Exploration: Info-Comp Core Revision — Tracing the History and the Evaluator Question

Status: Exploration. Reviewing the original info-comp core, comparing to information theory, and determining whether the evaluator distinction should have been caught.


1. The two primitive sets and their origins

1.1 Info-comp core (from v1 analysis)

Primitives: {St (State), Str (Structure), Tr (Transformation), Me (Measure), Co (Constraint), Pa (Parameter)}

Derived from: Integrating IT, CC, DS, NT — finding what information theory, computational complexity, dynamical systems, and network theory all share.

How it was derived: Each primitive abstracts across four domains:

Core triad: {St, Tr, Me} — "what is a stateful measurable dynamic system?"

1.2 Information theory (from v1 analysis, revised in v2)

v1 primitives: {Σ, P, C, K, μ, Γ} — Symbol, Distribution, Channel, Code, Measure, Constraint

v2 revised primitives: {Σ, P, C, Ke, Kd, μ, Γ} — K split into Encoder (Ke) + Decoder (Kd)

Derived from: Shannon's mathematical theory of communication and its extensions.

1.3 They are NOT the same domain

The info-comp core is the ABSTRACTION of four domains (IT, CC, DS, NT). Information theory IS one of those four domains. The relationship is:

Info-comp core {St, Str, Tr, Me, Co, Pa}
  ↓ configuration
  Information Theory {Σ, P, C, Ke, Kd, μ, Γ}  (= IT at specific settings)
  Computational Complexity {Π, I, A, M, R, Cl}
  Dynamical Systems {Ω, x, φ, t, λ, I}
  Network Theory {V, E, T, C, W, F}

The info-comp core is at a HIGHER abstraction level than information theory. Each IT primitive is a SPECIFIC CONFIGURATION of the info-comp core meta-primitives:

The previous session's analysis grabbed information theory and treated it AS the info-comp core. This was an error — they're at different abstraction levels.


2. The key structural question: Tr bundles too much

2.1 What Tr subsumes

The info-comp core's Transformation (Tr) primitive absorbs ALL of these across the four domains:

DomainWhat Tr absorbsCharacter
ITC (channel) — stochastic transmissionTRANSFER — moving states through a medium
ITK (code) — encoder/decoderEVALUATION — interpreting/translating states
CCA (algorithm) — computational procedureEVALUATION — deterministic computation
DSφ (flow) — dynamical evolutionTRANSFER — states evolving over time
NTE (edges) — connections between nodesTRANSFER — paths through topology
NTF (flow) — quantities moving through networkTRANSFER — directed movement

2.2 Transfer vs evaluation

Looking at what Tr absorbs, there's a clear split:

Transfer operations: C (channel), φ (flow), E (edges), F (flow) — states MOVING through media or topology. The medium may introduce noise but the operation is about MOVEMENT.

Evaluation operations: K (code — especially Kd decoder), A (algorithm) — states being INTERPRETED or COMPUTED. The operation is about MEANING or FUNCTION, not movement.

This is the SAME split that biology and entity system already make:

2.3 Would we have caught this from the original derivation?

The methodology.md §5.2 already noted:

Both substrates split the Transformation (Tr) role into transfer + evaluation.

So the structural evidence was VISIBLE at the time of the original analysis. Both biology and entity system explicitly separate transfer from evaluation as independent primitives. The info-comp core bundled them back together because:

  1. The info-comp core was derived from IT, CC, DS, NT — where the split is less prominent
  2. In IT: C (channel) and K (code) are both classified as "transformation" operations
  3. In DS: there's no clear evaluator — the dynamics IS the transformation (φ)
  4. In CC: the algorithm (A) is both transfer and evaluation bundled together
  5. In NT: edges and flows are transfers; no clear evaluation

The split is prominent in INFORMATION-PROCESSING substrates (biology, entity system, cognition) but less prominent in the original four info-comp domains (IT, CC, DS, NT).

2.4 Why the cognitive substrate forced the issue

The cognitive substrate analysis revealed that the evaluator can be SPLIT in a single system — formal mode (Kd3-4) vs linguistic mode (Kd1-2). This made the evaluator distinction URGENT because:

  1. You can have the same encoding (language) with different evaluation modes
  2. The reliability of the whole system depends on which evaluator mode is active
  3. Shannon's framework can't express this because it assumes deterministic decoders

Without cognitive substrates in the picture, biology and entity system both have deterministic evaluators (Kd4) — the distinction exists but doesn't cause problems. It's like a dimension that's always at the same value — you don't notice it until something varies along it.

