Noble, Jason, Seth Bullock, and Ezequiel A. Di Paolo. “Artificial life: discipline or method? Report on a debate held at ECAL’99.” Artificial Life 6, no. 2 (2000): 145-148.
URL1
How can artificial life (AL) advance scientific understanding? Is AL best seen as a new discipline, or as a collection of novel computational methods that can be applied to old problems? And given that the products of AL research range from abstract existence proofs to working robots to detailed simulation models, are there standards of quality or usefulness that can be applied across the whole field? On September 16th, 1999 in Lausanne, Switzerland, a debate on these questions was held as part of the Fifth European Conference on Artificial Life. As the organizers, we wanted to foster a constructive discussion regarding the scientific status, and future, of AL. We were well aware that some of these issues had been raised before (e.g., Miller [2]) but we felt that earlier treatments had perhaps not reached a wide enough audience. The format for the debate consisted of contributions from invited panelists followed by an open discussion. The panelists were Chris Langton, Mark Bedau, Simon Kirby, and Inman Harvey—Hiroaki Kitano was scheduled to participate but regrettably could not attend the conference. We started by sketching a continuum of approaches to AL as science. At one extreme, there are researchers who use techniques such as genetic algorithms and animat-style simulations to look at existing problems. Typically the problems come from within biology. The work of Kitano and colleagues [1] on morphogenesis in Drosophila is one example. We see this work as exemplifying the idea of AL as a method, or a collection of methods, that could (at least in theory) be put to use by investigators in many different fields. At the other extreme is the view that AL opens up whole new ways of thinking; that it is a discipline in itself. An example of this is Ray’s [3] work on Tierra, that Ray and some other commentators have taken to raise fundamental issues about what it means to call a system alive. If the existence of the continuum is granted, two questions are raised. First, are all of the possible positions along it tenable? The skeptic might ask whether computer simulations of the kind developed by AL researchers ever add anything to existing formal methods in a discipline like biology. At the opposite end of the spectrum, one might be cynical about the possibility of objectively studying “life as it could be.” The second question concerns quality: How can we distinguish good work from bad? The two questions are not independent: If one sees AL research as some kind of thought experiment, one’s quality criteria may well differ from those of someone who is interested in more-or-less precise models of real-world systems.