Is the Standard Model overfitting or am I curve-fitting?

3•albert_roca•1mo ago

I am developing a geometric model of physical interactions based on geometric constraints (w = 2, δ = √5 ) and topological invariants. No free parameters, just geometry. In your opinion, is this a legitimate geometric unification or just sophisticated curve-fitting?

Results:

Proton radius (r_p): Modeled as a tetrahedral structural limit (4 · ƛ) with spherical field projection loss (α / 4 · π).

  r_p  = 4 · ƛ_p · (1 - (α / (4 · π)))
  Pred: 8.407470 × 10^-16 m
  Exp:  8.4075(64) × 10^-16 m
  Diff: 3 ppm

Proton magnetic moment (g_p): Derived from the dynamic potential (δ = √5 ) damped by a golden friction term (α / Φ).

  g_p = (δ^3 / w) - (α / Φ)
  Pred: 5.5856599
  Exp:  5.5856947
  Diff: 6 ppm

Muon anomaly (a_μ): Derived as a hierarchical resolution of the icosahedral geometry: surface (α / 2 · π) + nodes (α^2 / 12) + vertex symmetry (α^3 / 5).

  a_μ = (α / (2 · π)) + (α^2 / 12) + (α^3 / 5)
  Pred: 0.00116592506
  Exp:  0.00116592059
  Diff: 4 ppm

α particle radius (r_α): Modeled as a 4-nucleon tetrahedron (8 · ƛ) with a linear nucleonic projection cost (α / π).

  r_α = 8 · ƛ_p · (1 - (α / π))
  Pred: 1.67856 × 10^-15 m
  Exp:  1.678 × 10^-15 m
  Diff: 330 ppm

Proton mass (m_p): Connecting the Planck scale to proton scale via a 64-bit metric horizon (2^64) and diagonal transmission (√2 ).

  m_p = ((√2 · m_P) / 2^64) · (1 + α / 3)
  Pred: 1.67260849206 × 10^-27 kg
  Exp:  1.67262192595(52) × 10^-27 kg
  Diff: 8 ppm

Neutron-proton mass difference (∆_m): Modeled as potential energy stored in the geometric compression of the electron (icosahedron, 20 faces) into the protonic frame (cube, 8 vertices). Compression ratio = 20/8 = 5/2.

  ∆_m = m_e · ((5/2) + 4 · α + (α / 4))
  Pred: 1.293345 MeV
  Exp:  1.293332 MeV.
  Diff: 10 ppm.

Gravitational constant (G) without G: Derived from quantum constants and the proton mass, identifying G as a scaling artifact of the 128-bit hierarchy (2^128).

  G = (ħ · c · 2 · (1 + α / 3)^2) / (m_p^2 · 2^128)
  Pred: 6.6742439706 × 10^-11
  Exp:  6.67430(15) × 10^-11 m^3 · kg^-1 · s^-2
  Diff: 8 ppm

Fine-structure constant (α): Derived as the static spatial cost plus a spinor loop correction.

  α^-1 = (4 · π^3 + π^2 + π) - (α / 24)
  Pred: 137.0359996
  Exp:  137.0359991
  Diff: < 0.005 ppm

Preprint: https://doi.org/10.5281/zenodo.17847770

Comments

bigyabai•1mo ago

If you have to ask people whether or not your preprint resembles curve-fitting, you have just self-reported that you are an AI user with no academic background.

