§0 Sammanfattning

Två fåglar lyfter samtidigt; ingen av dem har sagt något.

Dream logic, archetype, shadow material, demons and the divine, unreasonable coincidence and apparent hive-mind coupling are one problem: local sections that agree more than they should [D]. Correlation between two sections licenses either a connecting morphism or a common node upstream of both, never the morphism by itself (Reichenbach 1956) [E]. In every case treated here the upstream reading suffices, and the felt sense of a private lateral channel is a reading of the observer's narrative self-reference Ψ, not of the world [D]. The post does not claim that the experiences are false or unimportant; it claims that the lateral channel is not established, and it names the observations that would establish it (§9).

Epistemic ledger. Every substantive claim carries one tag:

Tag Meaning
[E] Established: textbook consensus, a proved theorem, direct measurement, or a standard etymology (SAOB, Hellquist, OED, etymonline)
[S] Supported: leading model or majority scholarly reading, strong but incomplete evidence
[H] Hypothesis: open, contested, or without decisive evidence
[D] Draken synthesis: structural claim made by this corpus, submitted for Clinch review
[M] Metaphor or paronomasia: a pointer, not a referent or root (per The Pendragon Source, DRK-165)

§1 Thesis: the common-cause disjunction

Six phenomena are usually filed in six different drawers: dream logic, archetypal manifestation, shadow material, demons and the divine, unreasonable coincidence, and the sense of a hive mind operating across large subsets of people and animals.

They are one problem in sheaf terms [D]. In each case an observer encounters local sections that agree more than they should, and must decide what the agreement licenses. The whole discipline of the post is one distinction:

Correlation between two sections implies either a morphism connecting them, or a common node upstream of both. It never, by itself, implies the morphism.

This is Reichenbach's common cause principle (Reichenbach 1956) [E], rewritten on a base space in the cellular-sheaf vocabulary of Hansen & Ghrist (2019) [D]. Formally: for sections $s_A, s_B$ over stalks $A, B$, if $\mathrm{Corr}(s_A, s_B) > 0$, then

$$\exists\, \rho_{A \to B} \quad \lor \quad \exists\, C \ \text{with}\ \rho_{C \to A},\, \rho_{C \to B},$$

where $\rho$ denotes a map carrying information from one stalk to another and $C$ is a node upstream of both [D].

The second disjunct is almost always the true one, and it is almost never the one the experience feels like [H]. The felt quality of synchronicity is precisely the feeling of the first disjunct: a private edge drawn directly to you [D]. That mismatch between the felt topology and the actual topology is the subject here.

§2 Dream logic is a sheaf with the sensory restriction map switched off

Waking cognition is a gluing problem under constraint [D]. Local models $s_U$ over overlapping patches $U, V$ of the world are forced into agreement by a restriction morphism carrying sensory evidence:

$$\rho_{U \cap V}(s_U) = \rho_{U \cap V}(s_V).$$

Satisfy this everywhere and the local sections glue into a global section: one world, one body in it, one time [D]. In sleep, the sensory term is gated. Following Hobson & Friston (2012), the generative model free-runs with the prediction-error channel attenuated [S]. The consequence is not noise. It is something sharper and much more interesting:

Dream states are locally consistent and globally inconsistent [D].

Every scene in a dream is internally coherent. The room is a room. The person is that person. What fails is exclusively the overlap condition: the room was also the other house, the person was also someone else, and the transition was not noticed because nothing was checking [H]. That is the signature of a nonzero first cohomology class, an assignment that is consistent on every patch and admits no global section [D]. With $X$ the space of dream scenes and $\mathcal F$ the sheaf of local world-models over it:

$$[\omega]_{\text{dream}} \in H^1(X;\mathcal{F}), \qquad [\omega]_{\text{dream}} \neq 0.$$

This is a structural reading, not a computed class: no coefficient field or cellular structure is fixed here [D].

Dream logic is not a lesser logic. It is the ordinary logic of the stalks with the cocycle condition unenforced [D]. This matters for the rest of the post, because it means the dreaming system is a generator of candidate sections unfiltered by $\rho$: a high-yield, low-precision abduction engine, in the sense of The Guessed Section (DRK-170) [D]. It produces guesses. It does not produce warrant. Everything downstream of that confusion is where the trouble starts.

