RCC: Why LLMs Still Hallucinate Even at Frontier Scale (Axioms Included)(effacermonexistence.com)
effacermonexistence.com
RCC: Why LLMs Still Hallucinate Even at Frontier Scale (Axioms Included)
http://www.effacermonexistence.com/rcc-hn-1
7 コメント
I’ve been working on something I call Recursive Collapse Constraints, or RCC.
It’s a boundary theory for any inference system that operates inside a larger manifold, including modern LLMs.
RCC is not an architecture and not a training trick. It’s a set of structural axioms that describe why hallucination, inference drift, and loss of long-horizon consistency appear even as models get larger.
Axiom 1: Partial Observability An embedded system never has access to the full internal state of the manifold it operates in.
Axiom 2: Non-central Observer The system cannot determine whether its viewpoint is central or peripheral.
Axiom 3: No Stable Global Reference Frame Internal representations drift over time because there is no fixed frame that keeps them aligned.
Axiom 4: Irreversible Collapse Each inference step collapses information in a way that cannot be fully reversed, pushing the system toward local rather than global consistency.
Several predictions follow from these axioms: • Hallucination is structurally unavoidable, not just a training deficit. • Planning failures after about 8 to 12 steps come directly from the collapse mechanism. • RAG, tools, and schemas act as temporary external reference frames, but they do not eliminate the underlying boundary. • Scaling helps, but only up to an asymptotic limit defined by RCC.
I’m curious how people here interpret these constraints. Do they match what you see in real LLM systems? And do you think limits like this are fundamental, or just a temporary artifact of current model design?
Full text here: https://www.effacermonexistence.com/axioms
RCC is not an architecture and not a training trick. It’s a set of structural axioms that describe why hallucination, inference drift, and loss of long-horizon consistency appear even as models get larger.
Axiom 1: Partial Observability An embedded system never has access to the full internal state of the manifold it operates in.
Axiom 2: Non-central Observer The system cannot determine whether its viewpoint is central or peripheral.
Axiom 3: No Stable Global Reference Frame Internal representations drift over time because there is no fixed frame that keeps them aligned.
Axiom 4: Irreversible Collapse Each inference step collapses information in a way that cannot be fully reversed, pushing the system toward local rather than global consistency.
Several predictions follow from these axioms: • Hallucination is structurally unavoidable, not just a training deficit. • Planning failures after about 8 to 12 steps come directly from the collapse mechanism. • RAG, tools, and schemas act as temporary external reference frames, but they do not eliminate the underlying boundary. • Scaling helps, but only up to an asymptotic limit defined by RCC.
I’m curious how people here interpret these constraints. Do they match what you see in real LLM systems? And do you think limits like this are fundamental, or just a temporary artifact of current model design?
Full text here: https://www.effacermonexistence.com/axioms
Modern LLMs do not fail because they are poorly engineered. They fail because they are embedded, partially-blind inference systems.
Modern LLMs fail to achieve "intelligence" --- and this is a direct result of their design and engineering --- or rather the lack thereof.
A "partial-blind inference system" has no effective sense of judgment and thus can't tell the difference between fact and fiction.
The most amazing part to me is the number of CEOs who have been convinced that a "partial-blind inference system" has the potential to replace their employees. Which likewise demonstrates a lack of judgment.
Modern LLMs fail to achieve "intelligence" --- and this is a direct result of their design and engineering --- or rather the lack thereof.
A "partial-blind inference system" has no effective sense of judgment and thus can't tell the difference between fact and fiction.
The most amazing part to me is the number of CEOs who have been convinced that a "partial-blind inference system" has the potential to replace their employees. Which likewise demonstrates a lack of judgment.
Just a quick comment on the “fact vs fiction” issue. Humans don’t reliably solve that either. For most of history, people believed the Earth was flat because every local observation they had access to pointed in that direction. Their frame of reference was simply too limited to reveal the error.
RCC isn’t claiming that LLMs are uniquely flawed. The point is that any system working with partial visibility(humans included)can’t guarantee globally correct judgments. What counts as “fact” only becomes stable when there is an external reference frame, and embedded agents don’t have access to one.
RCC just states these limits in geometric and observability terms.
RCC isn’t claiming that LLMs are uniquely flawed. The point is that any system working with partial visibility(humans included)can’t guarantee globally correct judgments. What counts as “fact” only becomes stable when there is an external reference frame, and embedded agents don’t have access to one.
RCC just states these limits in geometric and observability terms.
Humans don’t reliably solve that either.
They can and have. But they don't achieve this by using probability to guess the next word in a sentence.
Once again, judgment is an important but ill defined aspect of intelligence.
