This paper is about the possible resistance of GoL patterns to perturbations and the structures that could enable this to happen. They also want to know if resilience is a universal property of computational systems.
They test two types of resilience:
- Additive (add one or two live cells to the pattern)
- Negative (“kill” one or two live cells from the pattern)
They use 3 metrics for resilience:
- Equality: is the resulting pattern identical?
- Inclusion: is the resulting pattern larger?
- Cosine: amount of overlap between the two patterns.
Still life patterns were ranked according to the similarity metric, and this showed that a still life that ranks high on one metric rarely rank high in other metrics. Moreover, no still life was most resilient to more than one type of perturbation.
Still life seem to tend towards brittleness. Transient lengths were no longer than 2 steps most of the time. Symmetry was not found to help resilience, but patterns with density close to 1/2 seemed more resilient.