The Book I Promised Not to Open
A chemistry set, a middle school obsession, and the forty-year bet I just won.
There’s a book on my shelf that I haven’t opened, as of the writing of this article.
It’s a comic book, technically. La storia della chimica a fumetti... “The History of Chemistry in Comic Book Form”... by Luca Novelli and Cinzia Ghigliano. Published in the early eighties. Illustrated panels walking you from Democritus to Mendeleev, from alchemists burning their eyebrows off to the elegant logic of the periodic table. I got it when I was eleven or twelve, somewhere near Rome, and I must have read it a hundred times.
When I left Italy in my early twenties, the book stayed behind. At my mother’s place, on a shelf I assumed would always be there. But mothers move. Boxes get shuffled. Things disappear. At some point, the book was gone. Not thrown away... just lost in the chaos of life happening to other people’s shelves.
I didn’t think about it for years. Then, last year, I found the same edition on eBay. Same publisher. Same year. Same cover with the slightly off-register printing that Italian paperbacks always had. I bought it immediately. Didn’t hesitate. Didn’t haggle.
It arrived in Sardinia a week later. I put it on my shelf. And I didn’t open it.
That was deliberate.
I’ll tell you why.
The Kid With the Chemistry Set
If you grew up in Italy in the late eighties, you might remember Il Piccolo Chimico. The Little Chemist. It was this beautiful chemistry set... test tubes, litmus paper, a little alcohol burner, reagents in tiny bottles with labels you couldn’t quite read. Every kid who was even slightly curious about science wanted one. I got mine and basically disappeared into it.
I mixed everything with everything. Not because I had a hypothesis. Because I wanted to see what would happen. Colour changes, precipitates, things that fizzed, things that smelled terrible, things my mother told me to take outside immediately.
And at some point... I stumbled on a combination that instantly oxidized metal. I don’t even remember exactly what I mixed. But I remember my teacher’s face. Jaw on the floor.
“Come hai fatto?” How did you do that?
I didn’t know. I was just playing.
Around the same time, the coding obsession kicked in. I’d been programming on an MSX since I was ten... the little Japanese home computer that was everywhere in Italy back then. Monochrome green screen, BASIC, cursor blinking at you like it was waiting for something important. At twelve, I wrote a program that would take any element symbol and return the full name. Type “Fe,” get “Iron.” Type “Au,” get “Gold.” Every element in the periodic table, stored in an array.
My science teacher had me haul the MSX to school to demonstrate it. I remember carrying the CRT monitor through the corridors of a building that hadn’t been renovated since the sixties, plugging it in, the green glow lighting up a room full of kids who’d never touched a computer. The school didn’t have one. Most of them didn’t have one at home. Most them thought I was a crazy nerd…. (they were right, I still am!).
It was a trivial program. But something about it stuck. The idea that chemistry could live inside a machine. That the structure of elements, the logic of the periodic table, the patterns I saw in that comic book... all of it could be encoded. Queried. Computed.
That idea never left me.
The Promise
Fast forward a few decades. Through an illustration career with Marvel and DC. Through building a tech company in Vermont. Through moving back to Sardinia with my family. Through all of it, the chemistry question kept nagging.
Not just “what are the elements.” The deeper question. Why do bonds form the way they do? Why is a carbon-hydrogen bond exactly the length it is? Why does an SN2 reaction proceed and an E1 doesn’t, given the same starting materials? What’s the geometry underneath?
When I started working on the physics framework that would eventually become FluxMateria, something shifted. I wasn’t just curious anymore. I was close. The math was producing bond lengths from first principles. The mechanism predictions were matching experiment. The numbers were coming out right... not approximately right, not “within acceptable error for a first attempt” right, but sub-percent accuracy across hundreds of bonds right.
And I looked at that book on the shelf... the same edition, found and recovered after all those years... and I told myself: I’m not opening it until I’ve earned it. Until I’ve actually solved this. Until the kid who read the original a hundred times can point to something real and say, “I did what the book made me dream about.”
So it sat there. Closed. While I worked.
What “Solving Chemistry” Looks Like
Here’s what the engine can do now. The engine that started with a twelve-year-old’s periodic table program and ended up... here.
Bond lengths. 453 bonds validated across 64 elements. Single, double, triple, aromatic. Mean error: 0.079%. For hydrogen, the engine predicts 74.13 picometers. Experiment measures 74.14. The deviation is smaller than most instruments can reliably detect. Zero fitted parameters. Pure geometry.
