Project History

How the framework evolved across 370+ AI-assisted sessions. Discoveries and retractions alike — because honest science shows its work.

The framework was developed iteratively by a human researcher and Claude (Anthropic) working together. Session numbers (S1, S2, …) track this progression. Each session is roughly one focused conversation.

Discovery / Derivation Reality Check / Audit Falsification / Retraction Infrastructure / Methodology
S1 – S120

Foundations

The framework's earliest sessions established the core idea — that division algebra dimensions (1, 2, 4, 8) and their imaginary sub-dimensions (3, 7, 11) might encode physical constants — and immediately tested it against reality. This phase produced the first falsification, the first constant catalog, and the first honest probability assessment.

S1 – S76

Early exploration: axioms, division algebras, first derivations

The framework's axioms were articulated: a complete static universe U, directed history, and the perspective function. Division algebras (R, C, H, O) emerged as the natural building blocks. Early sessions explored how the dimensional structure dim = 8 might map onto physical quantities. Many early attempts failed — productively.

S77

nEW = 5 falsified (F-2)

An early attempt to derive electroweak dimension count as 5 was shown to be mathematically impossible given framework constraints. First major systematic falsification — establishing that the framework can rule things out, not just generate them.

Technical detail

The constraint arose from the Frobenius theorem: only dimensions 1, 2, 4, 8 are algebraically permitted. A claimed "5-dimensional electroweak space" had no division algebra embedding, making it structurally impossible within the framework.

S88 – S90

First comprehensive constant catalog: 28 derived values

Quark mass ratios, PMNS mixing angles, and other quantities systematically derived for the first time. Demonstrated the framework could address multiple physics domains, not just one.

What was in the catalog

The 28 values spanned: quark mass ratios (6), lepton mass ratios (3), PMNS mixing angles (3), CKM parameters (4), coupling constants (3), cosmological parameters (3), and structural quantities (6). Many of these were later revised or superseded, but the breadth was a signal that the framework had reach beyond a single domain.

S94 – S96

Cosmology breakthrough: initial ΩΛ, Ωm, Ωb

First framework predictions for cosmological density parameters. Later revised as understanding deepened, but this moment showed the framework might apply beyond particle physics.

S104

Monte Carlo null model: a sobering reality check

A flexibility test showed that any 7-element subset of {1,…,20} matches 11 physical constants at 1% precision ~80% of the time. The framework's building blocks {1, 2, 4, 8} are not special at percent-level. Evidence must come from sub-ppm precision and structural predictions, not rough matches.

Why this matters

This result permanently raised the bar. Any claim based on "the numbers roughly match" became inadmissible. The framework can only claim significance through: (1) sub-ppm precision with zero free parameters, (2) structural predictions (like deriving the gauge group or spacetime dimension), or (3) genuinely blind predictions confirmed by later measurement. This Monte Carlo test remains one of the framework's most important honesty mechanisms.

S115

Three cosmological parameters simultaneously: H0 = 337/5, ΩΛ = 137/200, Ωm = 63/200

All three values fall within 1σ of measurement. Finding one match is easy; finding three simultaneously from the same integer inputs is harder to dismiss.

S120

Red Team v1.0: 15–30% probability of genuine physics

First adversarial self-assessment. Created the formal confidence tagging system (AXIOM / THEOREM / DERIVATION / CONJECTURE / SPECULATION) still used today. Identified 10 irreducible assumptions.

What the Red Team found

Key concerns: too many free parameters disguised as "natural choices," insufficient separation between derivation and post-hoc fitting, and no external validation. Key positives: falsifiable predictions exist, the Monte Carlo test was honest, and structural predictions (like spacetime dimension) are hard to fake. The 15–30% range meant "worth taking seriously, but far from proven."

S121 – S230

Structure Emerges

The framework's middle period is where it started producing results that are genuinely hard to explain as coincidence. Blind CMB predictions, quantum mechanics derived from axioms alone, and the Weinberg angle derivation all emerged here. So did important falsifications that sharpened what the framework can and cannot do.