The cognitive substrate VARIES along the evaluator dimension (Kd1-4 depending on content type). That's what made the bundled K → separate Ke+Kd revision necessary.


3. Does the info-comp core need the same revision?

3.1 The case for splitting Tr

The info-comp core's Tr bundles transfer and evaluation. Both biology and entity system split them. Cognitive substrates make the split critical. The split should propagate UP to the info-comp core level because:

  1. The info-comp core is supposed to capture what ALL information-processing systems share
  2. ALL information-processing substrates (biology, entity system, cognition) distinguish transfer from evaluation
  3. The distinction has structural consequences (evaluator determinism determines system reliability)

If the info-comp core doesn't split Tr, it can't express the evaluator distinction — and that distinction is the most important structural variable we've found.

3.2 The proposed split

Transformation (Tr) splits into:

Transfer (Tf) — How states MOVE through media or topology. Channels, flows, transmission, transcription, emit. The medium may introduce noise but the operation is about movement/propagation.

Subsumes:

Evaluation (Ev) — How states are INTERPRETED, decoded, or computed. The operation that translates encoding into function or meaning.

Subsumes:

3.3 What about DS and NT evaluators?

Dynamical systems and network theory don't have obvious evaluators:

So DS and NT have IMPLICITLY deterministic evaluators — the dynamics/topology IS evaluation, and it's always deterministic. The evaluator is only a separate, visible concern in domains where it can be NON-deterministic (cognition, culture, AI).

This means: Ev can be at level 0 (absent/degenerate) for DS and NT. It's still a valid primitive — it just happens to be at a low or ambient level in those domains. This is like how the entity system's P (peer) is at P0 for single-peer implementations — the primitive exists, it's just not active.

3.4 The revised info-comp core

#PrimitiveWhat it isSubsumes
1State (St)What exists — configurations, symbols, problems, nodesΣ, Π, Ω, V
2Structure (Str)How states relate — distributions, topology, metricsP, T(NT), Ω-structure
3Transfer (Tf)How states move — channels, flows, transmissionC, φ, E, F
4Evaluation (Ev)How states are interpreted — decoding, computation, dispatchK(→Ke+Kd), A, ribosome, dispatch
5Measure (Me)How quantities are computed — entropy, complexity, invariantsμ, R, I, centrality
6Constraint (Co)What bounds apply — capacity, complexity classes, conservationΓ, R-bounds, conservation
7Parameter (Pa)What's externally varied — noise, input size, control parametersnoise, n, λ, W

7 primitives. Matching the information theory revision (6→7 by splitting K) and the SSA revision (6→7 by separating Vr from En).

3.5 Core triad — revised

Previously {St, Tr, Me}. With Tr split:

Candidate: {St, Tf, Ev} — "What is a dynamic information system?" States that are transferred and evaluated.

This works. And it parallels:

Alternative: {St, Ev, Me} — "What is a measurable evaluated system?" States that are interpreted and measured. This would keep closer to the original {St, Tr, Me} by replacing Tr with its more important half (Ev).

The better test: which triad defines "what the info-comp core IS?" The answer is probably {St, Tf, Ev} — because the TRANSFER+EVALUATION chain is the core operation of information processing. Measure (Me) is important but it's about QUANTIFYING the system, not about the system itself.

3.6 Mapping to information theory (checking consistency)

Info-comp core (7)Information theory (7)Quality
StΣ (Symbol)Direct — states ARE symbols
StrP (Distribution)Direct — structure IS statistical organization
TfC (Channel)Direct — transfer IS channel transmission
EvKe+Kd (Encoder+Decoder)Good — evaluation IS encoding+decoding. The IT level further splits Ev into encoder and decoder.
Meμ (Measure)Direct — measure IS entropy/information quantities
CoΓ (Constraint)Direct — constraints ARE bounds and rates
Pa— (no IT counterpart)Weak — parameters don't have a clear IT primitive

The mapping is clean EXCEPT for Pa. Information theory doesn't have an explicit Parameter primitive — noise levels, code parameters, etc. are SETTINGS of other primitives (C has a noise parameter, K has code parameters). This raises the question: is Pa genuinely irreducible at the info-comp core level, or is it a property of other primitives?

