Exploring emergence, persistence, learning, inheritance and evolution in computational systems.
Can life-like behaviour emerge from computation alone?
This book starts with the smallest possible systems — state, local interaction and time — and progressively asks harder questions about persistence, information, regeneration, learning, inheritance and cumulative improvement.
The rule throughout is simple:
we do not claim a life-like property until we can demonstrate and measure it.
Before we learn how artificial life works, look at what a few local rules can already produce.
Read chapter →Strip digital life down to a line of binary cells, a three-cell neighborhood and one tiny deterministic rule.
Read chapter →Change one rule number and watch a tiny deterministic system become unexpectedly difficult to predict.
Read chapter →Use Conway's Game of Life to ask when a recurring localized pattern becomes useful to treat as an entity.
Read chapter →Damage persistent digital patterns and separate survival, robustness and regeneration into experimentally distinct claims.
Read chapter →Ask what digital reproduction actually requires, from simple copying to causal reproduction, inheritance and lineage.
Read chapter →Variation, inheritance and differential reproduction can produce evolution in digital systems. But evolution is not the same thing as progress.
Read chapter →Turn appealing observations into experiments by defining properties, measurements, interventions, controls, confounds and bounded claims.
Read chapter →Stop treating biological life as the specification. Remove biological constraints and ask what organization actually requires in a digital substrate.
Read chapter →If digital life does not inhabit the same world as biological life, why should it inherit the same limits?
Read chapter →Begin with one seed on a hexagonal lattice and ask how far growth, damage, defects and history can take us before anything resembling life is required.
Read chapter →We examine Outlier, an unusually minimal artificial-life system in which causally verified self-replicating structures emerge from local cellular-automaton dynamics.
Read chapter →Can a local computational growth process turn an external signal into persistent morphology? We build a Digital Crystal, hide the signal that formed it, and test what the final structure actually preserves.
Read chapter →A Digital Crystal can preserve source statistics in morphology but not temporal order. We add state, checkpointing, event history, replay and branching to discover what a recoverable digital past actually requires.
Read chapter →A Digital Crystal can causally alter another through a one-bit event, but causal transmission is not yet sender-specific signalling.
Read chapter →A Digital Crystal has a causal past. Chapter 17 asks whether that past remains legible, and discovers that even the way randomness is held fixed changes the answer.
Read chapter →A pulse can leave a permanent mark in a Digital Crystal. Chapter 18 asks the harder question: when does a persistent mark remain causally usable?
Read chapter →Two different experiences can leave different material traces in the Digital Crystal. Chapter 19 asks whether those differences actually matter when the future arrives.
Read chapter →The Digital Crystal had always lived in a world where occupied material lasted forever. Chapter 20 removes that assumption and discovers that loss creates new construction opportunity.
Read chapter →Once material can disappear, the Digital Crystal can rebuild what is lost. Chapter 21 asks what happens when computation itself becomes scarce.
Read chapter →After finding stable turnover, causal locality and dynamically generated interfaces, Chapter 22 asks whether the Digital Crystal is best understood as a bounded thing or as a coherent process in space and time.
Read chapter →After failing to find a propagating process field, Chapter 23 replaces correlation with intervention and asks what one additional Digital Crystal attachment actually causes.
Read chapter →A reset of Chapter 24 reveals that local causal effects are redistributed through a finite global evaluation budget. The strongest result is not a local gain field, but a selector-mediated far-field effect that follows frontier change.
Read chapter →Chapter 25 isolates finite candidate selection as a control parameter for non-local causal redistribution in the Digital Crystal.
Read chapter →Chapter 26 tests how finite candidate selection changes the computational pathway of a local causal perturbation, while dynamically matching expected background construction.
Read chapter →Chapter 27 tests whether hidden material state can change the causal response of identical visible geometry, and whether that effect persists after the original trace has decayed.
Read chapter →Chapter 28 tests whether the Digital Crystal contains a privileged causal individual, and shows why strong apparent modularity can arise from observer-chosen spatial boundaries alone.
Read chapter →Chapter 29 turns the recent experiment failures into a reproducible epistemic method: distinguish invalid runs, unresolved questions, bounded negative results, and supported findings narrowed by stronger controls.
Read chapter →The final chapter synthesizes the experimental record into a provisional substrate-first specification of digital life, distinguishing what was supported from what biology tempted us to assume.
Read chapter →A cross-chapter ledger of recurring phenomena discovered during the Digital Life experiments, including what survived failed hypotheses, where each phenomenon recurred, what mechanisms may explain it, and what remains unearned.
Read chapter →