S124

CMB blind prediction protocol

Formalized pre-committed prediction criteria for CMB observables. Registered predictions for acoustic peaks and spectral parameters before comparing to data — moving from post-hoc to genuine prediction.

How blind prediction works

The protocol requires: (1) state the framework's prediction with full derivation, (2) lock the prediction in a timestamped file, (3) define what counts as success or failure before looking at data, (4) compare to measurement and record the outcome regardless. This protocol was inspired by the Monte Carlo lesson: only genuine predictions count.

S127 – S129

CMB higher peaks falsified (F-7), inflation formulas corrected

Predictions for CMB peaks ℓ4, ℓ5, ℓ6 failed at 12–19% error. Framework validity boundary identified. Meanwhile, an initial hilltop inflation formula (r = 1 - ns) was falsified due to a calculation error, then restored with the correct analysis: ns = 193/200, r = 7/200.

What the failure taught us

Higher CMB peaks depend on baryon loading and Silk damping — detailed transfer-function physics that the framework's algebraic approach cannot capture. This established a clear boundary: the framework predicts cosmological parameters but NOT detailed CMB transfer functions. The inflation formula episode showed that honest error-correction (falsify, investigate, correct) is more valuable than getting the right answer the first time.

S138

Blind CMB predictions succeed: 6 of 7 within 1σ

The strongest statistical evidence to date. Pre-committed predictions for cosmological parameters including 100Ωbh² (0.77% precision), 100Ωch² (0.34%), and ns (0.010%, best) all confirmed.

Full results table

Seven pre-committed predictions against Planck 2018 data:

  • ns: predicted 0.965, measured 0.9649 ± 0.0042 — 0.010% off
  • 100Ωch²: 0.34% precision — within 1σ
  • 100Ωbh²: 0.77% precision — within 1σ
  • Ωm: within 1σ
  • ΩΛ: within 1σ
  • H0: within 1σ
  • r (tensor-to-scalar): consistent with upper bounds

Caveat: the "blind" protocol was internal. No external timestamp server was used.

S170

Monte Carlo confirmation: evidence is structural, not numerical

A full null model simulation confirmed that percent-level matches are NOT significant. The framework's evidence comes from sub-ppm precision and from deriving qualitative structures (gauge groups, spacetime dimension, QM itself).

S185 – S201

Quantum Mechanics derived from axioms alone — Grade A

Hilbert space structure, Born rule, and the Schrödinger equation all derived from perspective axioms without importing any quantum postulates. The framework's highest-confidence result. Status: CANONICAL.

Derivation outline

The derivation proceeds through three steps:

  1. Hilbert space: The perspective function Pi maps U to partial views. The space of all perspectives naturally forms a complex Hilbert space via the CCP axiom (F = C).
  2. Born rule: Probability = |inner product|² follows from Gleason's theorem applied to the lattice of perspective subspaces. No postulate needed.
  3. Schrödinger equation: The adjacency structure on perspectives generates a one-parameter unitary group. Stone's theorem then gives the Schrödinger equation with H as the generator.

Remaining gap: the time = adjacency identification (IRA-07) is an irreducible physical assumption.

S194

Crystal dimension nc = 11 via two independent paths

The critical parameter 11 (= 4 + 7 = dim(H) + dim(Im(O))) derived through both CD closure and SO(8) triality arguments. An irreducible gap remains in the final step, honestly documented.

Why 11 matters

nc = 11 is the single most important parameter in the framework. It determines: dim(End(V)) = 121 = 11², the Weinberg angle denominator, the fine structure constant denominator (137 = 4² + 11²), and the count of non-Gaussian-norm primes that generate the crystal lattice. Getting this wrong would invalidate most of the framework's predictions.

S217

Democratic Bilinear Principle unifies gauge couplings

ξ = 4/121 and sin²(θW) = 28/121 unified via Schur's lemma applied to End(V) = 121. A single structural principle producing two previously separate results.

