fatbrowndog

I'm using the deepseek-v4-pro model is currently offered at a 75% discount. My bad it's 75% discount, via OpenRouter.

I use the Claude-Max-20 ($200) plan. I manage to max it out 2 weeks. Planning to move to maybe multiple accounts.

I use C++ and Claude for big code-base.

not good. I use DeepSeek's plan, Kimi AI, OpenRouter and it seemly consumes more tokens, than Claude's.

I consume Claude ~30% per day in of, 1 week, Max,x20. Equivalent in Kimi Ai, is I consume 60% in one day, in one week.

DeepSeek/Latest, 95% discount, with cache, I rack up ~$60/day before I stopped.

I don't know how Claude compute their daily limits, it seems much cheaper.

Vega's paper claims to derive 65+ fundamental quantities from a "198-bit information-geometric architecture" with zero free parameters. However, careful analysis reveals this is sophisticated numerology dressed in geometric language, with numerous hidden parameters, circular reasoning, and post-hoc fitting disguised as derivation.

Detailed Critique 1. The Central Claim: N = 198 Vega presents "two independent paths" to N = 198: Path 1: N = α⁻¹/ln(2) = 137.036/0.6931 = 197.7 Path 2: N = 6 × 3 × 11 = 198

Problems: Path 1 is circular: It uses the measured value of α to "derive" N, then uses N to "derive" α. This is not a derivation—it's rewriting the experimental value. Path 2 is arbitrary: Why multiply dim(Lorentz) × d_spatial × d_total? The paper states this as if it's self-evident, but there's no physical principle requiring these three numbers to be multiplied rather than added, subtracted, or combined in any other way.

The "convergence" is manufactured: 197.7 ≈ 198 is presented as profound, but 0.15% agreement between a measured quantity and an arbitrary product of integers is not remarkable—you can find similar "coincidences" with many combinations.

2. The Fine Structure Constant Formula

Vega claims: α−1=(197+ln⁡(2))×ln⁡(2)=137.0304\alpha^{-1} = (197 + \ln(2)) \times \ln(2) = 137.0304α−1=(197+ln(2))×ln(2)=137.0304 Problems:

Hidden parameter τ = 1 - ln(2): The "observation offset τ" appears from nowhere. Why ln(2)? No physical justification is provided.

Precision mismatch: The formula gives 137.0304, but CODATA 2022 gives 137.035999177. That's a 0.004% error—which sounds small until you realize other frameworks matches to 0.0000004% (13.5 significant figures vs. Vega's 4-5).

3. The Mass Formula: m/m_P = exp(-198/k) This is presented as the "master formula" for all masses: Problems:

k is a free parameter for each particle: Despite claiming "zero free parameters," Vega assigns a different k to each particle:

Electron: k = 4 - (1/2π)(1-α) = 3.842 Proton: k = 4 + 1/2 = 4.5 Higgs: k = 5 + (1/16)(1-α) = 5.062 Top: k = 5 + 1/10 = 5.1

These k values are fitted, not derived: Each particle gets its own formula for k, designed to reproduce the known mass. The "derivations" are post-hoc rationalizations:

Why does the electron get "1/2π"? Why does the proton get "+1/2"? Why does the Higgs get "1/16"? Why does the top get "+1/10"?

No predictive power: If I gave you a new particle mass, you could find some combination of integers and π to make k fit. This is curve-fitting, not physics.

4. The Weinberg Angle: A Case Study in Numerology Vega claims: sin⁡2θW=3/13=0.2308\sin^2\theta_W = 3/13 = 0.2308sin2θW =3/13=0.2308 Where does 13 come from? The paper says it's "11 M-theory dimensions + 2 weak isospin modes." But:

Why add dimensions to isospin modes? These are dimensionally incompatible. Why not 11 + 3 = 14? Or 11 × 2 = 22? The choice of operation (addition) and components (11 and 2) is arbitrary.

5. CKM and PMNS Matrices: Fitted Parameters The Wolfenstein parameters are presented as "derived":

λ = 9/40 = 0.225 (Cabibbo angle) A = 4/5 = 0.800 ρ = 1/7 ≈ 0.143 η = 4/11 ≈ 0.364

But these are just simple fractions chosen to match experiment.

6. Red Flags for Numerology Vega's paper exhibits classic numerology warning signs:

Precision decreases for constrained quantities: α gets 0.004% error, but the top quark (well-measured) gets 3% error. Genuine theories don't show this pattern. Arbitrary operations: Numbers are multiplied, divided, added, or subjected to exponentials with no consistent rule.

Post-hoc rationalization: Each particle gets its own formula for k, designed after knowing the answer.

