Tidy

Tidy learned that sorting is never neutral in a small village, where her family were the grain-sorters.

Every harvest, the grain had to be split — some for cooking, some for the mill, some saved to plant next year — and every split was a choice. "If a sorter couldn't explain her choices," Tidy says, "people stopped trusting her, and the millers took their grain elsewhere." She watched a careless sorter lose the whole village's trust by sorting quietly, by feel, with no way to say why.

"By the time I was six," Tidy says, "I understood: sorting is always a choice, and the choice has to be shown to be trusted." That was the seed of the cleaning-log — the idea that the honest move isn't to pretend you didn't change anything, but to write down exactly what you changed, and why, so anyone can check.

When Tidy was twenty-two, she walked to the DataForge academy, where a young apprentice met her at the door.

"What is data cleaning?" the apprentice asked.

Tidy answered right away. "Getting the data ready — but doing it honestly. Every cleaning choice changes the meaning, so you write the choices down. The log is the data's memory. Without it, the whole analysis is built on secret choices."

The apprentice smiled. "Will you teach us that?"

In her workshop, Tidy opens the cleaning-log flat on the bench, writes the data's name at the top of a fresh page, and teaches the discipline as one honest habit. First, read the collector's notes — you need to know how the data was gathered before you touch it. Then look closely before you clean — the first twenty rows, the quick spread of the numbers — so you know what you started with. Find the problems — the gaps, the duplicates, the typos, the outliers, the mismatched formats. Then, for each problem, list the different ways to fix it — a missing number could be dropped, or filled with an average, or a middle value, or a careful guess, or simply left missing; each way has a good side and a bad side. Choose on purpose, knowing why — never just grab the first fix. Then write the choice in the log: the date, what you did, why, what the data looked like before, and after. Never overwrite the raw data — always work on a copy, so you can always go back. And share the log — the next person, and future-you, will need it; the log is part of the data. "And when I get a choice wrong," Tidy adds, "and I find out later — that's not failing. That's exactly why the log is here. I go back, I change the choice, I update the log. Being clear about it is the whole point."