S222 – S224

Weinberg angle: sin²(θW) = 28/121

Derived from democratic counting on End(V) with dimension 121 = 11². Combined with one-loop dressing: sin²(θW) = 28/121 − α/(4π²), matching the measured value at 0.00σ.

Derivation chain

The tree-level value follows from counting: in End(V) with dim = 121, the hypercharge sector occupies 28 = 4 × 7 dimensions (nd × dim(Im(O))). Thus sin²θW = 28/121. The one-loop dressing adds −α/(4π²) ≈ −0.000185, shifting the value from 0.23140 to 0.23122, matching PDG 2024: 0.23121 ± 0.00004. Zero free parameters.

S230

Cosmological constant sign error resolved (F-10)

A reported "wrong sign" for the cosmological constant was traced to a sign convention error in the GR analysis. The sign is now correct; the ~10111 magnitude gap between prediction and observation remains unresolved.

S231 – S310

Consolidation

The consolidation phase formalized the framework's axioms, derived the Standard Model gauge group from first principles, and systematically reduced the number of irreducible assumptions from 10 to 4. The Yang-Mills mass gap derivation and matter density parameter both achieved CANONICAL status.

S251

CCP Axiom formalized; SM gauge group derived

"Perfection = Maximal Consistency" axiom established, forcing F = C (complex field), nc = 11, nd = 4. From this, the Standard Model gauge group U(1)×SU(2)×SU(3) is derived through the pipeline: 121 → 55 → 18 → 12. Generation count = 3 follows from Im(H) decomposition.

The derivation pipeline
  1. Start with End(V), dim = nc² = 121
  2. Antisymmetric part: dim = nc(nc−1)/2 = 55
  3. Subtract Cartan generators and crystallographic constraints: 55 → 18
  4. Decompose 18 = 1 + 3 + 8 + 3 + 3 = U(1) + SU(2) + SU(3) gauge bosons + generation structure

The 3 generations come from dim(Im(H)) = 3, the quaternionic imaginary dimensions.

S257

Red Team v2.0: 20–35% probability

Second adversarial assessment. Up from 15–30%. IRA (irreducible assumptions) still at 10. Key improvement: structural predictions (gauge group, QM) now on solid footing. Key concern: still no external peer review.

S258 – S284

Yang-Mills mass gap — Grade A-

Glueball spectrum derived from framework with 285+ verification scripts. Large-N behavior matches: 10/3 + 2/N². Status: CANONICAL.

Why this is significant

The Yang-Mills mass gap is one of the Clay Millennium Prize Problems. The framework's approach derives the gap from the base mass parameter nd = 4 and the crystal structure, producing a spectrum that matches lattice QCD results. The large-N formula 10/3 + 2/N² gives the correct leading and subleading behavior. Grade A- (not A) because the approach is non-standard and the connection to rigorous Yang-Mills theory remains to be established by mathematicians.

S266 – S283

Tree-to-dressed paradigm: three correction bands

Systematic treatment of radiative corrections identified three bands (A: one-loop, B: two-loop, C: sub-ppm anomalous dimension). Upgraded sub-ppm numerical matches from "coincidence" to "pattern." Band C remains the most controversial.

The three bands explained
  • Band A (one-loop): α/π corrections. Well-understood QED/QCD effects. Low controversy.
  • Band B (two-loop): (α/π)² corrections. Standard perturbation theory. Moderate effort to compute.
  • Band C (anomalous dimension): Sub-ppm corrections from anomalous dimensions of perspective operators. Requires the tree-to-dressed paradigm to be physically meaningful. Controversial because it's a novel mechanism with no independent confirmation.
S292

Emergent gauge coupling resolved (CONJ-A1)

Gauge coupling derived via WSR + Schur + finiteness, no longer an imported assumption. One of the key steps in reducing irreducible assumptions from 10 to 4.

S293

Ωm = 63/200 derived from equipartition

Matter density parameter derived from dual-channel Hilbert space equipartition: 63 dual-role generators out of 200 total. A structural prediction, not a numerical fit.