Good luck with the peer review, you're gonna need it.

albert_roca•1mo ago

I have reported nothing but numerical results. Making assumptions about me instead of looking at the numbers says more about your background than it does about mine.

bigyabai•1mo ago

I have done nothing but associate your "numerical results" with other numberslop I see from LLMs. Again, you're self-reporting.

albert_roca•1mo ago

Can you share the results of your analysis by association? Or was it an instant mental calculation?

yuuu•1mo ago

From the manuscript linked in your profile:

> The author declares the intensive and extensive use of Gemini 2.5 Flash and Gemini 3.0 Pro (Google) and sincerely thanks its unlimited interlocution capacity. The author declares as their own responsibility the abstract formulation of the research, the conceptual guidance, and the decision-making in case of intellectual dilemma. The AI performed the mathematical verification of the multiple hypotheses considered throughout the process, but the author is solely responsible for the final content of this article. The prompts are not declared because they number in the thousands, because they are not entirely preserved, and because they contain elements that are part of the author’s privacy.

albert_roca•1mo ago

This seems properly copied and pasted. Good job. I guess we agree that AI is already playing a central role in science, and physics is no exception.

yuuu•1mo ago

> AI performed the mathematical verification

That should be done by the human writing the manuscript, i.e., you.

albert_roca•1mo ago

Absolutely not. Results don't depend on who performed the calculation or how it was done. Can you solve 12,672 Feynman diagrams by hand?

rolph•1mo ago

i can. and i will take longer than you.

i will take longer, because at each step the process of lateral association occurs, this will foster imaginative variation of schema, and result in inspiration, an internally generated drive to pursue a goal, and experience the results.

i will not only complete the task, but will understand the many outcomes of task corruption as they relate to the components of the task.

you will obtain a set of right answers, i will discover the rules that govern the process.

albert_roca•1mo ago

Fair enough. However, it is practically impossible to complete such a task in a human lifetime. But even if it were possible, the main point stands: using computers to perform calcualtions is standard scientific practice. Discrediting a proposal solely because it uses AI is retrograde per se. It contradicts the history of technological progress and excludes potentially valid results based on intellectual prejudice.

rolph•1mo ago

who discredited your proposal?

albert_roca•1mo ago

I am referring to other comments in this thread that dismissed the proposal purely based on the use of AI tools. My comment about prejudice was not directed at you.

rolph•1mo ago

consider the conceptual model of particle as a polyhedral structure.

consider further, the [pred] values are an average, or a centroid of sort, related to a dynamic process, as a result, the straight edges, and faces of the polyhedron dont exist, they are virtual. what is actual is the variation of "curvature" as the object oscillates, further consider that [diff] is a measure of deviation that is in line with [exp] values.

albert_roca•1mo ago

Because AI has been in the center of the debate so far, I ran your comment through my AI system, and it concluded that you captured the essence of the model perfectly: the polyhedra are topological standing waves, and the edges are nodal lines. So [Pred] is the geometric attractor, and [Diff] is the amplitude of the oscillation around that limit. As I understand it myself, the polyhedra don't exist as real solids, but as an optimized way to distribute the intensity of the oscillation. Does this perspective make the results physically plausible in your view?

rolph•1mo ago

it is one plausible interpretation.

attached is the question of what is "oscillating" ?

is matter, composed of "spacetime" possessed of disequilibrial state?

or is matter something different than the surrounding "substance"?

where does the phenomenal energy originate to drive a proton for the duration of its existance [decay rate]. is there some topologic ultrastructure that constrains geometry and drives the process of being a proton?

pavel_lishin•1mo ago

Based on your pre-previous post, this is nothing.

albert_roca•1mo ago

Your contribution is the opposite of "something".

rolph•1mo ago

a much more revelatory exercise would be to compare these derived values with measured values, then construct testable hypotheses regarding disparities.

albert_roca•1mo ago

That's precisely what the numbers show. "Pred:", predicted value. "Exp:", experimental value. "Diff", difference.

rolph•1mo ago

the next step is, why?

what assumptions does your current model make. what could change that would eliminate disparity. What plausible mechanisms explain [Diff]?

albert_roca•1mo ago

The model shows that the surface and volume of an object scale with mass such that electrostatic and gravitational acceleration can be explained through this scaling relationship. This is considered a geometric or structural cost:

  C_s ~ m^(1/3) + m^(-2/3)

In terms of intrinsic acceleration, surface and volume scale with mass as:

  a_i ~ m^(1/3) + m^(-5/3)

This relationship holds for any object with charge ≠ 0 across electrostatic and gravitational regimes, so the free fall principle is strictly recovered only for mathematically neutral objects.