3.7 Pa (Parameter) under scrutiny

Pa subsumes: noise level (IT), input size (CC), bifurcation parameter λ (DS), edge weights (NT).

These are EXTERNAL INPUTS that change system behavior without changing system structure. But are they a separate PRIMITIVE, or are they partial-level settings of other primitives?

These all look like PROPERTIES of other primitives rather than independent primitives. Pa might be reducible to partial-level parameters within St, Str, Tf, etc.

Counter-argument: Parameters are what you VARY in experiments. They're the independent variables. They have their own structure (parameter space topology, bifurcation diagrams). They're not the same as partial levels — a partial level is about HOW MUCH of a primitive is present, while a parameter is about WHICH CONFIGURATION of a primitive is active.

This is genuinely unclear. Pa might survive or might be absorbed. Let me flag it and move on.


4. Checking the mapping to biology and entity system

4.1 Biology mapping (revised info-comp core)

Info-comp (7)BiologyQuality
StG (Genome)Good — genetic state
StrReg (Regulation)Moderate — regulatory structure organizes genome expression
TfT (Transcription)Good — RNA transfer
EvR (Ribosome)Direct — deterministic evaluation (Kd4)
Me— (implicit)Weak — biology doesn't have an explicit "measure" primitive
CoMem (Membrane)Moderate — membrane as boundary/constraint
Pa— (implicit)Weak — environmental parameters, but not explicit

Issues: Me and Pa don't have clean biology counterparts. This suggests either:

  1. Me and Pa are NOT universal across all information-processing domains (they're specific to the mathematical domains IT/CC/DS/NT)
  2. Biology DOES have measure and parameter equivalents but we haven't identified them

Option 1 seems more likely — biology's primitives were identified through biological literature, not through info-comp mapping. Biology has 6 primitives {G, T, R, P, Reg, Mem} that cover its structure. Not everything in the info-comp core needs a biology counterpart — the info-comp core abstracts a DIFFERENT set of domains.

4.2 Entity system mapping (revised info-comp core)

Info-comp (7)Entity systemQuality
StE (Entity)Direct — typed data as state
StrT (Tree)Good — tree namespace as structure
TfM (Emit)Good — observable mutation as transfer
EvX (Execution)Direct — handler dispatch as evaluation (Kd4)
MeI (Identity)Moderate — content hash as measure of identity
Co— (implicit)Weak — constraints are designed, not explicit primitive
PaP (Peer)Moderate — peer as external context/parameter

Better mapping than biology but still imperfect. I → Me is a stretch (identity isn't really "measure" in the info-comp sense). P → Pa is a stretch (peer is a full primitive with capabilities, not just a parameter).

4.3 What the imperfect mappings tell us

The info-comp core doesn't perfectly abstract biology or entity system. This makes sense:

  1. The info-comp core was derived from IT, CC, DS, NT — MATHEMATICAL domains
  2. Biology and entity system are PHYSICAL/COMPUTATIONAL substrates
  3. The SSA (Situated Substrate Architecture) was derived to abstract biology + entity system + cognition
  4. The info-comp core and SSA are COMPLEMENTARY abstractions, not the same thing

The info-comp core abstracts the INTERNAL INFORMATION-PROCESSING structure. The SSA abstracts the EXTERNAL SITUATED ARCHITECTURE. They meet at the substrate level but from different angles:

The Ev → Vr (evaluator) mapping IS clean across both frameworks. This is the most important mapping — the evaluator/decoder distinction shows up in BOTH abstractions.


5. What needs to happen

5.1 The info-comp core should split Tr into Tf + Ev

The case is strong:

5.2 Pa needs scrutiny

Parameter might be reducible to partial-level settings of other primitives. Needs the 3/3b iteration test — can we absorb Pa without losing analytical capability?

5.3 The info-comp core is a different domain from information theory

They should not be confused. The info-comp core abstracts IT + CC + DS + NT. Information theory IS one of the four. The relationship is configuration: IT is info-comp core at specific settings.

5.4 The info-comp core and SSA are complementary

They abstract the same substrates from different angles. The key shared finding: the evaluator distinction (Ev in info-comp, Vr in SSA, Kd in IT). This is the structural invariant that shows up at every abstraction level.

5.5 Full 12-step analysis needed

The revised info-comp core at 7 (or 6 if Pa absorbs) needs the full treatment: dependencies, pairs, sublattice, compositions, manifestation landscape. The Pa question specifically needs the 3/3b iteration to resolve.


