Darwin, 1859), has been redefined over the years to its current meaning: the differential reproduction of genotypes. This genetic selection (GS) amounts to a gene and genotype sorting process: genotypes whose net effect on phenotypes results in a better-than-average design for survival and or reproduction in a given environment will increase in frequency in a population. Such genotypes, and the specific genes within them, are said to be “favored” or “selected for” by the environment. In this way, genetic selection acts as one major force of evolutionary change (Durham, 1991). Consider the following scenario:
A polymorphic (or metamorphic) species of virus named “Stanfurd” encounters a vulnerable host system – the environment. Unlike the vast majority of computer viruses, Stanfurd differentially self-mutates in order to avoid early detection by anti-virus software – the predator. Immediately, Stanfurd has a reproductive advantage over “second-generation” viruses that utilize static, unique sequences of bytes – also known as signatures – to transmit the infected status of susceptible files or systems to related kin (Spafford, 1997). Each polymorphic progeny requires a new detection program; that is, each child copy of Stanfurd looks different than the other copies, thereby undercutting a predator’s ability to compare catalogued viral signatures with abnormal code on the host system. In view of GS, it is evident that the host machine favors, or selects for Stanfurd’s genotype. The polymorphic virus is more reproductively “fit” precisely because its offspring survive and thrive in greater numbers relative to its competitors. In time, though, anti-virus software detects and eradicates certain Stanfurd variants. Other robust Stanfurd strains remain on the host system, differentially copy, and propagate within the current and neighboring environments. Assuming that different host systems signify different environmental constraints, genetic selection will again “favor” certain Stanfurd variants. This process continues so long as reproductive resources persist.
To put it bluntly, polymorphic computer viruses evolve. Contrary to various Artificial Life scholars, it is the opinion of this document that polymorphics undergo evolutionary processes once released into cyberspace. Indeed, they typify anagenetic evolutionary change: the sequential transformation within a population or species. Insofar as this is true, it is possible to speak of directional and disruptional selection, speciation, Hardy-Weinberg equivalence, Mendelian genetics, and myriad other evolutionary terms. Even so, it is important to raise a pair of cautionary points about the role of genetic selection in evolution.
(1) The optimized computer virus does not represent the best of all possible solutions to cyberspace survival. This misconception – termed the “adaptationist programme” and the “Panglossian paradigm” – rightly identifies natural selection of genes as the primary agent of change, but ignores nonoptimizing evolutionary forces as significant sources of variation. Consider for example, the variability of operating systems and file formats – Windows, Macintosh, Unix, and Linux systems will demonstrate differing architectural vulnerabilities, thereby influencing the kinds of genotypes beneficial to polymorphic viruses. (2) “The current design of an organism does impose real limits upon the ‘choice’, or sorting process of genetic selection”. Otherwise, the virus “could be regarded as a sphere, ready to roll in any direction and always subject to changes over which its current form had little or no influence” (Durham, 1991). Viruses resulting from parental-self-modification do inherit a set of constraints.
In short genetic selection, like Darwin's theory of descent with modification is materialistic, opportunistic, and non-teleological: there are no nonmaterial forces at work in the evolutionary process, and GS is not goal-driven. Once a human programmer releases his polymorphic virus into cyberspace, GS, claims Richard Dawkins, “does not plan for the future. It has no vision, no foresight, no sight at all… it is the blind watchmaker” (Dawkins, 1987). The aforesaid concerns and clarifications are further proof that polymorphics evolve. But do they live?
Parts is parts is parts
Humans and polymorphic viruses exemplify nonlinear systems. That is, “the key feature of [them] is that their primary behaviors of interest are properties of the interactions between parts, rather than being properties of the parts themselves” (Langton, 1987). It is convenient then, that the computer virus can be broken down into its basic pieces – namely, lines of code – in an attempt to synthesize intricate conduct.