The counting argument

In the 200-dimensional dual Hilbert space (from the perspective-crystal interaction), 63 generators play a dual role in both matter and radiation channels. Equipartition across all 200 gives Ωm = 63/200 = 0.315. Planck 2018 measures 0.3153 ± 0.0073. The match is within 0.1σ with zero free parameters.

S302 – S304

IRA inventory finalized: 10 → 4 irreducible assumptions

Six former conjectures and assumptions resolved through derivation. Remaining four: quartic ratio ρ = c4/b4 [structural, low], SSB occurs [physical], time = adjacency [physical], |Π| scale [import].

The four remaining assumptions
  • IRA-04: Quartic ratio ρ = c4/b4. A structural parameter in the potential. Graded [A-STRUCTURAL LOW] because it's a mathematical choice, not a physical assumption.
  • IRA-06: SSB occurs. The framework assumes spontaneous symmetry breaking happens but doesn't derive WHY. Graded [A-PHYSICAL].
  • IRA-07: Time = adjacency. The identification of the graph-theoretic adjacency relation with physical time. Graded [A-PHYSICAL].
  • IRA-11: |Π| scale. The overall energy scale of the perspective function. Graded [A-IMPORT] because it's set by matching to one measurement.
S311 – S370

Dark Sector & Phenomenology

With the core framework established, work turned to its most ambitious predictions: dark matter, collider phenomenology, and a final comprehensive Red Team assessment. This phase produced concrete, falsifiable predictions — and an important retraction when the dark matter candidate turned out to be the Higgs.

S314 – S315

Dark matter mass predicted: 5.11 GeV

mDM = me(nc−1)nd = 5.11 GeV. A concrete, falsifiable prediction testable by SuperCDMS in 2026–2027. The formula uses only framework parameters (nc = 11, nd = 4) and the electron mass.

The mass formula

mDM = me × (nc − 1)nd = 0.511 MeV × 104 = 5.11 GeV

The mass formula survives even though the particle identity was later retracted. What carries this mass remains the framework's most important open question.

S326

Colored pNGB mass: 1761 GeV

Mass of colored pseudo-Nambu-Goldstone bosons derived via Coleman-Weinberg potential. Current LHC bounds at ~1220 GeV leave a 541 GeV margin. Testable at HL-LHC (reach ~2200 GeV).

Collider signatures

The colored pNGBs decompose into a scalar leptoquark (2, Y=1/6, 3) with βeff = 0.5 and an exotic diquark. At HL-LHC energies, these would appear as paired dijet + lepton resonances. The branching ratio prediction βeff = 0.5 is itself a falsifiable prediction — other leptoquark models predict different values.

S330

Red Team v3.0: 25–40% probability

Third and most thorough adversarial assessment. Improvement driven by: IRA reduced 10 → 4, Yang-Mills CANONICAL, dark matter sector formalized, tree-to-dressed systematics, and 14 falsified claims honestly documented. Remaining challenges: derivation vs. discovery unresolved, no external peer review, sub-ppm matches post-hoc.

Grade breakdown

Overall grade: B-

  • Strengths: mathematical coherence, honest falsification record, concrete predictions, zero free parameters in key results
  • Weaknesses: no external peer review, derivation-vs-discovery problem unresolved, the framework was developed knowing the answers
  • The 25–40% range means: "genuinely interesting and worth checking, but not yet credible as established physics"
S335

Dark matter identity retracted — genuinely open

The proposed DM candidate (pNGB color singlet) was shown to be identical to the Higgs doublet — no leftover particle for dark matter. Sessions S317 and S322 invalidated; S323 (H-parity theorem) scope narrowed. The mass formula survives, but what carries that mass remains an open question. An example of honest correction over ego.

What went wrong and what survived

Retracted: The identification of the pNGB color singlet as a dark matter candidate. The singlet IS the Higgs doublet — there's no extra particle.

