How a data project turned into a book
I want to tell you how something I started as a dry data project turned into a book I now care about more than almost anything I've worked on.
It began with mine safety records. In the United States, whenever a miner is killed on the job, the government investigates and writes a report. There are hundreds of these reports, going back decades. They are full of useful information, but they are hard to read and scattered all over the place, so I started building a way to organize them — to pull out the facts cleanly and turn them into clear, reliable summaries. Honestly, it started as a business idea: something I could sell to law firms and insurers.
To test whether my summaries were any good, I picked a few real cases to work through. One of them was the Jim Walter No. 5 mine disaster in Brookwood, Alabama, in September 2001.
An explosion went off deep underground and trapped one miner. Twelve of his coworkers went toward him. Fifty-five minutes later, after a second, bigger explosion, thirteen men were dead.
It happened twelve days after September 11th, and the whole country was looking somewhere else. Fifteen hundred people came to the funeral. The country never did. In twenty-five years there had been no book, no film and no documentary about that day.
The more I read, the more I couldn't let it stay just a data entry. These weren't statistics. They were thirteen men who went to work on a Sunday afternoon, and twelve of them died going toward a man who was hurt. That is not a report. That is a story someone should tell.
How the book is written
I wrote it like a real story, not a report — so that a stranger who never heard of Brookwood might read it and come to know who these men were. But everything about what actually happened comes from the official records. I do not invent dialogue, and I do not claim to know what any man thought or felt.
Building it was slower and more careful than I expected, and that was on purpose. I started the way I know how: with the official government records, the mine-safety reports, the investigation findings and the court decisions, pulling the hard facts out of them — how deep the mine was, how the air moved through it, what the investigators found, who was where. Then I did the same with old newspaper stories and the published accounts of people who were there: survivors, rescuers, families.
I learned quickly to be careful with all of it. A headline I found online was not the same as an article I had actually read. A quick summary was not the same as a checked fact. So I built myself a way to keep score — to mark every claim in the book and check it against a real source before it was allowed to stay. If something could not be traced back to a record, it either got fixed or it came out.
I would rather the book be smaller and true than bigger and guessed-at. These men deserve to have it right.
A note on method and tools
This book was researched with the help of software I built myself, including AI tools used to extract, sort and organize the facts in hundreds of pages of government reports and court records. An AI writing tool also drafted prose, working to my outline and my rules, under my direction and editing. The book carries its own page saying so. None of these tools decided what is true.
Every factual claim was checked against the record before it was allowed to stay. Where the records disagree, the book says so — and in places it gives both accounts rather than choosing between them. Where the record is silent, the book stays silent rather than filling the gap. Responsibility for every fact, judgment and omission is mine.