[Paper Review] Digital Genesis: Computers, Evolution and Artificial Life

Tim Taylor, Alan Dorin+1 more|arXiv (Cornell University)|Dec 7, 2015

Computability, Logic, AI Algorithms11 references3 citations

TL;DR

This paper traces the historical development of digital evolution and artificial life, from von Neumann’s theoretical foundations and Barricelli’s early computational experiments to modern efforts in open-ended evolution. It argues that integrating ecological principles—such as energy, matter, and environmental dynamics—is essential for creating truly open-ended digital life systems.

ABSTRACT

The application of evolution in the digital realm, with the goal of creating artificial intelligence and artificial life, has a history as long as that of the digital computer itself. We illustrate the intertwined history of these ideas, starting with the early theoretical work of John von Neumann and the pioneering experimental work of Nils Aall Barricelli. We argue that evolutionary thinking and artificial life will continue to play an integral role in the future development of the digital world.

Motivation & Objective

• To trace the historical lineage of digital evolution and artificial life from foundational theoretical work to contemporary research.
• To identify the core limitations of current digital evolutionary systems, which often stall in quasi-stable states after an initial burst of activity.
• To argue that incorporating ecological principles—such as energy, matter, and environmental dynamics—is essential for achieving open-ended evolution in digital systems.
• To advocate for a more principled, ecologically inspired modeling approach to bridge the gap between biological and digital evolution.
• To position evolutionary computation and artificial life as central to the future development of adaptive, autonomous digital systems.

Proposed method

• Analyzing the historical development of digital evolution through key theoretical and experimental milestones, including von Neumann’s cellular automaton framework for self-replicating machines.
• Examining Barricelli’s pioneering work using one-dimensional cellular automata to simulate symbioorganisms that exhibit self-reproduction, parasitism, and genetic exchange.
• Reviewing Turing’s early proposals for evolutionary machine learning via mutation and human feedback, despite limited success and lack of formal publication.
• Mapping the shift from artificial life goals to optimization-focused research in the mid-20th century, followed by the resurgence of artificial life in the 1980s via the ALIFE conference series.
• Evaluating modern attempts at open-ended evolution by identifying missing ecological components such as resource dynamics and environmental feedback mechanisms.
• Proposing that future systems must model physical dynamics and organism-environment interactions more rigorously to achieve sustained evolutionary innovation.

Experimental results

Research questions

• RQ1 What historical foundations underlie the application of evolutionary principles in digital systems, and how did they emerge independently across multiple researchers?
• RQ2 Why do most digital evolutionary systems fail to sustain open-ended evolution, despite satisfying basic conditions of variation, inheritance, and differential reproduction?
• RQ3 What ecological and physical dynamics are missing in current digital evolutionary models that prevent the emergence of complex, adaptive digital life?
• RQ4 How can the relationship between organisms and their environment be more accurately modeled to support long-term evolutionary innovation in digital systems?
• RQ5 To what extent can evolutionary thinking and artificial life concepts continue to shape the future of artificial intelligence and adaptive computing systems?

Key findings

• John von Neumann’s theoretical work on self-replicating cellular automata laid the first formal foundation for digital evolution, though it was never implemented during his lifetime.
• Nils Aall Barricelli’s experiments with one-dimensional cellular automata successfully demonstrated self-reproduction, parasitism, and genetic exchange among digital entities, coining the term 'symbioorganisms'.
• Turing’s early work on evolutionary machine learning, though dismissed at the time, anticipated modern neuroevolution and reinforcement learning approaches.
• From the 1950s to 1980s, most research shifted toward optimization, sidelining the original artificial life goals of open-ended evolution.
• Despite progress, most digital evolutionary systems reach a stable state after initial innovation, failing to sustain open-ended evolution due to insufficient ecological modeling.
• The paper concludes that future success in digital artificial life depends on integrating ecological principles such as energy, matter, and environmental dynamics into digital evolutionary frameworks.