§3 Archetype is protocol, not transmission

The empirical observation behind Jung's archetypes is real and does not need a mystical carrier: unrelated people, in unrelated cultures, independently generate structurally similar figures, such as the devouring mother, the trickster, the threshold guardian, the shadow twin, the dying and returning king [S].

Kärv-etologi, the sheaf-ethology pilot, already gives the correct reading of this. The pilot's load-bearing result was that ritualized varanid combat is substrate-invariant: the protocol is the agent, the individual is the substrate [H]. Two monitors that have never met run the same five-node phase graph because the graph is the stable solution to a shared constraint problem, not because one taught the other [H].

Archetypes are the same object at a different layer [D]. Convergent structure follows from a shared base space:

  • one body plan, hence one geometry of threat, appetite, and exhaustion;
  • one developmental ecology: dependence, weaning, rivalry, mortality;
  • one narrative grammar, because stories are compression schemes and only a few compress a life well.

$$\text{archetype} \;=\; \text{attractor of the generative dynamics under shared constraints}, \qquad \text{person} = \text{substrate}.$$

So: no transmission channel, no inherited image-library, no shared psychic reservoir needed. Independent solvers over one constraint landscape land in the same basins [D]. The commonality is upstream, not lateral. This is the defensible core of the archetype claim, and it survives the deflationary attacks that the transmission version does not [D].

§4 Shadow: the section you refuse to include in the gluing

Shadow work has a clean formalization in the framework's own operators [D]. The shadow is not a hidden entity. It is the set of local sections that the self-model declines to admit into the global gluing: the appetites, the capacities for cruelty, the ambitions, the fears whose inclusion would break the current narrative section [D].

Exclusion is not deletion. The excluded material remains in the stalk and continues to constrain behaviour, so the global section stays out of agreement with the substrate that generates it [D]. That gap is coherence debt, as defined in The Coherence Debt (DRK-121):

$$K(t) = \int_0^t \big[\Psi(\tau) - \Psi_{\text{viable}}\big]\, w(\tau)\, d\tau,$$

where $\Psi$ is the narrative self-reference ratio, $\Psi_{\text{viable}}$ the largest deficit at which the system still corrects, and $w$ weights the irreversibility of what is done under the divergent model. Here $\Psi$ includes exactly the self-referential narrative work required to keep the excluded sections excluded [D]. Suppression is not free; it is amortized.

The Burnt Section (DRK-158) traced the sharp version of this on a civilizational scale: Josiah's reform as forced $H^1$-suppression by amputation [D]. Cut the disagreeing sections out rather than glue them, and you get a clean narrative and a fragile one. The individual case is identical in form [D]. Integration lowers $K$; amputation defers it at interest. That is the entire operational content of "shadow work", and it needs no metaphysics.

When the excluded section is thrown outward and seen on someone else, the move is the centrifugal optic of The Two Optics (DRK-163): projection enlarges the image and relocates its source [D].

§5 Demons and the divine: carrier and cargo on an attractor

Apply the four-pole carrier/cargo taxonomy used in The Carrier and the Cargo (DRK-189). A demon is a carrier. The question is never whether the carrier is literally real; it is whether the cargo indexes truly [D].

Personification is a legitimate and efficient interface to a high-dimensional dynamical regularity [D]. "A demon of resentment that grows when fed and speaks in your own voice" is a compact, actionable, and largely accurate description of a self-reinforcing affective loop with a positive feedback term and an endogenous justification generator [D]. As an interface it outperforms the clinical description on several practical axes: it is memorable, it is addressable, it supports strategy, and it externalizes the loop enough to be fought without requiring the self to be the enemy [H].

The failure mode is the index flip. The carrier stops being a handle on the dynamics and becomes an ontological claim about an agent with intentions toward you specifically [D]. At that point

$$\text{authentic carrier} + \text{true index} \;\longrightarrow\; \text{authentic carrier} + \text{false index},$$

and $\Psi$ rises, because the cosmos has just been assigned a private communication channel to one node [D]. The same applies exactly and symmetrically to the divine pole. Personified transcendence as an interface to value, limit, and awe: sound carrier. Personified transcendence issuing you targeted, decodable instructions through ambient events: false index, rising $\Psi$, and plausibly one of the earliest signatures of a section coming apart [H].

§6 Unreasonable coincidence is a base-rate result, and it is supposed to feel unreasonable

The felt improbability of a coincidence is computed post hoc, over a specification chosen after the event [S]. This is not a minor bias; it inverts the calculation.