This allows drawing an intrinsic acceleration curve for objects with homogeneous density, and the minimum point of this curve is identified at:

  m_ϕ ≈ 4.157 × 10^−9 kg

If the surface and volume of a not strictly neutral object determine its dynamic behavior, this would theoretically allow measuring m_ϕ with precision and deriving G without the historical dependence on the Planck mass. In this sense, it is a falsifiable proposal.

The geometric logic of the model allows establishing a geometric or informational saturation limit that eliminates GR singularities. At the same time, fundamental particles are not treated as dimensionless points but as polyhedral objects, which also eliminates the quantum gravity problem. The concept of infinity is considered, within the model, physically implausible.

From here, the model allows making the derivations included in this post, which I have not presented categorically, but as a proposal that seems at least statistically very unlikely to be achieved by chance.

The model does not question the precision of the Standard Model but postulates that the particle zoo represents not a collection of fundamental building blocks, but the result of proton fragmentation into purely geometric entities. The fact that these entities are not observed spontaneously in nature, but only as a consequence of forced interactions, seems to support this idea.

proteal•1mo ago

Hey - plugged this into chatGPT 5.2 and it seems to think this theory needs more work.

“As written, this looks closer to sophisticated curve-fitting (numerology with constraints) than a legitimate geometric unification, mainly because the claimed “ppm agreement” is often not assessed against experimental uncertainties and because several integer/constant choices function like hidden degrees of freedom.”

Thank you for sharing and happy holidays!

albert_roca•1mo ago

Thanks for running this on GPT 5.2. It is fascinating to see AI critiquing AI-assisted work.

The critique regarding hidden degrees of freedom is a fair point. However, in curve-fitting, parameters are continuous: one can choose 4.1 or 3.9 to make the data fit. In this model, parameters are topological invariants (integers like 4 faces, 12 vertices, 20 faces). They are discrete and cannot be tuned.

The fact that this unadjustable logic yields results agreeing with experimental data within ppm implies either a massive statistical coincidence or a structural aspect.

It would be very interesting to run independent tests on different AIs with the whole context of the model and a standardized, consensual prompt. Beyond formal verification, this methodology could open paths that are difficult to navigate without AI assistance, helping to determine if the model stands as a possible foundation for a 'broad explanation of the observable', since the term 'ToE' instantly raises red flags. Kind of a pioneer peer-centaur-review. Just an idea.

Thanks for your comment and happy holidays!

yongjik•1mo ago

> sophisticated curve-fitting (numerology with constraints)

lol ChatGPT feeling sassy today, though I think it was well deserved.

albert_roca•1mo ago

Undefined/non-consensual prompt.

bigyabai•1mo ago

Nah, ChatGPT cooking you on this one. You're lucky it didn't call it gematria.

fatbrowndog•1mo ago

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...

albert_roca•1mo ago

- Why 4? It's not random. It is derived from the structural constant w = 2 as a topological constraint of the three-dimensional topology. Radius scales as w^2 = 4.

- Why tetrahedron? Mass is defined as volume. The tetrahedron is the simplest closed 3D volume. Mathematically, the derived proton radius corresponds to the exact geometric circumradius (edge · √6 / 4) of this volumetric structure.

- Why α / 4 · π? It represents the linear interaction cost (α) distributed over the spherical solid angle (4 · π) of the protonic surface.

- Incorrect QED terms? The model explicitly and intentionally diverges from QED. It doesn't treat particles as points, but as three-dimensional objects. The model excludes the notion of physical infinities or singularities.

- Why α^2 / 12? It derives from nodal friction distributed over the 12 vertices of the lepton's icosahedral topology.

- Why α^3/5? It derives from the local 5-fold symmetry of the icosahedral node.

