On the face of it, these arguments suggest that it might be possible to build an artificial world that exhibits these simple processes on a computer, such that entities evolve within it which could be regarded as living organisms. Most people would agree that the chances of success of such an endeavour are slim. However, there is less agreement over the reasons why the enterprise might fail. Some would argue that life is a phenomenon which is fundamentally associated with the material world; for example, they might argue that some of the processes associated with living organisms, such as metabolism, could be simulated on a computer, but that a computer program could never be regarded as really metabolising. On the other hand, others would argue that life is fundamentally a process (or set of processes) and is quite independent of its particular medium of implementation; they would be quite happy to accept that artificial life could evolve on a computer. However, even some of these people would predict the failure of the endeavour, but for more practical reasons such as the lack of processing speed, memory, etc., of today's computers.
An attempt to create artificial life might therefore seem ill-fated from the start, but I believe it is still worth pursuing for a number of reasons. First, the approach of building a synthetic evolutionary system is very different to the traditional approach taken in theoretical biology of analysing evolution by tracking changes in population-level measures (e.g. gene frequencies) using models based upon differential equations. Different approaches involve looking at systems in different ways; one approach might suggest answers (and, indeed, questions) whose significance is not apparent from another approach. In this way, the synthetic approach can complement the more traditional approaches of theoretical biology, and lead us to ask different sorts of questions about evolution and life. This being the case, the endeavour might be worthwhile pursuing even if the attempt ultimately fails in achieving its grand goal. Secondly, we do not know how much of a barrier the practical limitations imposed by today's computers, mentioned above, really are. These are issues which can be resolved empirically.
The potential benefits of this work therefore include a contribution to our scientific knowledge of biological evolution. Even if we are ultimately unsuccessful in reaching the grand goal of synthesising artificial life, the nature of the ways in which the attempt fails will be instructive. In addition, any success in the endeavour would have tremendous implications in the longer-term for many kinds of computer applications. The ability to evolve an unlimited variety of complex adaptations is of obvious significance to areas such as machine learning systems, evolutionary design, computer games and genetic art, to name but a few.