An Overview of Open-Ended Evolution: Editorial Introduction to the Open-Ended Evolution II Special Issue

Norman H. Packard, Mark A. Bedau, Alastair Channon, Takashi Ikegami, Steen Rasmussen, Kenneth O. Stanley, Tim Taylor

Abstract

Nature's spectacular inventiveness, reflected in the enormous diversity of form and function displayed by the biosphere, is a feature of life that distinguishes living most strongly from nonliving. This article introduces the second of two special issues on current research in Open-Ended Evolution (OEE) and provides an overview of the contents of both special issues. We present a simplified categorization of OEE and summarize progress in the field as represented by the articles in this special issue.

Introduction

In the four billion years of Earth's history, biological evolution has not merely solved the problem of survival; it has continuously invented new forms, functions, and entire ecosystems. This creative productivity is known as Open-Ended Evolution (OEE). While biological evolution remains the gold standard for inventiveness, artificial life (ALife) simulations often stall, reaching a plateau where no further "interesting" developments occur. This paper addresses the lack of a unified framework for defining and measuring OEE, which has historically made it difficult for researchers to compare results or identify the mechanisms driving the creative process.

The Tokyo Categories: A New Taxonomy

A primary contribution of this editorial is the synthesis of the Tokyo categories, a refined four-pillar taxonomy designed to classify OEE research. This framework simplifies earlier, more complex categorizations into four distinct kinds of ongoing generation:

  1. Interesting new kinds of entities and interactions: Focuses on the emergence of novel individuals or behaviors.
  2. Evolution of evolvability: Examines how the evolutionary process itself becomes more efficient or creative over time.
  3. Major transitions: Investigates the emergence of new levels of hierarchy, such as single cells forming multicellular organisms.
  4. Semantic evolution: A novel category focusing on how the "meaning" or function of an entity changes relative to its environment.
Figure 1: Mapping of research papers to the four Tokyo categories of Open-Ended Evolution.

Key Insights and Methodological Approaches

The papers within this special issue explore various mechanisms for achieving OEE. A recurring theme is the distinction between evolution as optimization (finding the best solution in a fixed space) and expansion (where the search space itself grows). True OEE is fundamentally about expansion.

Semantic Evolution and Functionality

Semantic evolution suggests that OEE is not just about new "shapes" or "code," but about how the functional relationship between an entity and its context evolves. For instance, a molecule's "meaning" might change as the chemical environment shifts. This highlights the importance of modeling how components interact to perform functions rather than just existing as passive structures.

Quantifying "Interestingness"

A major hurdle in OEE research is the subjectivity of "novelty." Researchers are now attempting to quantify "interestingness" using machine learning to detect statistically significant new structures in data streams. The MODES Toolbox, a standardized C++ toolset, has been introduced to measure OEE dynamics, including metrics for change, novelty, and complexity potential.

Figure 2: Visualization of a spatial ALife simulation, illustrating the complex interactions and hierarchical transitions studied in platforms like DISHTINY.

Results and Diverse Applications

The research presented spans synthetic and real-world datasets:

  • Cardinality Leaps: Sayama demonstrated that certain artificial chemistries can undergo "cardinality leaps," where the number of possible fit entities explodes.
  • Technological Evolution: By applying OEE frameworks to human patent data, Bedau et al. found that technological innovation follows patterns of niche innovation and major transitions similar to biological life.
  • Digital Ecosystems: Ikegami et al. used web services tags to study semantic evolution in digital environments.

Future Directions

Despite significant progress, no current ALife model matches the degree of open-endedness seen in biology. Future work must address three critical areas:

  1. Functionality: Better modeling of active functional interactions.
  2. Information Processing: Understanding the eons-long increase in the biosphere's total computational sophistication.
  3. Thermodynamics: Linking OEE to non-equilibrium physics and how energy flow drives the creation of complex, low-entropy structures.

The Tokyo categories provide a provisional roadmap for these efforts, expected to evolve as our understanding of these complex dynamics matures.

Summary updated

Abstract

Nature's spectacular inventiveness, reflected in the enormous diversity of form and function displayed by the biosphere, is a feature of life that distinguishes living most strongly from nonliving. It is, therefore, not surprising that this aspect of life should become a central focus of artificial life. We have known since Darwin that the diversity is produced dynamically, through the process of evolution; this has led life's creative productivity to be called Open-Ended Evolution (OEE) in the field. This article introduces the second of two special issues on current research in OEE and provides an overview of the contents of both special issues. Most of the work was presented at a workshop on open-ended evolution that was held as a part of the 2018 Conference on Artificial Life in Tokyo, and much of it had antecedents in two previous workshops on open-ended evolution at artificial life conferences in Cancun and York. We present a simplified categorization of OEE and summarize progress in the field as represented by the articles in this special issue.
BibTeX
@inproceedings{packard2019overview,
  author = {Norman H. Packard and Mark A. Bedau and Alastair Channon and Takashi Ikegami and Steen Rasmussen and Kenneth O. Stanley and Tim Taylor},
  title = {{An Overview of Open-Ended Evolution: Editorial Introduction to the Open-Ended Evolution II Special Issue}},
  booktitle = {Artif. Life},
  year = {2019},
  url = {https://openreview.net/forum?id=jp3yKcsiak},
}
Figure