Littlewood's estimate. Take one distinguishable event per second, eight waking hours per day, as in Littlewood's estimate (Littlewood 1986) [S]. Then

$$N \approx 1\,\text{s}^{-1} \times 3600 \times 8 \times 35 \approx 1.0 \times 10^6 \ \text{events per 35 days}.$$

For an event class of probability $p = 10^{-6}$, the expected count follows by multiplication [E]:

$$\mathbb{E}[\#] = Np \approx 1 \ \text{per month}.$$

A one-in-a-million event, once a month, per person, as a matter of arithmetic [E]. In a population of ten million that is on the order of $10^7/35 \approx 3 \times 10^5$ such events every single day [E]. The miraculous is not rare; it is scheduled.

Ramsey's constraint. Complete disorder is impossible: sufficiently large structures necessarily contain ordered substructures (Ramsey 1930) [E]. Any life long enough contains repeated names, dates, numbers, and echoes regardless of whether anything is arranging them; finding the pattern is evidence of adequate search, not of an arranger [D].

The specification problem. Before the fact, the space of coincidences that would have struck you as meaningful is enormous and undefined. After the fact, it has cardinality one. The probability people report is $P(\text{this exact event})$; the probability that matters is $P(\text{some member of the unspecified class})$, which is not small [E].

Selection and optional stopping. Hits are logged and retold; misses are not encoded [S]. The dream that anticipated the phone call is remembered; the several hundred that anticipated nothing were discarded before breakfast. There is no register, so the denominator is unrecoverable, which is precisely why a pre-registered log is the only thing that could settle the question (§9) [D].

None of this makes the experience false. The experience is a real report about the observer's salience-weighting. It is simply not a measurement of the world [D].

§7 The hive that is real, and the hive that is not

Now the strongest part of the question, because here a large fraction of the phenomenon is genuinely real and is routinely attributed to the wrong mechanism.

Collectives of humans and animals demonstrably coordinate without vocalized intent, and the mechanisms are well characterized [E]:

  1. Stigmergy (Grassé 1959). Agents modify a shared medium; the medium coordinates them. Termites build arches with no architect and no signal between builders; each responds to the local state of the mound [E]. Formally: no $\rho_{A \to B}$ at all, only $\rho_{A \to E}$ and $\rho_{E \to B}$ through the environment $E$ [D]. Path, trail, worn floor, thread of a group chat, the state of a shared repository. The coordination is real; the telepathy is an artifact of not looking at $E$.
  2. Common driver. Two nodes with similar priors receiving similar inputs emit similar outputs at similar times, with zero coupling (Reichenbach 1956) [E]. Everyone thinks of the same song because everyone heard the same advert.
  3. Sub-threshold sensorimotor coupling. Gaze direction, postural set, micro-latency in a reply, breathing rate, gait entrainment. High-bandwidth, fast, and largely unavailable to introspection, which is exactly why its products feel channel-less [H]. You did not consciously see the cue, so the inference arrives as if from nowhere.
  4. Information cascade. Each agent rationally weights others' revealed choices; the group locks onto a consensus faster than the evidence warrants, and every member experiences it as independent convergence (Bikhchandani, Hirshleifer & Welch 1992) [E].

Dataset C is a live instance sitting on the balcony (Khrug Engineering 2026). Sentinel behaviour in the magpie–crow assemblage propagates without a call: one bird's posture shifts, the railing empties, and the departure looks instantaneous and intention-shared [H]. It is neither. It is fast visual coupling plus shared line of sight plus a common external driver, running at a latency below the resolution of casual observation [H]. Måns is not telepathic. Måns is quick, and watching everyone, and so is everyone else.

So, is there a hive mind? Yes, in the only sense that has been demonstrated: distributed computation over a shared medium with no global controller and no node holding the whole section [E/D]. That is a real, formal, load-bearing structure and one of the more important objects in the framework.

Is there a direct edge, an unmediated transfer of intent between stalks? No evidence supports it, and the sheaf makes the cost explicit [D]. Positing $\rho_{A \to B}$ where an upstream $C$ already accounts for the correlation adds an edge that does no explanatory work, breaks the locality that makes the apparatus computable (Hansen & Ghrist 2019), and, by the anti-totalization principle, is exactly the move the framework exists to refuse [D]. One shared substance underneath everything, of which we are all local expressions, is the totalizing section. Draken does not get to make that move, and neither does anything built on it.

