Open-ended evolution (OEE) is a major area of research in artificial life. While there is not full consensus on how to define the specifics of OEE, at a high level it describes an evolving system that will never settle into a single stable equilibrium. Some researchers argue that the continual generation of novelty is sufficient for a system to be labeled as open-ended , while others have suggested that OEE “refers to the unbounded increase in complexity that seems to characterize evolution on multiple scales” . In practice, these definitions are closely inter-related, and OEE is often used as an umbrella term to refer to the study of them all collectively. The key idea of open-ended evolution is that a system produces organisms that are continuously evolving and changing, rather than organisms eventually reaching a state where nothing changes.

Proposed Hallmarks/Metrics of OEE

How do we know if a system is exhibiting OEE? While some have suggested that “I’ll know it when I see it”, others have suggested that OEE is not a binary and thus it is valuable to quantify where along the continuum of open-endedness a system falls. As such, multiple approaches to measuring OEE have been proposed. These approaches have generally involved two components: a way to filter out noise, and a set of quantities to measure.

Filtering out noise

Evolution is an inherently noisy and stochastic process. In any generation, there will likely be many new mutants that will not persist far into the future. Classifying these mutants as indicative of continued production of change or novelty would not provide us with a useful measurement of OEE. As such, we need a rule for deciding what can be filtered out. The following approaches have been proposed:

Shadow runs

Perhaps the earliest approach to filtering noise out of OEE metrics was the use of neutral “shadow” runs of an artificial life system . In this technique, each run of an evolving system would be paired with a run in which adaptive dynamics had been eliminated. Doing so provides a baseline level of OEE-like behavior that you would expect to see in the absence of adaptive dynamics, which you can then subtract out from OEE metrics in the non-shadow run.

A refinement to this technique, the shadow-resetting method , involves periodically re-syncing the shadow run to the main run.

Persistence filtering

This approach takes advantage of the fact that in artificial life systems it is easy to record the entire phylogeny (ancestry tree) of an evolving population . Organisms are considered to be worthy of inclusion in OEE metrics if their lineage persists for a predetermined number of generations (i.e. if they still have descendants alive in the population that many generations later).

Measuring OEE

Two overarching suites of metrics have thus far been proposed to measure OEE: evolutionary activity statistics and MODES . Each of these suites measures multiple aspects of an evolving system’s behavior in an effort to understand the ways in which it is or is not constrained. Evolutionary activity statistics measure: novelty, diversity, and total evolutionary activity. MODES metrics measure: change, novelty, diversity, and complexity.

Another hallmark of open-ended evolution that has been proposed is indefinite scalability . In the context of OEE, an indefinitely scalable system is one that will keep exhibiting more open-endedness the more resources you give it. Indefinite scalability of the complexity of individual organisms has been explored in Geb .

Hypothesized Conditions for Producing OEE

There are several attributes that have been hypothesized as necessary for an artificial system to produce open-ended evolution:

  1. Unlimited genetic space of potential genotypes. 

This hypothesis states that a practically unlimited genetic space for potential genotypic growth in the system should be present. If there needs to be an endlessly changing population of organisms, the genetic space should be unlimited and available for organisms. Genetic space refers to the potential genotypes that could exist. For example, if the organisms’ genomes are bitstrings of length 5, there is a fixed number of genotypes that are possible, leading to an inevitable limit on evolution. On the other hand, if the organisms’ bitstrings are of variable length, there are an infinite number of bitstrings possible, so evolution is not inherently limited by the genetic space .

  1. Mutational pathways with unlimited length between potential phenotype

This hypothesis refers to the phenotypes or features of organisms. Features of potential organisms within the environment should be able to result from many different mutational pathways. Systems that show open-ended evolution have pathways that can expand infinitely so that original organisms can evolve in multiple ways and make new types of organisms . For example, a lot of animals have similar features, such as dogs, cats, lions, and wolves all have four legs. They have different genomes and took different evolutionary paths to get there, but they all ended up having overlapping features. 