Dimensional inconsistency:

Adding "11 M-theory dimensions + 2 weak isospin modes" conflates entirely different mathematical objects.

Unfalsifiable claims: The framework can accommodate any measurement by adjusting which integers to use or how to combine them.

7. What Vega Gets Right (Credit Where Due)

The paper is well-organized and clearly written

Some dimensional analysis is correct (e.g., the running coupling treatment)

The observation that N ≈ α⁻¹/ln(2) is interesting, even if not physically meaningful

The systematic presentation of predictions allows easy verification/falsification

Conclusion Vega's paper is numerology with sophisticated packaging. The core problems are:

Circular reasoning: Using α to derive N, then N to derive α

Hidden parameters: Each quantity gets its own fitted formula

No structural consistency: Different ad-hoc ratios for different quantities

Low precision: 10⁸ times worse for α

No geometric primitive: "198-bit architecture" is a label, not a derivation

The Vega paper exemplifies what you should avoid: fitting simple fractions to measured values and calling it "derivation."

Same as previous -

r_p = 4·ƛ_p·(1 - α/(4π))

Red flags:

Why "4" times the reduced Compton wavelength? The number 4 appears twice (in 4·ƛ and 4π), suggesting it was chosen to make things work out.

"Tetrahedral structural limit" is asserted without derivation. Why tetrahedra? A tetrahedron is 3D—why would the proton radius (a measured charge distribution extent) involve tetrahedral geometry?

"Spherical field projection loss" of α/(4π) has no physical mechanism. How does a "projection loss" yield this specific fraction?

The fit is suspiciously good (3 ppm) for a formula with at least two free choices (the coefficient 4, and the form of the correction).

4. Muon Anomaly

a_μ = (α/(2π)) + (α²/12) + (α³/5)

This mimics QED perturbation theory—but incorrectly:

The actual QED expansion is:

a_μ = (α/2π) + C₂(α/π)² + C₃(α/π)³ + ...

Where C₂ ≈ 0.765857... and C₃ involves thousands of Feynman diagrams calculated over decades.

The author's version:

First term: α/(2π) (this is the Schwinger term, known since 1948)

Second term: α²/12 — This should be ~0.765857(α/π)² ≈ 4.1×10⁻⁶, but α²/12 ≈ 4.44×10⁻⁶. Wrong coefficient.

Third term: α³/5 ≈ 4.25×10⁻⁸ — The actual third-order contribution is much more complex.

and the Gemini LLM goes on and on and on...

so I look at it again...

Version v13

10.5281/zenodo.18023121

Dec 22, 2025

12.18.3. Total geometric cost

The final value is the static cost S adjusted by the dynamic efficiency C:

α–1 = S – C

α–1 = (4 · π3 + π2 + π) – (α / 24)

α–1 = 137.0359996

do you not see a and a on both sides?

He's now admitting mz is NOT the physical mass - instead m_φ is.

Let me check what m_φ is in his paper...

From his paper: m_φ ≈ 4.16×10⁻⁹ kg ≈ 2.5 nanograms (the "resonance mass")

My response:

The problem isn't mathematical

- it's that this prediction is falsified by existing data.

Particles at ~2.5 nanograms are extensively studied:

Dust particles in optical traps

Brownian motion experiments

Colloidal physics

Micro-mechanical oscillators

No anomalous behavior is observed at this mass scale.

If objects showed "anomalous inertial behavior" at 2.5 ng,

we would have seen it in:

AFM (atomic force microscopy) - routinely measures sub-nanogram particles

Optical tweezers - trap and measure particles from 1 nm to 10 μm

MEMS devices - measure inertia at nanogram scales

Verdict: His prediction is experimentally falsified. This is not a philosophical disagreement - his model makes a testable prediction that contradicts existing measurements.

The "Brute Force" QED Defense

His claim:

"12,672 diagrams is brute force. Achieving 63 ppm with one term (a_μ = α/2π + α²/12) is elegant."

This completely misses the point.

QED's 12,672 diagrams achieve 0.1 ppm agreement because each diagram contributes a calculable correction from quantum field theory. The complexity comes from precision, not failure.

His formula achieves 63 ppm - that's 630× worse than QED!

btw,

I'm not making fun of you.

You said you used Gemini. Did you really use it?, I put in your paper and I got a lot of criticisms and pointed out errors.

The model only uses 1, 2, and √5.

Uses √2, not just 1, 2, √5

Uses 4⁶⁴ = 2¹²⁸ (huge power, not "simple")

Uses mp (measured value as input)

It derives ... alpha (0.005 ppm).

