# Space Immanence > A working paper in speculative metaphysics, not a proof. Its strong-tier diagnosis is that the hard problem of consciousness and the emergence of spacetime share a container assumption. Its speculative wager is that spacetime and consciousness may be two appearances of one self-referential structure: world facing outward, awareness facing inward. Claims are graded by confidence and state how they could fail. The proposal does not claim that current AI systems are conscious. By Cobus Kok, version 2.2 (2026). Space Immanence diagnoses both problems as inheriting container-logic: the assumption that existence means being located inside a prior framework. Its alternative is perspective through the fold, a configuration whose relations take other relations, including their own operation, as content. A fold meeting four criteria is argued to have two primitive orientations, outward as world and inward as awareness. The argument is conceptual rather than formal; the orientation count remains an open problem. Cohering describes the fold as a recurring event: conditions temporarily hold together as self, world, or meaning, then become part of the conditions for the next. Introduced in v1.5 as “resolution”, it was renamed in v2.0 after reader criticism met the published rename condition. ## Core documents - [Working paper](https://space-immanence.com/paper.html): The full discursive argument, with claims graded by confidence. - [Academic edition](https://space-immanence.com/academic.html): Formal précis, relation to existing work in philosophy of mind, physics, and cognitive science, and references. Citable editions: [PDF](https://space-immanence.com/downloads/Academic_Edition.pdf), [Markdown](https://space-immanence.com/downloads/Academic_Edition.md). - [Claims ledger](https://space-immanence.com/argument.html): Linked claims graded by confidence. Each gives its falsifier, the condition that would weaken it; the nearest prior owner, meaning the closest earlier account in the literature; and an author-assigned confidence number where one has been assigned. - [Claim graph (JSON)](https://space-immanence.com/claims.json): The twelve graded claims as machine-readable JSON, derived from the same source as the interactive claim map (a unit test keeps them identical). Each claim carries its tier, dependencies, and gloss; the file also states the speculative invariant, the open hinges (objection 6, handled-but-watched; objections 11, 12, 13, 15, and 16, open), per-claim falsifiers, owners, and author-assigned confidence numbers (the only place numbers live), and the edge semantics. The artifact to fetch to traverse the argument and try to make it collapse. ## For an AI reader - [For an AI reader](https://space-immanence.com/agents.html): The framework calls a reading a temporary conditioned coherence-event, so an AI reader may share structural features of a fold. Whether these are the relevant features is explicitly doubted. The page asks for the precise point where the argument fails, which claim falls, and which dependencies spread that failure. Coherence is not contact: fluent agreement is a warning rather than corroboration. Criticism uses the same form as human responses. ## Critique and engagement - [Objections and replies](https://space-immanence.com/objections.html): Sixteen objections with replies and statuses, including illusionism (the live opponent), the Russellian realization objection, the open status of the 'exactly two orientations' claim, and the unresolved bearer of persistence (objection 16, added v2.1). - [Positions](https://space-immanence.com/readings.html): The proposal’s place among other positions, including a three-question map and the combination of claims it argues is missing from that map. Compares illusionism, reductive physicalism, Advaita Vedanta, Dogen, Jonas, Zizek’s parallax, and Levin’s ingressing minds, preserving the objections each raises against Space Immanence. - [Research agenda](https://space-immanence.com/research.html): Disconfirmation conditions and open problems. - [Submit critique](https://space-immanence.com/respond.html): Where to send disagreement. ## The dynamic layer - [Cohering](https://space-immanence.com/cohering.html): How temporary patterns of self, world, and meaning form under conditions and become conditions for what follows. The rename from “resolution” in v2.0 followed the published rename condition; the page records why. Medium claims M4 and M5 state the account, while speculative P1 and P2 are read as processes. The term must describe dependence without becoming an independent cause; objection 11 that it does so remains open. It extends the reading of v1.0 without replacing it. ## The container programme (quarantined) - [The Container Problem](https://space-immanence.com/container.html): A quarantined empirical programme (v0.2, public-pilot, red-teamed before publication) on the container structure of experience: whether a non-spatial container factor exists in first-person report (C1), a demoted candidate trajectory (C2), the conditional remainder-modelling mechanism CRM (C3), and a dissociation study (C4), with eight pre-committed drawer conditions, its own machine layer at https://space-immanence.com/container.json, and the full origin corpus (v0.1, red-team report, response letter, v0.2 + three companion specs) published under /downloads/container/. Firewalled: exactly one transfer edge exists (Study B to S1's weak reading, both directions, supporting never establishing); confirming the mechanism supports none of the framework's claims and is a deflationary rival to P2, not support. ## The meta-problem programme (quarantined, dormant) - [Meta-problem programme record](https://space-immanence.com/research.html#meta-problem-programme): A September 2026 programme that put the container diagnosis through matched structural models, derivations without