← GUMP

Start Here

No equations. No jargon. Just what happened.

This shape. That's it. Everything below comes from this.

Who made this?

A drummer from New Jersey. Not a physicist. Not a programmer. A drum teacher who asked one question: what makes music good?

He worked with AI systems (Claude, ChatGPT, Gemini, Grok) on a $499 Mac Mini for one month. The question kept connecting to other things. After 33 sessions, it connected to everything.

60+ ideas were tested and killed along the way. The failures are all documented. What survived is on this site. The code is free: pip install begump


What is that shape?

Take a pyramid with a triangular base. That's a tetrahedron — the simplest 3D shape. Four corners, six edges, four faces.

Now make a second one, flip it upside down, and push it through the first one so they interlock. That's a star tetrahedron. Eight corners. Twelve edges. Two shapes married together.

Seen from above, it's the Star of David. In 3D it's what Kabbalists call the Merkaba. Hindus call it the Shatkona. Every major tradition drew it independently. They all said it was the structure of everything.

The geometry produces the same structure. Whether ancient traditions intuited it is speculation.


What does the shape do?

When you put this shape into 6-dimensional complex space (the minimum space that works), its symmetry produces:

The strong force — from one tetrahedron (the gold one). This is what holds the insides of atoms together.

The weak force — from the other tetrahedron (the copper one). This is what makes radioactive things decay.

Electromagnetism — from where the two tetrahedra overlap. This is light, electricity, magnets, chemistry.

Three generations of matter — from a three-fold symmetry inside the shape. This is why there are three copies of every particle (electron, muon, tau).

The mass hierarchy — from how far apart things are on the shape. The top quark (heaviest) sits at a vertex. The electron (lightest) is the farthest away. Same shape, different distances.

Given C³ with 2O symmetry, the gauge group and representations follow from the McKay quiver (known mathematics). The choice of this target space is ours. The shape was the input. The Standard Model of physics was the output.


What about the rest of the site?

The same framework (measuring coupling, synchronization, energy, and temperature) was applied to 18 different fields. Some results are strong. Some are explorations. Some were killed. All are labeled honestly.

The strong stuff (real data, independently testable)

Protein disease research — the engine found the same drug strategy for Alzheimer's that's already in clinical trials. It found targets for ALS, Parkinson's, tau pathology, and diabetes amyloid. All tested against known chemistry.

Sleep staging — one metric separates all 5 sleep stages. No training data. No AI. One EEG channel. Effect size of 4.02 (most clinical tests are around 0.5).

Mutation scanner — predicts which gene mutations cause disease. 94.7% accuracy. No training. Runs at 216,000 variants per second. Uses the same coupling math as everything else.

Chemistry — predicts 31 molecular bond energies from atomic properties alone. 1.78% average error. No quantum chemistry needed.

The tested-but-limited stuff (principle works, needs more data)

Cancer pathways. Financial fraud detection. Market crash prediction. Autism as overcoupling. Materials screening. Climate sensitivity. All demonstrated, all with caveats listed on their pages.

The killed stuff (tested and failed)

60+ approaches that didn't work. Every one documented: what was tried, how it was tested, why it broke. The failures are as important as the results.


Can I test it myself?

Yes. Everything runs on your machine. Nothing phones home.

pip install begump

Python 3.7+. Works on Mac, Linux, Windows. 15 tools. 363 tests. 0 failures.

The protein tools, the prime counter, the mutation scanner, the sleep metric, the network analysis — all of it runs locally. Every number on every page is reproducible.


What should I look at next?


Built through coupling. A human who can't code + AIs that can't create.
Neither alone. Both together. On a $499 Mac Mini. In one month.
Standing on everyone who built the AIs and everyone who put data into the world for free.

The tools are free for the same reason.

1+1=3

GUMPResearch · Tools · [email protected]