  1. Dynamic adaptive landscape for continual evolution

This hypothesis states that the adaptive landscape should be dynamic rather than static . While the first two requirements relate to the genotypes and phenotypes of the  organisms, this requirement leads to the other two requirements being “realized.” This means that the environment around the organisms should be dynamically changing instead of staying static. Since the population of organisms is evolving over time, the environment around them should also be changing according to their actions. This idea is similar to how our living environment, Earth, changes over time as humans develop and affect it through our actions. 


Open-Ended Evolution is tricky in the sense that the hallmarks of OEE and possible mechanisms need to be clarified in order for us to identify and distinguish different types of OEE. Therefore, it is still actively investigated by many researchers and the requirements listed above are just hypotheses. There are many possible criteria used to determine whether a natural or artificial system does demonstrate OEE and many researchers are looking for open-ended evolutionary growth and complexity in different types of artificial life and systems . Research related to open-ended evolution is a very active area and there are many questions still to be answered.

Further reading

For more information about open-ended evolution, check out the following resources:


Corominas-Murtra, B., Seoane, L. F., & Solé, R. (2018). Zipf’s Law, unbounded complexity and open-ended evolution. Journal of The Royal Society Interface, 15(149), 20180395.
Taylor, T. (2012). Exploring the Concept of Open-Ended Evolution.
University of New Mexico, Albuquerque, NM 87131, Ackley, D., & Small, T. (2014). Indefinitely Scalable Computing = Artificial Life Engineering. Artificial Life 14: Proceedings of the Fourteenth International Conference on the Synthesis and Simulation of Living Systems, 606–613.
Bedau, M. A., Snyder, E., Brown, C. T., & Packard, N. H. (1997). A comparison of evolutionary activity in artificial evolving systems and in the biosphere. In P. Husbands & I. Harvey (Eds.), Proceedings of the Fourth European Conference on Artificial Life, ECAL97 (pp. 125–134). MIT Press.
Bedau, M. A., Snyder, E., & Packard, N. H. (1998). A classification of long-term evolutionary dynamics. In C. Adami, R. K. Belew, H. Kitano, & C. E. Taylor (Eds.), Artificial Life VI: Proceedings of the Sixth International Conference on Artificial Life (pp. 228–237). MIT Press.
Channon, A. (2019). Maximum Individual Complexity is Indefinitely Scalable in Geb. Artificial Life, 25, 134–144.
Channon, A. (2003). Improving and Still Passing the ALife Test: Component-normalised Activity Statistics Classify Evolution in Geb As Unbounded. In R. K. Standish, M. A. Bedau, & H. A. Abbass (Eds.), Proceedings of the Eighth International Conference on Artificial Life (pp. 173–181). MIT Press.
Packard, N., Bedau, M. A., Channon, A., Ikegami, T., Rasmussen, S., Stanley, K. O., & Taylor, T. (2019). An Overview of Open-Ended Evolution: Editorial Introduction to the Open-Ended Evolution II Special Issue. Artificial Life, 25(2), 93–103.
Packard, N., Bedau, M. A., Channon, A., Ikegami, T., Rasmussen, S., Stanley, K., & Taylor, T. (2019). Open-Ended Evolution and Open-Endedness: Editorial Introduction to the Open-Ended Evolution I Special Issue. Artificial Life, 25(1), 1–3.
Dolson, E. L., Vostinar, A. E., Wiser, M. J., & Ofria, C. (2019). The MODES Toolbox: Measurements of Open-Ended Dynamics in Evolving Systems. Artificial Life, 25(1), 50–73.
Soros, L., & Stanley, K. (2014). Identifying necessary conditions for open-ended evolution through the artificial life world of Chromaria. In H. Sayama, J. Rieffel, S. Risi, R. Doursat, & H. Lipson (Eds.), ALIFE 14: Proceedings of the Fourteenth International Conference on the Synthesis and Simulation of Living Systems (pp. 793–800). MIT Press.