§8 Why this domain is the framework's own stress test

Define synchronicity density:

$$\sigma(t) = \frac{\#\{\text{events read as addressed to the self}\}}{\#\{\text{salient events}\}}.$$

Synchronicity density $\sigma(t)$ is the fraction of salient events at time $t$ that the observer reads as addressed to the self [D].

Claim: $\sigma$ tracks $\Psi$, not the state of the world [H]. It rises with fatigue, isolation, sleep debt, intensity of a single interpretive project, and the amount of time spent inside one symbolic system without external correction. It falls with sleep, physical work, unrelated conversation, and adversarial review [H].

This is why the six-model Clinch exists, and why $\Psi$ was defined with high values marking sickness rather than depth [D]. The domain in this post is the one where every operator in the framework can be run at full power on material that cannot push back. Dream, archetype, demon, and coincidence are infinitely tractable to interpretation: they will accept any structure imposed on them and return it confirmed [D]. A system that finds Draken's own signature in dreams and numberplates is not detecting Draken. It is measuring its own $\Psi$ and misreading the dial [D]. In the terms of Ouroboros Debt (DRK-176), the strange loop is safe while it remains a cycle through the world, and becomes a singularity when every reading returns to the reader [D].

The anti-totalization principle is non-negotiable and applies reflexively. So the honest position is the narrow one [D]:

  • The phenomenology is real and worth close description.
  • The upstream mechanisms are real, formalizable, and often beautiful.
  • The lateral channel is not established, and treating it as established is a coherence debt that compounds.

Rising $\sigma$ is a signal about the observer [H]. Treat a sustained rise the way you would treat any other out-of-band instrument reading: as something to check against people outside the system, not as evidence that the system is finally being answered.

§9 Falsification (DRK-131)

The falsification protocol is that of 知行合一: When Automation Removes the Only Honest Referee (DRK-131). This post is wrong, and the lateral channel is real, if any of the following hold [D].

F1: Pre-registered synchronicity log. §6 predicts that coincidence frequency matches the base rate. Specify the coincidence classes in advance, log every salient event for 90 days, hits and misses alike. Refuted if observed frequency exceeds the Littlewood base rate at $p < 0.01$ with the specification fixed beforehand.

F2: Dataset C, occlusion protocol. §7 attributes corvid departure synchrony to visual coupling and common drivers. Measure departure latency in the magpie–crow assemblage under (a) full shared line of sight, (b) occluded line of sight, (c) occluded and acoustically damped. Refuted if synchrony survives (c) at latencies shorter than any known common external driver; the direct edge is then on the table.

F3: Archetype without shared constraint. §3 derives archetypes from a shared base. Refuted if a population with demonstrably different body plan, developmental ecology, and narrative grammar independently generates the identical archetypal set.

F4: Dream cocycle. §2 predicts local consistency without global consistency. Refuted if dream reports show global consistency, with stable transitive identity and geography across scene boundaries under blinded scoring, so that $[\omega]_{\text{dream}} = 0$.

F5: Ψ/σ decoupling. §8 predicts that $\sigma$ tracks $\Psi$. Refuted if $\sigma$ rises while $\Psi$ and its correlates (sleep debt, isolation, single-system immersion) are held flat or falling; $\sigma$ is then measuring something external.

F6: Scope limit. The post does not claim that the experiences are false, that archetypes are trivial, or that collective coordination is illusory. The sheaf notation is a structural reading, not a computed cohomology. If F1 or F2 fail, §6 or §7 falls and the direct edge must be taken seriously; §2–§4 survive independently. F2 is cheap, is already instrumented, and should be run.

