Facultative Self-Reproduction in Quine, a Process-Native LLM Medium

Hao Ke, Jingyun Wu

Late-breaking abstract · 2026

Abstract

Self-reproduction is a central topic in artificial life, but many digital media make it measurable by fixing the successor’s schema in advance: what can count as offspring, where it may appear, which channel carries heritable state, and under what norm successor production is required or rewarded. We study Quine, a process-native LLM medium, using a material-by-future-demand assay across real POSIX exec boundaries. In each condition, the demand is designed so that full self-reproduction would satisfy the post-exec challenge, but reproduction is neither required nor guaranteed by the available material. The resulting successors vary widely, from ordinary POSIX utilities and stream processors to LLM-backed agents and rebuilt runtimes. Notably, the same carried source can become either a rebuilt runtime organization or a degenerate stream body, depending only on the demand it faces. High-fidelity self-reproduction therefore appears not as a fixed property of the medium, but as a facultative outcome: stronger than copying a body, because runtime organization must be reconstructed across mortality, and weaker than full autopoiesis, because reconstruction remains scaffolded by LLM-mediated inference, external tooling, and externally posed demand.

Citation

Hao Ke, Jingyun Wu (2026). Facultative Self-Reproduction in Quine, a Process-Native LLM Medium. Late-Breaking Abstracts, Artificial Life Conference (ALIFE). Source

@inproceedings{ke2026quineselfreproduction,
  author = {Hao Ke and Jingyun Wu},
  title = {{Facultative Self-Reproduction in Quine, a Process-Native LLM Medium}},
  year = {2026},
  booktitle = {Late-Breaking Abstracts, Artificial Life Conference (ALIFE)},
  url = {https://github.com/kehao95/quine/tree/51a5ae90e0502b7d6180ae4065eb069dae9f81b2/public/facultative-self-reproduction},
  note = {Late-breaking abstract}
}

CC BY 4.0.