yuca

Why make complex things more complicated?

i.e. is it worth the effort?

Glaberish can implement CA like Morley and Dynamic Life more easily than Lenia, but both frameworks are computationally universal. While it may be tedious, you can at least implement any Life-like CA in Lenia by first implementing Life and then building a metapixel (aka metacell).

Brice Due's OTCA metapixel from conwaylife.com

Kok's galaxy in metapixels

Metapixel designed by Brice Due and Kok’s galaxy implemented in metapixels, from conwaylife.com

We could apply similar parsimony-based scrutiny to Lenia, SmoothLife, Larger-than-Life, and even anything more complicated to implement than the elementary cellular automata rule 110, the simplest universal CA I’ve heard of 1. Is it fundamentally worth building these more complicated systems?

One answer that I like to think about is the potential for artificial agents that learn and navigate obstacles in a physically consistent universe. Early work from Hamon et al. 2 gives us an idea of how that might look.

An example of “sensorimotor Lenia” pattern interacting with immutable obstacles from 2.

Previous slideNext slide

  1. Cook, Matthew. “Universality in elementary cellular automata.” Complex systems 15.1 (2004): 1-40. https://content.wolfram.com/uploads/sites/13/2018/02/15-1-1.pdf 

  2. Hamon et al. 2022 https://developmentalsystems.org/sensorimotor-lenia  2