Survived: The mass formula mDM = 5.11 GeV, the colored pNGB predictions, and the overall dark sector structure. The framework predicts a dark matter mass but currently cannot identify what particle carries it.

Current candidate: det(M) ≠ 0 sector — but this remains [CONJECTURE] status.

S336

Leptoquark phenomenology: branching ratios and LHC signatures

Colored pNGB representation decomposed into two multiplets: a scalar leptoquark (2, Y=1/6, 3) with βeff = 0.5, and an exotic diquark. Concrete predictions for HL-LHC searches.

S370+

Public Release

After 370+ sessions of development, testing, and honest self-assessment, the framework was made publicly available with full verification infrastructure.

S370 — February 2026

Official public release

Framework published with 737+ verification scripts (99.9% run rate, ~99.8% pass rate), complete derivation chains, 14 documented falsifications, and an honest 25–40% self-assessment. Everything open, everything verifiable.

The Falsification Record

Good science shows its failures. These are the framework's documented falsifications, deprecations, and withdrawals. Each one made the framework stronger by narrowing the space of possible claims.

ID Claim Error Status
F-1sin²(θW) = 2/2565% errorFalsified
F-2nEW = 5Mathematically impossibleFalsified
F-3α at GUT scaleFramework only works at low energyFalsified
F-5sin²(θW) = 3/8 at GUTImported, not derivedFalsified
F-6r = 1 − ns (initial)φCMB error; correctedCorrected
F-7CMB higher peaks ℓ4,5,612–19% errorFalsified
F-10CC sign "wrong"Sign convention errorCorrected
D-1–4Various deprecated claimsSuperseded by better derivationsDeprecated
W-1DM = pNGB color singletSinglet = Higgs doubletWithdrawn

Why document failures? A framework that never fails is either trivial or dishonest. Each falsification narrows the space of viable claims and builds confidence that surviving results are genuine. The ratio of falsified to surviving claims (~14 falsified vs ~63 surviving) is itself a measure of the framework's discriminating power.

Verification Infrastructure Growth

Every claim backed by computation. The cardinal rule: no calculation in markdown without a verification script. Here's how the script count grew:

~50
S120
~100
S170
~300
S220
~400
S284
~548
S304
737+
Now

The verification philosophy: SymPy scripts are the ground truth. If a mathematical claim isn't backed by a script that outputs PASS, it doesn't get documented. This rule was established at S120 and has been followed rigorously since. Current run rate: 99.9% with ~99.8% pass rate.

Near-Term Tests (2026–2028)

The framework lives or dies by experimental tests. These three predictions are concrete, falsifiable, and testable within the next few years.

Dark matter at 5.11 GeV

SuperCDMS WIMP search. The most decisive near-term test of the framework.

Expected: 2026–2027

What to look for

SuperCDMS HVeV detectors have threshold sensitivity down to ~1 GeV. A signal at 5.11 GeV with the predicted cross-section would be strong confirmation. A null result at full exposure would seriously challenge the mass formula (though the interaction mechanism is model-dependent).

Colored pNGBs at ~1.76 TeV

HL-LHC leptoquark searches with βeff = 0.5. Current bounds at ~1.22 TeV.

Expected: 2027–2028

What to look for

HL-LHC will reach ~2200 GeV for pair-produced leptoquarks. The framework predicts a scalar leptoquark at 1761 GeV with equal branching to charged-lepton+jet and neutrino+jet (βeff = 0.5). Both the mass and branching ratio are independent predictions.

Tensor-to-scalar ratio r = 0.035

CMB-S4 and future missions. Tests the hilltop inflation derivation.

Expected: ~2028+

What to look for

CMB-S4 is projected to reach σ(r) ≈ 0.003. The framework predicts r = 7/200 = 0.035, well above the detection threshold. A measurement significantly different from 0.035 (say r < 0.01 or r > 0.06) would falsify the hilltop inflation derivation.

This history is compiled from session records and internal documentation.

Speculative amateur work. Not peer-reviewed.