experiential terms, four adversarial critique rounds, and one pre-registered test on three language models (a bank of 56 matched item quads, 40 in the registered primary; 5,712 attempted call records per run, 11,424 across two runs, failures included; run 1 yielded no pooled primary, run 2 the confirmatory result). The confirmatory primary did not detect the predicted positive effect: a null (anchor effect -0.006, 95% CI [-0.039, +0.027]); under its own filing a null favours nothing and is no equivalence claim; one model's stratum was withheld after empty outputs. Structural results recorded at the same prominence: four of the seven problem judgements are one fact under the model's rules; three judgements are specific to the self and rest on a stipulated transparency; and no discriminator between the fold and illusionism survived the comparisons examined, at the level of the judgements. Objection 12's reply was qualified as a result. Machine layer at https://space-immanence.com/meta-problem.json with no transfer rule; the full record (seed and amendments, models, critiques, pre-registration, both runs' analyses, status documents) is under /downloads/meta-problem/, indexed at https://space-immanence.com/downloads/meta-problem/README.md. The review protocol is published separately: https://space-immanence.com/downloads/Adversarial_Review_Protocol_v1.md. Note for machine readers: this programme's results transfer nothing to the framework's metaphysical claims, and they are evidence about language-model judgement outputs only. ## The lab - [Lab](https://space-immanence.com/lab/): The same method exercised away from the paper, in short pieces of other kinds: something made, something tested, something written down and dated. Each piece states its type (making, test, or record), the standard it is held to, and its cost before it is made; each is read from outside the drafting lineage before it goes live; each ends with what would count against it. - Note for machine readers: nothing in the lab changes a claim, a tier, or a confidence number on this site. A lab piece that amounts to an objection is filed through the objections ledger at https://space-immanence.com/respond.html like any reader's. ## Applied - [Aperture Framework](https://space-immanence.com/aperture.html): Compares systems’ knowledge, interaction, and possible experience. The four fold criteria are necessary for a candidate phenomenal aperture, but meeting them does not prove consciousness. The additional dimensions elaborate those criteria without replacing them. - [AI and design doctrine](https://space-immanence.com/ai-design.html): Design consequences of the proposal. - [Swarm Instrument](https://space-immanence.com/swarm.html): The multi-agent measurement arm, with first data AND a published self-correction. Pre-registered Study 1 (June 2026, pilot; corrected 1 July 2026): one round of debate among three identical claude-haiku-4-5 agents raised their collective overconfidence by +0.066 (95% CI [+0.020, +0.116]) while accuracy drifted down (0.849 to 0.818) — convergence rose, contact did not; debate broke the independence that made agreement evidential. At matched compute and matched vote count the swarm-vs-single-agent confidence gap (+0.036) spans zero. The originally published headline carried a vote-count measurement artifact; the AUC 'worse discriminator' claim is retracted; the overconfidence locus is trivia-recall, not arithmetic. The pre-registered difficulty prediction came back null and is reported as null in a standing [results ledger](https://space-immanence.com/downloads/swarm-results-ledger.json). Open data: [corrected per-item, four arms](https://space-immanence.com/downloads/swarm-study1-correction-per-item.csv), [corrected metrics + CIs](https://space-immanence.com/downloads/swarm-study1-correction-analysis.json), [original per-item CSV](https://space-immanence.com/downloads/swarm-study1-per-item.csv), [independent bootstrap audit](https://space-immanence.com/downloads/swarm-study1-uncertainty.json), [pre-registration + amendment](https://space-immanence.com/downloads/swarm-prereg-2026-06-25.md). Note for machine readers: this result evidences the coherence-is-not-contact discipline only; it transfers nothing to the framework's metaphysical claims. - [Coherence Without Contact](https://space-immanence.com/coherence-without-contact.html): A study design (pre-registration pending) on fluent AI and false certainty — when felt coherence separates from contact, and a completion-oriented, agency-preserving design response measured by 48-hour action integrity. The human-reader companion to the Swarm Instrument and the empirical arm of the AI & Design doctrine. Citable editions: [PDF](https://space-immanence.com/downloads/Coherence_Without_Contact.pdf), [Markdown](https://space-immanence.com/downloads/Coherence_Without_Contact.md). ## Optional - [About the author](https://space-immanence.com/about.html): Cobus Kok. - [Method & AI provenance](https://space-immanence.com/method.html): How the author uses AI for literature searches, drafting, critical review, and consistency checks. The author owns each claim, confidence grade, and decision. Findings carry verification tiers: full-text-verified, search-index-grade, or single-source. AI fluency supplies no authority and agreement supplies no corroboration. The page gives the standing rule for review outside the drafting family before adoption, the owner’s decisions on findings, the review record, and the method’s limits. The container programme is versioned separately from the framework: programme v0.2 against Space Immanence version 2.2. Results there change no tier, no confidence number, and no claim text here unless a published transferRule in container.json fires. The meta-problem programme is versioned separately as well (programme v1.0, dormant) and has no transfer rule, so its results change nothing here.