An LLM has none. Instead, it relies on probability --- which we all know can easily produce incorrect results that sound plausible.
Tempered with human judgment, LLMs can still prove useful in some strictly restrained cases but it's general purpose reliability is highly suspect --- in my judgment. And this lack of reliability counters the logic for applying them in a lot of cases.
They can and have. But they don't achieve this by using probability to guess the next word in a sentence.
Once again, judgment is an important but ill defined aspect of intelligence.
An LLM has none. Instead, it relies on probability --- which we all know can easily produce incorrect results that sound plausible.
Tempered with human judgment, LLMs can still prove useful in some strictly restrained cases but it's general purpose reliability is highly suspect --- in my judgment. And this lack of reliability counters the logic for applying them in a lot of cases.
The argument that “LLMs lack judgment because they only guess the next token probabilistically” starts from an overly simplistic model of how human judgment actually forms.
Humans also begin as probabilistic next-word predictors. Look at early language formation in infants:
“Mom → food” “Mom → poop”
This is literally a next-token model. There is no semantics, no reasoning—only repeated patterns, reinforced predictions, and gradual abstraction. As children grow, they expand the sequence window:
“Mom I’m hungry” → “Mom can you go to the store and get the ice cream I like”
This is the emergence of abstraction → generalization → specialization, the exact loop LLMs run internally.
Human cognition is biochemical; LLMs are computational. Different substrate, similar functional loop.
And “judgment” is not a mystical faculty. It can be decomposed into: 1. forming a generalized baseline, 2. comparing specific cases to that baseline, 3. updating through iteration, 4. selecting an output.
LLMs do exactly this. Pretraining forms the baseline, attention performs comparison, decoding performs selection.
If your definition of judgment is “access to a global, external truth-frame,” then humans do not possess judgment either. For most of history people believed the Earth was flat because their local frame of reference made it the most reasonable inference.
Judgment is always local for embedded agents—biological or computational.
This is precisely what RCC explains: LLM failures are not due to “probabilistic prediction,” but due to embeddedness and partial observability, the same geometric constraint that applies to humans.
The reliability issue is structural, not moral or mystical.
Humans also begin as probabilistic next-word predictors. Look at early language formation in infants:
“Mom → food” “Mom → poop”
This is literally a next-token model. There is no semantics, no reasoning—only repeated patterns, reinforced predictions, and gradual abstraction. As children grow, they expand the sequence window:
“Mom I’m hungry” → “Mom can you go to the store and get the ice cream I like”
This is the emergence of abstraction → generalization → specialization, the exact loop LLMs run internally.
Human cognition is biochemical; LLMs are computational. Different substrate, similar functional loop.
And “judgment” is not a mystical faculty. It can be decomposed into: 1. forming a generalized baseline, 2. comparing specific cases to that baseline, 3. updating through iteration, 4. selecting an output.
LLMs do exactly this. Pretraining forms the baseline, attention performs comparison, decoding performs selection.
If your definition of judgment is “access to a global, external truth-frame,” then humans do not possess judgment either. For most of history people believed the Earth was flat because their local frame of reference made it the most reasonable inference.
Judgment is always local for embedded agents—biological or computational.
This is precisely what RCC explains: LLM failures are not due to “probabilistic prediction,” but due to embeddedness and partial observability, the same geometric constraint that applies to humans.
The reliability issue is structural, not moral or mystical.
Hallucinated slop about hallucinating slop. chef's kiss
RCC can be written as a set of geometric / partial-information constraints:
A1. Internal State Inaccessibility Let Ω denote the full internal state. The observer only ever sees a projection π(Ω), with π: Ω → Ω′ and |Ω′| < |Ω|. All inference happens over Ω′, not Ω.
A2. Container Opacity Let M be the manifold containing the system. Visibility(M) = 0. Global properties like ∂M or curvature(M) are, by definition, not accessible from inside.
A3. No Global Reference Frame There is no Γ such that Γ: Ω′ → globally consistent coordinates. Inference runs in local frames φᵢ, and the transition φᵢ → φⱼ is not invertible over long distances.
A4. Forced Local Optimization At each step t, the system must produce x₍ₜ₊₁₎ = argmin L_local(φₜ, π(Ω)), even when ∂information/∂M = 0.
From these, the boundary condition is pretty direct:
No embedded inference system can maintain stable, non-drifting long-horizon reasoning when ∂Ω > 0, ∂M > 0, and no Γ exists.
This is the sense in which RCC treats hallucination, drift, and multi-step collapse as structural outcomes rather than training failures.
If anyone wants the longer derivation or the empirical predictions (e.g., collapse curves tied to effective curvature), I’m happy to share.