The criticisms fail to identify that the model presents a first-principles framework where these numbers are geometric consequences, not free parameters. The model is not intended to be orthodox, but mathematically and geometrically coherent.

Looking for 4 Autistic Co-Founders for AI Startup (Equity-Based)

AI-native capabilities, a new API Catalog, and updated plans and pricing

What changed in tech from 2010 to 2020?

From Human Ergonomics to Agent Ergonomics

Advanced Inertial Reference Sphere

Toyota Developing a Console-Grade, Open-Source Game Engine with Flutter and Dart

Typing for Love or Money: The Hidden Labor Behind Modern Literary Masterpieces

Show HN: A longitudinal health record built from fragmented medical data

CoreWeave's $30B Bet on GPU Market Infrastructure

Creating and Hosting a Static Website on Cloudflare for Free

"The Stanford scam proves America is becoming a nation of grifters"

Elon Musk on Space GPUs, AI, Optimus, and His Manufacturing Method

X (Twitter) is back with a new X API Pay-Per-Use model

Zlob.h 100% POSIX and glibc compatible globbing lib that is faste and better

Show HN: Deterministic signal triangulation using a fixed .72% variance constant

Scientists Discover Levitating Time Crystals You Can Hold, Defy Newton’s 3rd Law

When Michelangelo Met Titian

Solving NYT Pips with DLX

Baldur's Gate to be turned into TV series – without the game's developers

Interview with 'Just use a VPS' bro (OpenClaw version) [video]

EchoJEPA: Latent Predictive Foundation Model for Echocardiography

Disablling Go Telemetry

Effective Nihilism

The UK government didn't want you to see this report on ecosystem collapse

No 10 blocks report on impact of rainforest collapse on food prices

Seedance 2.0 Is Coming

Show HN: Fitspire – a simple 5-minute workout app for busy people (iOS)

Dexterous robotic hands: 2009 – 2014 – 2025

Interop 2025: A Year of Convergence

JobArena – Human Intuition vs. Artificial Intelligence

Looking for 4 Autistic Co-Founders for AI Startup (Equity-Based)

AI-native capabilities, a new API Catalog, and updated plans and pricing

What changed in tech from 2010 to 2020?

From Human Ergonomics to Agent Ergonomics

Advanced Inertial Reference Sphere

Toyota Developing a Console-Grade, Open-Source Game Engine with Flutter and Dart

Typing for Love or Money: The Hidden Labor Behind Modern Literary Masterpieces

Show HN: A longitudinal health record built from fragmented medical data

CoreWeave's $30B Bet on GPU Market Infrastructure

Creating and Hosting a Static Website on Cloudflare for Free

"The Stanford scam proves America is becoming a nation of grifters"

Elon Musk on Space GPUs, AI, Optimus, and His Manufacturing Method

X (Twitter) is back with a new X API Pay-Per-Use model

Zlob.h 100% POSIX and glibc compatible globbing lib that is faste and better

Show HN: Deterministic signal triangulation using a fixed .72% variance constant

Scientists Discover Levitating Time Crystals You Can Hold, Defy Newton’s 3rd Law

When Michelangelo Met Titian

Solving NYT Pips with DLX

Baldur's Gate to be turned into TV series – without the game's developers

Interview with 'Just use a VPS' bro (OpenClaw version) [video]

EchoJEPA: Latent Predictive Foundation Model for Echocardiography

Disablling Go Telemetry

Effective Nihilism

The UK government didn't want you to see this report on ecosystem collapse

No 10 blocks report on impact of rainforest collapse on food prices

Seedance 2.0 Is Coming

Show HN: Fitspire – a simple 5-minute workout app for busy people (iOS)

Dexterous robotic hands: 2009 – 2014 – 2025

Interop 2025: A Year of Convergence

JobArena – Human Intuition vs. Artificial Intelligence

Is the Standard Model overfitting or am I curve-fitting?

Comments