Bond energies. 908 bonds. 0.289% mean error. 870 out of 906 within 1% of experimental values. Every single bond, double bond, and triple bond across the p-block, d-block, and s-block. Computed in under a millisecond each.
Reaction mechanisms. 336 out of 336 experimental test cases classified correctly. SN1, SN2, E1, E2, E1cb. 100% accuracy. With activation barriers at 7.4 kJ/mol mean absolute error. The engine doesn’t just tell you which mechanism dominates... it evaluates all competing pathways simultaneously and shows you why.
Synthesis planning. 29 reaction types. 3.1% barrier MAE. 200 specific reactions validated, 72 exact matches. Retrosynthetic analysis with physics-derived barriers, not database lookups. Under 50 milliseconds per plan.
Spectroscopy. UV-Vis at 6.2% error, IR under 1%, NMR at 0.3 to 0.5 ppm. Predict the spectrum before you synthesize the compound. About 25 milliseconds per prediction.
Solvation. 0.06 logS MAE on nearly 10,000 compounds... currently the top result in the field. Real molecules exist in solution, not in gas-phase textbook problems. The engine handles that.
All of this from the same computational kernel. Same physics. Same input, same version, same output. Deterministic. Reproducible. Auditable. No neural network. No training data. No fitted parameters.
The twelve-year-old with the MSX would lose his mind trying the Interactive Demos.
What Changed
I’ll tell you what the kid with Il Piccolo Chimico didn’t know, mixing reagents on his desk near Rome.
He didn’t know that the entire field of computational chemistry would grow up around a sixty-year-old equation that’s too expensive to solve. That density functional theory would become the workhorse of the field and also its bottleneck. That a single DFT calculation on a modest molecule would still take hours in 2026, and that teams would screen 200 compounds when they needed to screen 200,000, because the compute budget ran out before the science did.
He didn’t know that the “pattern” he saw in the periodic table wasn’t just a lookup. It was geometry. That the bond lengths he would eventually compute come from the same geometric kernel that determines everything from electron mass to the gravitational constant. That the reason chemistry works the way it does is the same reason physics works the way it does... same structure, same math, same single principle, all the way down.
He definitely didn’t know that he’d end up building a platform that computes all of this at 3.6 million times the speed of DFT, from a studio in the same Sardinia where he was born.
Ma vedi come gira il mondo. You see how the world turns.
Today, the Book Opens
I’m publishing this article the day after FluxMateria went live.
The chemistry modules are up. The bond length engine. The mechanism discovery system. The solvation models. The spectroscopy predictions. The synthesis planning tools. The benchmarks are published. The demos are live. You can paste a SMILES string right now and get results.
And today, I’m opening the book.
Not because the work is finished. It’s not. There are modules still in beta, edge cases to sharpen, validation campaigns to expand. The solvation work has more solvents to cover. The spectroscopy benchmarks need more molecules. I know this.
But the promise was never about perfection. It was about showing up. About the kid with the chemistry set and the green-screen computer earning the right to say: I didn’t just memorize the periodic table. I computed it. From scratch. From geometry. And it works.
So the book is open. The pages are yellow. The illustrations are exactly as I remember them... Democritus wondering what happens when you keep cutting something in half, Lavoisier measuring gases, Mendeleev’s table slowly taking shape.
And now there’s a new chapter. One that isn’t in the book.
But maybe it should be.
The comic book is open on my desk as I write this. The page it fell open to is about Dalton’s atomic theory. Fitting.
The kid who read it a hundred times finally did something about it.
OK. I need to tell you something. I wrote everything above, including the line about Dalton, before I touched the book. It was a guess. A narrative choice. I thought it would make a nice ending.
Then I walked over to the shelf, picked up the book for the first time in years, and opened it to a random page. Pure instinct. No peeking.
The first word, top left: “Dalton.”
I stared at it for about ten seconds. Then I started laughing like a lunatic. My wife heard me from the other room and knew immediately that something very Roberto had just happened.
I don’t have an explanation for this. I’m a physicist, not a mystic. But I do have the photograph. And I have a theory about the universe that says geometry determines everything.
Maybe it determines Substack articles too.
Stiamo a vedere.
About the Author
Roberto Campus is the creator of FLUX Theory and architect of FluxMateria. Born in Sardinia, raised in Rome, he’s spent over 35 years asking stubborn questions about how the universe works... and recently started getting answers. He’s either onto something profound or needs a better hobby. The benchmarks suggest the former.
FluxMateria: fluxmateria.com · FLUX Theory: thefluxtheory.org