Take a single line of polymorphic code. This could be a reference to a parent function, part of a search algorithm, or a simple printf command. Individually the ASCII text is not behaviorally impressive, nor does it impress upon the observer any sort of “existence”. But taken together, the total code represents a polymorphic virus who mutates, migrates, invades, evolves, and dies. Akin to any living organism, the virus demonstrates authentic “existence” activity. Insofar as life is made up of behavior, and not of stuff, that computer viruses consist of simple stuff does not preclude their exhibiting complex behavior under the right conditions. Keep in mind, “The ‘artificial’ in Artificial Life refers to the component parts, not the emergent processes”. And so, “If the component parts are implemented correctly, the processes they support are genuine – every bit as genuine as the natural processes they imitate” (Langton, 1987). Consequently, this document contends that polymorphics, like organic life forms, live.
Richard Dawkins notes, “computers… are machines that do exactly what you tell them but often surprise you in the result” (Dawkins, 1987). So do viruses strictly adhere to their programming, but surprise the world with their inherited penchant for propagation, frustration, and destruction. Indeed, according to Computer Economics, the economic impact of virus attacks totaled $17.1 billion last year, up from $12.1 billion in 1999. But who is responsible for this binary sabotage? At whom do we point our mouse-clicking fingers? Ourselves. Humans create computer viruses. We create life.
The process of evolution – Dawkins’s Blind Watchmaker – has produced you and I: partially seeing watches with the capacity to create other, smaller watches. However, “we cannot foresee all of the possible consequences of the kinds of manipulations we are now capable of inflicting on the very fabric of inheritance”. We are but “nearsighted watchmakers” (Langton, 1987). In the same way computer programmers cannot anticipate virus behavior. Instead, polymorphics enjoy a semi-autonomous existence after human agents “breathe” life into their virtual bodies: they manipulate and proliferate independently of their creators. Hence, it is imperative that viral actuators reach for a set of contacts before polymorphic life extends beyond their control.
Artificial Life Online 2.0. Alife Online. 19 October
CNN.com. Technology. 17 October 2001.
Darwin, Charles. On the Origin of Species. London: Harvard University Press, 1859.
Dawkins, Richard. “Intelligence tests”. Times Newspapers Limited. 18 October 1992.
Dawkins, Richard. The Blind Watchmaker. New York: WW Norton & Company, 1987.
Dawkins, Richard. The Extended Phenotype. London: Oxford University Press, 1982.
Dietrich, Bill. “It’s Artificial, But is it Really Alive”? The Seattle Times Company. 6 July
Durham, William H. Coevolution. Stanford: Stanford University Press, 1991.
E-Commerce News. ‘Love’ Virus Damage Could Top $1B. 20 October 2001.
Gleick, James. “Artificial Life: Can Computers Discern the Soul”? The New York Times.
29 September 1987.
Godlovitch, Ilsa. “Creatures that can breed on the desktop”. Times Newspapers Limited.
20 November 1996.
Langton, Christoper. Introduction to Artificial Life. Santa Fe Institute Studies in the
Sciences of Complexity, 1988 (Alife I).
Nass, Clifford. The Media Equation. Stanford: CSLI Publications, 1996.
Polymorphism. A rough history of the computer virus. 16 October 2001.
Spafford, Eugene H. Computer Viruses as Artificial Life, in C. Langton, ed. Artificial
Life: An overview, Cambridge, MA: MIT Press, 1997, pp. 179-209
Symantec. “Understanding and Managing Polymorphic Viruses”. The Symantec
Enterprise Papers. Symantec Corporation, 1996.
Woolley, Ben. “Computers: Invasion of the Artificial Body Snatchers”. The Guardian. 7
 A Switzerland native, Tequila’s author did not intend to dole out the code; instead, an acquaintance stole it and contaminated millions of personal computers viz. his father’s shareware business.
 As defined by Richard Dawkins, universal biology is “the study of the principles of life in general, regardless of which planet we happen to be looking at. What we call biology is a subdivision of universal biology, dealing with the form of life that happens to have arisen on Earth” (Dawkins, 1992).