Aesthetic Selection of Naked Genes Milos Rankovic Independent Scholar Bergen, Norway +47 55 698511 milos@asnakedgene.com ABSTRACT The problem of genetic representation in (creative) evolutionary systems that aspire to taking part in cultural production is reformulated in the more general terms of the necessary structuring of the search space. Cumulative (aesthetic) selection of “naked genes” is discussed as an alternative approach whereby no germ-line/somatic-line distinction is explicitly implemented. The feasibility of this approach is illustrated on a computer application ASNakedGene, which therefore allows evolutionary runs to be seeded by arbitrary sets of digital (or digitised) images and generates variation through direct operation on the arrays of pixel values (as opposed to “genetic” representations of them). Categories and Subject Descriptors J.5 [Computer Applications]: Arts and Humanities – fine arts; I.3.6 [Computer Graphics]: Methodology and Techniques – interaction techniques. General Terms Algorithms, Human Factors, Measurement, Theory. Keywords Aesthetic selection, creativity, genetic representation, naked gene. 1. AESTHETIC SELECTION 1.1 From Nature with Creativity, Darwin On our unlikely planet, we find life capriciously negotiating the energy slopes, filling this film of space between the rock and the empty place with its intricate hierarchies. We see life ceaselessly discovering new niches and then hold Nature’s way of exploring the adjacent possible as the ultimate model of creativity. Since the initial Darwinian understanding, evolutionary principles have been applied anywhere from physical to cultural universes (e.g. [8], [2]), whenever a “non-question-begging” account is sought of phenomena that otherwise “beg” the evocation of some original intent [5]. This very generality reveals their computational nature, as is so conspicuous in the level of abstraction with which various computer programs model evolutionary processes. We therefore account for, or model, the “creative-like” search through the space of possible solutions by pointing to, or actually implementing, different evolutionary algorithms. Indeed, the more complex the distribution of relative value, or “fitness”, across a given space of alternatives, the more is evolutionary computation the toolbox of choice for its exploration. The area of interactive or collaborative evolutionary systems that aspires to provide the user with effective means of browsing the spaces of aesthetically distinct alternatives will be referred to in this paper as aesthetic selection and any resulting artists’ tools as meme breeders. Historically, though somewhat inadvertently, Richard Dawkins found himself effectively at the origin of both memetics and A- life art. In The Selfish Gene [3] he illustrated the universality of Darwinian principles by pointing to the cultural analogue of the gene — a unit of cultural heredity he called the meme — thus providing evolutionary approaches to the study of tradition with an unencumbered unifying term. Similarly, in The Blind Watchmaker [4] he illustrated the power of cumulative selection by describing the workings of a little program which has ever since been emulated by the growing community of scientifically inclined artists and artistically disposed scientists (see [9] for a historical review). Indeed, his may well be the first meme breeder [9] (in the sense of computer-aided creative practice, of course, for breeding of memes has always taken place at the scale of cultural traditions). 1.2 The Problem of Genetic Representation More than two decades later, elaborations of Dawkins’ original program are many (see [1], [9]), but the curious thing is that one can often tell with ease the program used just by the look of the images it helped generate [6]. Even when different people use the program, the outcome is still recognisably of that program and, typically, not at all of the user. To illustrate this problem, say, there was an exhibition of digital work all generated using one of four different image breeders. You are told that on show are works of one hundred artists, each of whom created one image using each of the four programs — the exhibition thus comprises four hundred digital images. The question is what would you expect to see: four styles exemplified a hundred times over, or a hundred styles represented each by four images? The ultimate goal of this research area is the latter scenario, but current systems almost entirely fit the former (see [6], [7]). This is potentially a fatal problem for the cultural relevance of meme breeders and has already caused the initial enthusiasm to all but fizzle out. Those that still believe in the potential of the original idea generally define the problem in terms of the Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. MM’07, September 23–28, 2007, Augsburg, Bavaria, Germany. Copyright 2007 ACM 978-1-59593-701-8/07/0009...$5.00.