Table 1.
All previously described classes of evolutionary dynamics as measured with evolutionary activity statistics.
ClassMedian2evolutionary activity ($A¯$cum)ChangeNovelty (Anew)Diversity (or ecology) (D)ComplexityType of evolutionary dynamicsDescribed in
zero zero bounded bounded None Bedau et al. [6]
unbounded zero bounded bounded Uncreative Skusa and Bedau [42]
bounded positive positive bounded bounded Bounded Bedau et al. [6]
4a bounded positive positive unbounded bounded Unbounded Bedau et al. [6]
4b unbounded positive positive bounded Unbounded Channon [12]
4c unbounded positive positive unbounded Unbounded Channon [12]
ClassMedian2evolutionary activity ($A¯$cum)ChangeNovelty (Anew)Diversity (or ecology) (D)ComplexityType of evolutionary dynamicsDescribed in
zero zero bounded bounded None Bedau et al. [6]
unbounded zero bounded bounded Uncreative Skusa and Bedau [42]
bounded positive positive bounded bounded Bounded Bedau et al. [6]
4a bounded positive positive unbounded bounded Unbounded Bedau et al. [6]
4b unbounded positive positive bounded Unbounded Channon [12]
4c unbounded positive positive unbounded Unbounded Channon [12]

Notes. For each class, we show the response of all quantities measured for evolutionary activity statistics and in our proposed metrics (novelty and diversity should behave equivalently between the two systems). Note that we expect bounded evolutionary activity to imply bounded complexity, as any scenario in which complexity is growing without bound should imply that evolutionary activity is too. Question marks indicate that the value of a given metric is not specified in the description for a class of evolutionary activities. Higher-numbered classes are generally believed to fall further along the continuum of open-endedness than lower-numbered classes. In principle, classes 4b and 4c could each be further split into subclasses based on whether complexity is bounded or unbounded. Likewise, classes 1 and 2 could be further subdivided based on whether change is 0 or positive. In the absence of further data on the behavior of real-world systems, it is unclear how helpful such increased precision would be.

Close Modal