6. Summary

The info-comp core {St, Str, Tr, Me, Co, Pa} and information theory {Σ, P, C, K, μ, Γ} are DIFFERENT DOMAINS at different abstraction levels. The previous session's analysis confused them.

The info-comp core's Transformation (Tr) bundles transfer and evaluation — the same bundling that information theory's Code (K) had before it was split into Ke + Kd. Both biology and entity system already split this distinction as separate primitives (T/R and M/X respectively).

The evaluator distinction COULD have been caught from the original substrate mappings (both biology and entity system split Tr) but WASN'T because:

  1. The info-comp core was derived from IT, CC, DS, NT where the split is less visible
  2. In DS and NT, the evaluator is implicitly deterministic (ambient)
  3. The cognitive substrate analysis made it urgent by varying along the evaluator dimension (formal vs linguistic mode)

The proposed revision: Tr → Tf (Transfer) + Ev (Evaluation), giving 7 info-comp core primitives. Pa (Parameter) may be reducible and needs 3/3b testing. A full 12-step analysis of the revised info-comp core is needed.


7. The abstract information substrate — a different domain

7.1 The distinction

There are THREE different analytical objects that have been conflated:

  1. Information theory {Σ, P, C, Ke, Kd, μ, Γ} — a specific mathematical domain. One of IT/CC/DS/NT.

  2. Info-comp core {St, Str, Tr(→Tf+Ev), Me, Co, Pa} — the abstraction of IT/CC/DS/NT. What all four mathematical info-comp domains share.

  3. Abstract information substrate — the abstraction of biology, entity system, cognition. What all three CONCRETE information substrates share. This is what needs to be derived.

The previous session's analysis used information theory (#1) in the role of the abstract substrate (#3). This was an error — they're at different abstraction levels and abstract different domains.

The abstract information substrate should be derived the same way the abstract surface was derived — by looking at 3+ concrete instances at the same type-position and finding what they share.

7.2 Derivation from three concrete substrates

The three analyzed information substrates:

Primitive roleBiologyEntity systemCognition
Encoding coreG (genome)E (entity) + I (identity)Rp (representation)
Encoding structureT (transcription)T (tree)Ct (categorization)
EvaluatorR (ribosome)X (execution/dispatch)Sy (symbolization)
DirectionReg (regulation)M (emit — triggers evaluation)Ev (evaluation — what matters)
OutputP (protein — functional product)M (emit — observable change)As (association — connections)
BoundaryMem (membrane)P (peer — capability boundary)Sq (sequence — temporal boundary)

Six abstract roles recur across all three substrates. Each concrete substrate fills them with different primitives, but the ROLES are consistent.

7.3 The abstract information substrate primitives

#PrimitiveWhat it doesWhy it's irreducible
1Encoding (En)Stores/represents information in a specific mediumWithout encoding, no information exists in the system
2Structure (St)Organizes encoding — how encoded information is arrangedWithout structure, encoding is unorganized — no addressing, no hierarchy
3Evaluator (Ev)Translates encoding into function in a different mediumWithout evaluation, encoding is inert — stored but non-functional
4Direction (Dr)Guides what gets evaluated — regulatory controlWithout direction, evaluation is undirected — everything evaluated equally
5Output (Op)Functional product of evaluation — what the system producesWithout output, evaluation is internal — no observable effect
6Boundary (Bd)Defines system scope — what's inside vs outsideWithout boundary, no distinct system — merges with environment

7.4 Core triad: {En, St, Ev}

"What is an information substrate?" → Information is ENCODED (En), ORGANIZED (St), and EVALUATED (Ev) — translated from encoding into function.

All three pairs:

This maps to the concrete core triads:

7.5 The evaluator IS the critical variable

The abstract information substrate confirms: the evaluator (Ev) is where substrates differentiate.

SubstrateEv levelConsequence
BiologyEv4 (ribosome — deterministic molecular machine)Hard substrate — same codon always → same amino acid
Entity systemEv4 (dispatch — deterministic type-checked routing)Hard substrate — same typed data always → same handler
Cognition (formal)Ev3-4 (proof checking — near-deterministic)Hard mode — same proof steps always → same validity
Cognition (linguistic)Ev1-2 (interpretation — variable)Soft mode — same text → different meanings for different readers

The Ev3→Ev4 transition IS the hard-substrate gate. This is visible at the abstract level without reference to information theory or Shannon.