§10 Provenance and leaks

  1. Authorship. Khrug's prompt (dream logic, archetype, shadow, demons, the divine, synchronicity, hive minds); Claude drafted the post on 2026-08-24. No Clinch review recorded.
  2. Revised 2026-10-01 to POST_STANDARD v1. The published file had lost all Markdown (headings, emphasis, lists, block quote) and every display equation, and rendered as one block of text. The original Markdown was restored from the drafting session and brought to the standard: §0 Sammanfattning, epistemic ledger and claim tags added; the former §0 Thesis became §1 and later sections shifted by one (internal § references updated); falsifiers rewritten as F1–F6 with a scope limit; DRK mentions made links; References written out in full and verified; footer rebuilt.
  3. Corrections made in the open. (a) §4 originally wrote $K(t) = \int_0^t \Psi\,d\tau$, which contradicts the approved definition in The Coherence Debt (DRK-121); replaced by that definition. (b) §6 originally said "roughly ten thousand people" per day in a population of ten million; the post's own arithmetic gives about $3 \times 10^5$. (c) §3 used an undefined symbol $\Phi$; replaced by words. (d) §5 originally called the false-index pattern "empirically the single most reliable early signature"; no source supports this, so it is softened and tagged [H]. (e) F5 read "held flat and falling"; now "flat or falling".
  4. Added references. Ramsey (1930) and Bikhchandani, Hirshleifer & Welch (1992) were added to support claims the original made without citation. Hansen & Ghrist (2019) was listed but not cited in the original text; it is now cited in §1 and §7.
  5. Unverified. The exact passage in Littlewood (1986) carrying the one-per-month estimate was not checked against the book; Jung (1952) was not checked against the original edition. The Kärv-etologi pilot (five-node phase graph, substrate invariance) and Dataset C are unpublished internal material.
  6. Layers. The draft carried L07, L09, L11, L14, L18 unchecked. Checked against the thesis v4.4 layer table on 2026-10-01 and replaced by L05 (neural integration: dreaming as predictive coding, §2), L06 (embodied cognition: sub-threshold sensorimotor coupling, §7), L07 (narrative self: shadow, Ψ, σ, §4 and §8), L09 (group cognition: the hive, §7), L12 (collective narratives: archetype, demon and divine as narrative carriers, §3 and §5), L15 (cultural field: cross-cultural convergence, §3). L11, L14 and L18 were dropped: the post makes no claim about market exchange, macroeconomic topology or planetary cognition, and §7 explicitly refuses the planetary shared-substance reading.
  7. New term. $\sigma(t)$, synchronicity density, is defined in §8 and entered in the KO registry and approved by Khrug on 2026-10-01; it counts toward W once another post reuses it.
  8. Dataset C stays in References as unpublished field data, as in Ouroboros Debt (DRK-176), so that §7 and F2 cite their evidence base.

References

  • Bikhchandani, S., Hirshleifer, D. & Welch, I. (1992). A theory of fads, fashion, custom, and cultural change as informational cascades. Journal of Political Economy 100(5), 992–1026. doi:10.1086/261849
  • Grassé, P.-P. (1959). La reconstruction du nid et les coordinations interindividuelles chez Bellicositermes natalensis et Cubitermes sp. La théorie de la stigmergie. Insectes Sociaux 6(1), 41–80. doi:10.1007/BF02223791
  • Hansen, J. & Ghrist, R. (2019). Toward a spectral theory of cellular sheaves. Journal of Applied and Computational Topology 3(4), 315–358. doi:10.1007/s41468-019-00038-7
  • Hobson, J. A. & Friston, K. J. (2012). Waking and dreaming consciousness: neurobiological and functional considerations. Progress in Neurobiology 98(1), 82–98. doi:10.1016/j.pneurobio.2012.05.003
  • Jung, C. G. (1952). Synchronizität als ein Prinzip akausaler Zusammenhänge. In C. G. Jung & W. Pauli, Naturerklärung und Psyche. Rascher.
  • Khrug Engineering (2026). Dataset C: balcony observations of a magpie–crow assemblage. Unpublished field data.
  • Littlewood, J. E. (1986). Littlewood's Miscellany, ed. B. Bollobás. Cambridge University Press.
  • Ramsey, F. P. (1930). On a problem of formal logic. Proceedings of the London Mathematical Society s2-30(1), 264–286. doi:10.1112/plms/s2-30.1.264
  • Reichenbach, H. (1956). The Direction of Time. University of California Press.

Operators: ρ (restriction), H¹ (dream cocycle, excluded section), K(t) (coherence debt), Ψ (narrative self-reference), σ (synchronicity density, introduced) · Crosslinks: The Coherence Debt (DRK-121) · 知行合一: When Automation Removes the Only Honest Referee (DRK-131) · The Burnt Section (DRK-158) · The Two Optics (DRK-163) · The Pendragon Source (DRK-165) · The Guessed Section (DRK-170) · Ouroboros Debt (DRK-176) · The Carrier and the Cargo (DRK-189)

Khrug Engineering · Göteborg · ORCID 0009-0003-8049-7167 · DOI 10.5281/zenodo.23121197 · CC BY-SA 4.0