What are we
not seeing?
I keep coming back to galaxy rotation: what we can measure, what the models predict, and what a correlation can actually tell us.
EXPERIMENTS, MODELS, AND RESEARCH
My G500 has eight cylinders and sixteen spark plugs. You can take the V8 apart and see where each plug sits.
01 / FOLLOW A THREAD
I keep coming back to galaxy rotation: what we can measure, what the models predict, and what a correlation can actually tell us.
Light can look very different depending on how you’re moving. What does the observer actually measure? How does that change around a black hole?
STILL FIGURING THIS ONE OUTI want to see what a smaller model can do on hardware I can run myself. Then I can try changing it and find out what’s worth the memory and time.
Playing trumpet is one way into this. Harmonics and resonance are another. I want to understand what gives a sound its shape.
Russian, Mandarin, pronunciation, and all the times a sentence is technically correct but sounds strange. I want to know how people actually say it.
STILL FIGURING THIS ONE OUTThere’s an electric motor in the back of my G500 that opens the sunroof over the front seats. A cable runs the length of the cabin. You can follow it here.
Make a little world, give it a few rules, and see what happens when you play. Or just try to beat your last run.
A tin of white beans, a lemon, and some feta can carry dinner. I keep that recipe here alongside kimchi fried rice and garlic mushrooms on toast.
A model can get the label right and still be wrong about how sure it is. These three experiments look at that, prediction intervals, and missing data.
02 / THE CURIOSITY PLAYGROUND
There’s a galaxy to mess with, a car to work on, a beat to make, and probably something for dinner. Pick one.
Compare the orbits with and without a halo. Launch a tracer and see where it goes.
OPEN THE GALAXY ↗Follow the G500’s exhaust from either cylinder bank, open a Porsche crankcase, or trace a trawler’s shaft from the engine to the propeller.
EXPLORE THE COLLECTION ↗Fly through the rings, find your way out of the maze, or see how high you can stack.
ENTER THE ARCADE ↗Eight pantry recipes, ingredients to pick from, and a shopping list for what’s missing.
MAKE SOMETHING ↗Make a beat, add a melody, and mess with the sound. Save the loop if you like it.
MAKE A LOOP ↗Pack model weights into a memory crate and try a few tradeoffs. See how much room the weights, cache, and reserve take up.
PACK THE CRATE ↗Three medical-AI experiments with actual held-out results. Explore the errors, the uncertainty, and the assumptions behind the scores.
OPEN THE MEDICAL AI LAB ↗THE SOURCE IS HERE TOO
The six browser projects come with setup notes and tests. Run them locally and mess with them. The medical lab and dark-matter paper have their own reproducible research downloads.
03 / WORKING PAPERS
A new dark-matter synthesis and computational study, alongside the earlier SPARC notes. Each includes methods, sources, and limitations. These are working papers, not peer-reviewed publications.
One dark-matter candidate at 248 keV, about fifty papers explaining it. Load any of 24 published explanations and change the assumptions behind it.
24 BENCHMARKS · 14 PAPERS · COMPUTED IN YOUR BROWSERMechanisms, new constraints, and a fresh analysis of 100 galaxies. What can their rotation curves actually tell us?
100 GALAXIES · 2,742 MEASUREMENTS · REPRODUCIBLE COMPUTATIONA multivariable SPARC pilot: testing whether extra galaxy properties add predictive value beyond baryonic acceleration.
149 GALAXIES · 2,700 RADIAL MEASUREMENTS · SOURCE-BASED DRAFTWhat would a next test need to measure? Lensing, gas physics, and the challenge of moving between astrophysical systems.
PROPOSED TESTS · NO NEW MEASUREMENTSThe pilot’s results are reproduced from the saved research note, not independently recomputed here. The next study is a proposal, not a claim that dark matter has been ruled out.
Read the original pilot note ↓WHAT ELSE SHOULD WE TRY?