7.6 How this fits the invariant topology

The invariant topology discovered in the biology analysis:

    En → Ev → Mc → Sf → Cm ← Se
                    ↕        ↕
                    Cx ←→ Cm

The abstract information substrate {En, St, Ev, Dr, Op, Bd} sits at the En + Ev position in the SSA. Specifically:

The SSA's {En, Vr, Mc, Sf, Cx, Cm, Se} is the TOPOLOGY. The abstract information substrate {En, St, Ev, Dr, Op, Bd} is the CONTENT of the substrate nodes in that topology.

7.7 How this relates to the info-comp core

The info-comp core {St, Str, Tf, Ev, Me, Co, Pa} and the abstract information substrate {En, St, Ev, Dr, Op, Bd} are DIFFERENT abstractions of overlapping systems:

Abstract info substrateInfo-comp coreOverlap?
En (encoding)St (state)Related — encoding IS state in a medium
St (structure)Str (structure)Close — both about how things are organized
Ev (evaluator)Ev (evaluation)Direct match — same concept
Dr (direction)No counterpart — direction is architectural, not mathematical
Op (output)Tf (transfer)Related — output IS transferred result
Bd (boundary)No counterpart — boundary is architectural
Me (measure)No substrate counterpart — measure is mathematical
Co (constraint)No substrate counterpart — constraint is mathematical
Pa (parameter)No substrate counterpart — parameter is mathematical

What they share: encoding/state, structure, and evaluator. These are the UNIVERSAL concepts that appear in BOTH mathematical and architectural views of information processing.

What they don't share: the info-comp core has mathematical primitives (measure, constraint, parameter) that don't appear in the architectural substrate view. The abstract substrate has architectural primitives (direction, output, boundary) that don't appear in the mathematical info-comp view.

The EVALUATOR (Ev) is the bridge — it appears in both because evaluator determinism is simultaneously a mathematical property (decoder reliability) and an architectural property (translation mechanism fidelity).

7.8 The relationship between them

The abstract information substrate and info-comp core are connected by a role-identification edge — both describe information processing, from complementary perspectives:

A specific information substrate (biology, entity system, cognition) has positions in BOTH lattices:

Biology position in abstract info substrate lattice:
  En=Full, St=Full, Ev=4-Full, Dr=Full, Op=Full, Bd=Full

Biology position in info-comp core lattice (as information processing):
  St=Full, Str=Full, Tf=4, Ev=4-Full, Me=2, Co=2, Pa=1

The two positions describe the SAME system (biology) from different analytical perspectives. The role-identification edge connects them — showing which abstract substrate roles correspond to which info-comp properties.

7.9 What this means for the analysis

We now have THREE abstract domain mirrors at the substrate/surface/ecosystem levels of the invariant topology:

LevelAbstract domainPrimitivesDerived from
SubstrateAbstract information substrate6: {En, St, Ev, Dr, Op, Bd}Biology + entity system + cognition
SurfaceAbstract surface7+2: {St, Or, Rg, Pr, Ac, Pt, Ex, [Rs], [Gn]}Organism + app + cognitive arch
EcosystemAbstract ecosystem9: {Pd, Tf, Cy, Dv, In, Rg, Sp, Tp, Ct}Biological + cultural + (digital) ecosystem

And the info-comp core sits ALONGSIDE this as a complementary abstraction:

AbstractionWhat it abstractsPerspective
Abstract info substrateBiology, entity system, cognitionARCHITECTURAL — components and their roles
Info-comp coreIT, CC, DS, NTMATHEMATICAL — information-theoretic properties
SSAThe topology connecting substrate → surface → ecosystemSTRUCTURAL — how the architecture connects

These are three views of the same landscape:

7.10 Next steps

  1. Full 12-step of abstract information substrate {En, St, Ev, Dr, Op, Bd} — dependencies, pairs, sublattice, compositions, manifestation landscape

  2. Validate the mappings — check that each concrete substrate's primitives cleanly fill the 6 abstract roles

  3. Info-comp core revision — the Tr→Tf+Ev split stands, but this is a SEPARATE domain from the abstract info substrate. Both need analysis, but they're different domains connected by role-identification.

  4. Check: does the abstract info substrate have the same tight filter as concrete substrates? If the abstraction is right, the abstract domain should share structural properties with its instances (like the abstract surface shares loose filter with concrete surfaces).