The Independence Test

The Independence Axiom, formalised by von Neumann and Morgenstern in 1944 and stress-tested by the Allais Paradox in 1953, states that a genuine preference must remain stable regardless of which irrelevant alternatives are present. Applied to vocabulary, the same principle holds: a term that has been genuinely internalised will mean the same thing whether the speaker is in a formation conversation or an arena conversation. A term that has only been surface-adopted will shift its meaning to match the dominant frame of the current audience.

Michael Polanyi's formulation — "we can know more than we can tell" — captures the residue of genuine internalisation. Tacit knowledge is the knowledge that cannot be reproduced from the Lexicon definition alone; it is the knowledge that emerges from having used the term in practice, made errors with it, and corrected those errors through feedback. Wenger and Lave's concept of Legitimate Peripheral Participation (1991) describes the mechanism: newcomers acquire vocabulary not through instruction but through practice at the edge of the community, gradually moving toward full participation as their usage becomes indistinguishable from that of experienced members.

The Independence Test has three levels. The first is Definitional: can the person state the term's meaning? This tests surface adoption and is the level at which most onboarding processes stop. The second is Contextual: can the person identify a correct and incorrect use of the term in a novel situation not previously discussed? This tests internalisation — whether the term has become a stable cognitive tool rather than a reproduced phrase. The third is Generative: can the person use the term to illuminate a situation that the Decalogy has never addressed? This tests genuine ownership — the point at which the vocabulary has become part of the person's thinking rather than a layer on top of it.

Johan's observation is structurally precise: machine intelligence will always score the maximum on a vocabulary consistency test. The reason is architectural. Token embeddings are deterministic: "Abel Progress" maps to the same vector regardless of whether the conversation is formation-oriented or arena-oriented. MI has no social cost for using terms correctly, no upbringing that instilled competing meanings, and no arena that contaminates its usage. It is, in this narrow sense, perfectly independent.

But this precision is statistical rather than experiential. MI knows the distribution of how "Abel Progress" is used across the Decalogy corpus. It does not know the referent — the lived experience of choosing between Abel and Cain paths under conditions of scarcity and social pressure. The paradox is that MI scores maximum on the first two levels of the Independence Test but fails structurally on the third. It cannot generate novel applications of the vocabulary because it has no experiential grounding from which to generate them — only a statistical model of how the vocabulary has been used before.

This is the structural gap that MI must bridge with biological intelligence. The gap is not about vocabulary consistency — MI wins that test. The gap is about experiential grounding: the capacity to use a term to illuminate a situation that has never been illuminated before. Johan's question — "is this an impossible task?" — is answered by the structure of the test itself. MI can bridge the gap at levels 1 and 2 by providing consistent, precise vocabulary to BI users who are contaminated by arena pressure. But at level 3, the bridge must run in the other direction: BI provides the experiential grounding that MI lacks, and MI provides the consistency that BI loses under arena pressure.

Johan's observation that "we created standards and the forging of humans through education and community/company language" is confirmed by three independent research traditions. Pierre Bourdieu's concept of "linguistic habitus" (1977) describes how the language of the family and school is internalised at the pre-reflective level — as a bodily disposition rather than a conscious choice. This is why arena contamination is so persistent: it was installed before the person had the conceptual tools to evaluate it. Vygotsky's Zone of Proximal Development (1934/1978) describes the mechanism: vocabulary is internalised through scaffolded social interaction, not through instruction alone. The community provides the scaffold; the individual internalises the vocabulary through practice within it.

The corporate equivalent is explicit and intentional. McKinsey's "MECE" (Mutually Exclusive, Collectively Exhaustive) is a term that only has meaning inside McKinsey's frame — it forges identity through exclusion of non-initiates. Military jargon (SITREP, FUBAR, ROE) operates the same way: the vocabulary is a membership test that simultaneously forges identity and excludes outsiders. The Decalogy Lexicon is structurally similar, but with a critical difference: its terms are defined by their referents (negentropic development, mimetic escalation, formation-oriented conversation) rather than by their exclusionary function. The test of a Decalogy term is not "do you know this word?" but "can you use this word to see something you could not see before?"

Johan's forecast — that vocabulary standardisation will accelerate as MI becomes the dominant communication medium — was confirmed by a landmark study published in Trends in Cognitive Sciences on 11 March 2026, three days before this dialogue. Sourati, Ziabari, and Dehghani synthesised evidence across linguistics, psychology, cognitive science, and computer science to show that LLMs "reflect and reinforce dominant styles while marginalising alternative voices and reasoning strategies." The mechanism is structural: LLMs are trained on dominant patterns and amplify convergence as all people increasingly rely on the same models across contexts. The paper identifies the risk as "epistemic collapse" — the flattening of the cognitive diversity that drives collective intelligence.

The scale of the effect is already measurable. As of mid-2025, users send over 2.6 billion messages per day to LLMs (Handa et al., 2025). Each of those messages is a vocabulary interaction in which the LLM's consistent, precise usage exerts pressure on the user's context-dependent usage. Over time, the LLM's vocabulary becomes the reference frame — not because it is more accurate, but because it is more consistent. This is the forging process operating at civilisational scale, without the scaffolding of a Formation Seed to preserve the experiential referent.

The Decalogy's position is not that standardisation is inherently Cain. The Benedictine Rule (529 AD) was a vocabulary standardisation that preserved negentropic development across fifteen centuries. The distinction is between standardisation that preserves the referent (Abel) and standardisation that replaces the referent with the label (Cain). When MI standardises "Abel Progress" to mean "growth that beats competitors," it has replaced the referent with an arena label. When the Formation Seed standardises "Abel Progress" to mean "development that increases the complexity of the cosmos," it has preserved the referent across contexts.

The Formation Seed solution is not to resist MI's vocabulary precision but to use it as a mirror. MI's consistent usage reveals where BI's usage has drifted — not as a judgment but as a diagnostic. The Independence Test is the instrument: when a Formation Seed member's usage of "Abel Progress" diverges from MI's usage, the divergence is a signal that arena pressure has contaminated the term. The task is not to correct the member but to ask: what arena pressure produced this drift, and what does it reveal about the member's current formation context?

The distinction between precision and standardisation is structural. Standardisation removes variation by replacing diverse referents with a single label. Precision preserves the referent while allowing variation in expression. A Formation Seed community that uses the Independence Test correctly will have members who express "Abel Progress" in five different ways — because they have five different experiential referents for the same underlying concept. MI can only express it in one way, because it has one statistical distribution. The Formation Seed's vocabulary is richer than MI's precisely because it is grounded in diverse experiential referents.

This is the answer to Johan's question about whether the MI-BI bridge is an impossible task. It is not impossible — but it requires the bridge to run in both directions simultaneously. MI provides consistency to BI users under arena pressure; BI provides experiential grounding to MI's statistical precision. The Formation Seed is the structure that makes both directions possible — the community that uses MI's mirror without becoming its reflection.