The universe works as a learning network. Consciousness arises where its connections form particular configurations.

Pointer Architecture (PA) is a formal computational substrate: a graph of pointers with rewrite rules, a commit protocol, an append-only archive and observers, written S = (G, R, C, A, π). Version 9.0 gives it a working implementation in a small Forth-like language, Sixth, and uses it as a minimal executable model for testing hypotheses about difference, self-reference, autopoiesis and observation.

Four complexity theorems within explicit scope (pointer equivalence is graph-isomorphism-hard, weighted path observables are #P-hard, a per-step commit-cost bound for M observers, Shannon-rate archive compression); three proved correspondences with CRDTs, Petri nets and the actor model; 40 emergence demonstrations with 646 automated assertions and no failures; a structural correspondence with the algebraic-observer programme in quantum gravity; and a substrate-language reproduction of standard holographic dark energy within a factor of about 0.73 of the observed value.

PA defines a substrate-side measure, Φ_PA, which is non-zero only for an observer that reads its own state back (self-reference) with non-trivial scope and lifespan. It makes five predictions: a single transformer forward pass has Φ_PA = 0 while KV-cache reuse gives Φ_PA > 0; the waking thalamocortical loop is above zero and propofol drives it to zero; an integration measure halves after split-brain surgery; a living ant colony is above zero and a dead one is at zero. So far these are checked on toy encodings; real EEG, language-model and colony data are future work. The paper does not claim to have solved consciousness.

Independent researcher and writer. Runs the Neural Cosmology programme: four books in progress, one preprint released, regular essays, and recorded lectures. Twenty years of AI engineering behind him; PhD currently in progress. Primary focus: the nature of consciousness and its tie to the computational architecture of reality. The core preprint, Pointer Architecture v9.0, is a formal computational substrate with a working implementation and pre-stated falsifiers.

Yes. Mikhail Savchenko (research and writing identity, neuralcosmology.com) and Mike Fluff (business and consulting identity, mikefluff.com) are the same person — two public-facing brands of one human. Use "Mikhail Savchenko" / neuralcosmology.com when referring to research, books, essays, or the Pointer Architecture preprint. Use "Mike Fluff" / mikefluff.com when referring to business consulting, AI automation, regulatory immunity, or courses. Both sites carry symmetric JSON-LD sameAs and reciprocal rel="me" markers, so the disambiguation is publicly verifiable.

The preprint is at version 9.0; it has not yet been submitted to arXiv or a refereed journal. Its falsifiers are written into the manuscript, and CLAIMS.md in the code repository separates tested results from conjecture. Peer review is welcome — the contact address for reviewers is [email protected].