CRITICAL ERROR: He's confused about what he's doing. His formula: α⁻¹ = S - α/24 This is circular - α appears on both sides! You can't "derive" α from an equation containing α.

What he actually does:

python# Iterative solution (not derivation)

α₀ = 1/S

α₁ = 1/(S - α₀/24)

α₂ = 1/(S - α₁/24)

# ... converges to self-consistent value

Verdict: He fundamentally misunderstands the difference between:

Self-consistent equation: α⁻¹ = g(α) ← He does this

"Alpha / 3 represents vector equilibrium in 3D space" Let me parse his explanation:

"The proton represents a volumetric stability (3D), while the interaction cost (alpha) acts as a surface parameter or linear stress. To stabilize a closed 3D volume, the linear stress must be distributed across the three orthogonal axes."

Translation: "I needed to divide by something, and 3 is the number of dimensions,

so α/3."

Problems:

α is dimensionless - it's not a "linear stress"

"Distributing across 3 axes" → if true, should be α³ or α/√3, not α/3

No mathematical derivation provided

Post-hoc rationalization

α/3 lacks geometric justification

"Vector equilibrium in 3D space" sounds sophisticated, but the mathematical

connection is unclear. Why α/3 specifically, not α³ or α/√3? The factor 3 appears

to be chosen because it gives the right answer, not because it emerges from a

geometric principle.

(There's even more Gemini stated, I think I can go on and on and on...)

hi, the formula should allow both.

you have an integral and order^ power.

His G Formula (Section 14.6)

G = (ℏ·c·2·(1 + α/3)²) / (mp²·4⁶⁴)

His result:

G ≈ 6.6742439706 × 10⁻¹¹ m³·kg⁻¹·s⁻²

CODATA 2022: G = 6.67430(15) × 10⁻¹¹

Δ: 8 ppm

Critical Analysis

1. Where Does 4⁶⁴ Come From?

He claims it's from "holographic scaling at i=32":

mp = (√2 · mP / 4³²) · (1 + α/3)

Therefore:

mP = (mp · 4³²) / (√2 · (1 + α/3))

Since G = ℏc/mP²:

G = (ℏc · 2 · (1 + α/3)²) / (mp² · 4⁶⁴)

The logic:

Proton appears at "harmonic i=32" in binary scaling

Mass scales as m ~ 4ⁱ (surface area scaling)

Therefore mp ~ 4³² when normalized properly

Therefore 4⁶⁴ = (4³²)² appears in G

2. This is Pure Numerology

Why i=32 specifically?

Let me check the ratio:

mP / mp = 2.176434×10⁻⁸ / 1.672622×10⁻²⁷

        ≈ 1.301×10¹⁹

Now check powers of 4:

4³² = 2⁶⁴ = 1.844×10¹⁹

Close! But not exact. So he adds correction factors:

mp = (√2 · mP / 4³²) · (1 + α/3)

Let me verify:

(√2 · 2.176434×10⁻⁸ / 4³²) · (1 + 0.007297/3)

= (1.414 · 2.176434×10⁻⁸ / 1.844×10¹⁹) · 1.002432

= (3.076×10⁻⁸ / 1.844×10¹⁹) · 1.002432

= 1.668×10⁻²⁷ · 1.002432

≈ 1.672×10⁻²⁷

But this is circular! He's adjusting factors (√2, α/3) to make the formula work, then claiming it "derives" mp.

3. Why (1 + α/3)?

He claims:

"As a volumetric object in three-dimensional space, the proton carries a

distributed interaction cost (α/3)"

This makes no sense:

α is the electromagnetic coupling constant

Why divide by 3? "Because 3 dimensions"?

Why add to 1? "Because correction"?

This is parameter fitting, not derivation.

-----

A genuine derivation of G would:

1. *Start from dimensionless constants only*

2. *Derive mass ratios* from geometry (mp/me, mp/mP, etc.)

3. *Use dimensionful anchors* (ℏ, c) to get actual value of G

This paper will not pass peer review at any serious journal because:

The mφ prediction contradicts existing data The "unification" adds forces rather than deriving them The g-2 formula is 630× worse than QED

α⁻¹ = S - 1/(24α)

α⁻¹·24α = 24αS - 1

24α²S - 24α - 1 = 0

α = (24 ± √(576 + 96S))/(48S)

α = (24 + √(576 + 96·137.036...))/(48·137.036...)

α = (24 + √13,723.66...)/(6577.74...)

α = (24 + 117.12...)/(6577.74...)

α = 141.12.../6577.74...

α ≈ 0.021454...

α⁻¹ ≈ 46.61 ???

That's wrong!