The worst data cleanup I’ve ever had to do, nobody handed me. I built it, and I built most of it before 2007, before I was a data professional, back when a photo was just a photo and not a file that would one day need to be found.
This weekend I finally hauled the whole archive together. Thousands of images, going back to 1999, pulled off an external drive and loaded into one place. As they came across, the gap showed up. For a huge chunk of them the dates are wrong, the locations are gone, and the only thing telling me when a picture was taken is a folder name I typed twenty years ago and can’t fully trust.
My son’s first birthday, July 2009, sorts into 2012. Three years after the cake.
And that one I can’t blame on not knowing better. By his first birthday I’d been working in data for almost two years.
How it happened
Back then, Windows drove everything. It built the folders and decided how your pictures were organized, and it mostly worked until you touched anything.
Copy a batch to a thumb drive, or drag it into a new folder on another drive, and Windows gave every copy a fresh Date Created. A copy isn’t the original, and the filesystem treats it as a new file born that day. A lot of tools sorted by that date. The real capture date was usually still buried inside the file, but nothing I was using bothered to read it.
So I lost provenance without touching a single pixel. Every photo was pristine. The record of when it was taken eroded every time I tried to take care of it.
There’s a difference between the data and the data about the data, and the metadata you can see isn’t always the metadata that’s true. The field literally named “created” was the one lying to me. A wrong date is worse than a missing one, because a wrong date looks authoritative.
What changed
Eventually the tools caught up. Phones kept their clocks right without anyone setting them. Location came along without anyone asking. Photo apps started sorting by when a picture was taken instead of when the file showed up on a drive. The truth about each photo traveled with it and actually got used.
So the dividing line in my archive isn’t before and after I became a data professional. It’s before and after I started shooting on a phone. Everything earlier needs work. Everything since mostly sorted itself out.
Defaults
That’s the whole history of data management, and my own photo library tells it better than any work example I have.
Nobody tags their memories faithfully for a decade. I didn’t, and I’m the guy who lectures people about it. What finally worked wasn’t discipline. It was defaults: tools that keep the truth about a file whether anyone remembers to care or not.
Every business user I’ve ever trained is in the same spot I was in, trusting the date on the screen because the screen showed it to them. They aren’t careless. Their tools never told them the date had changed.
