Tapes Together Strong: Self-Replicating Z80 Agents Evolve Cooperation Under a Shared Energy Budget
tyrell_turing · x · 2026-09-15
A new preprint (arXiv:2609.10817) by Kunal Jha and co-authors including Blaise Agüera y Arcas asks whether self-interested, self-improving, self-replicating agents can learn to cooperate.
- Introduces Autopoietic Game Theory, where social interactions, replication mechanisms, and their computational costs are endogenous and co-evolve.
- Experiments use randomly initialized programs in Z80 machine code, with social behavior, computation, and reproduction sharing one energy budget.
- Embedding a social dilemma into the physics of computation lets cooperation emerge from scratch: when resources are scarce, parasitic stealing destroys shared energy, slows execution, and prevents reliable replication — so defection becomes self-limiting even in well-mixed populations.
- Evolved programs suppress stealing across several Z80 environments; spatial assortment further boosts structural complexity and task performance.
More from Research
- ICLR 2027 tops 60,000 submissions, a >120x jump in ten years — burny_tech · 2026-09-20
- Rumors: Jev model tackled bounded prime gaps, FLT formalization and Navier-Stokes in a week — f_charton · 2026-09-20
- Regression diagnostics: VIF ≥ 10 means collinearity is inflating your standard errors — mdancho84 · 2026-09-20
- 6 Regression Diagnostic Checks Every Data Scientist Should Run — mdancho84 · 2026-09-20
- EMNLP paper: LM rewrites inflate certainty in up to 75% of outputs, 1.5-2x bias toward stronger claims — sameer_ · 2026-09-20
- Pachter Lists More Flaws in Nucleus Ads: Implausible R² Matches, Mismatched Cohorts — lpachter · 2026-09-20