PSA: Don't Trust Meta's Data Export Feature

tl;dr: Both Facebook and Instagram give you back corrupted and/or incomplete data exports. Check and verify every single byte of what you have received, or suffer data loss. (Guess which one happened to me?)

I decided to delete my Instagram account a couple of months ago, when they tried to pull the privacy-hostile (and probably very illegal) Pay or Okay manoeuvre. Asking me to pay 15$/month for my fundamental right is extortion, and I don't negotiate with privacy terrorists.

Before I deleted my account, I used Meta's data export tool (which, under GDPR, they have to provide), so that I have a copy of my data. The conversations I've had, things I've posted, that sort of stuff. I downloaded the multi-gigabyte file, and did a quick check whether it contained things. I verified that my oldest post, my newest post, and a few conversations were included in the archive. So, I went ahead and requested the deletion of my account.

A few weeks after my account was deleted, I needed some information from the data export: I needed the title of a track I had posted on Instagram as part of jamuary 2021. So I unzipped the file, and dove into the directory.

The first thing I've noticed was the lack of proper structure. Some videos I have posted were filed under posts, some under igtv, and some were under reels. Some videos had the proper .mp4 extension, while some did not have any extension at all.

But that's not what I was after, so I ignored this smell and continued digging. Found the relevant json file that contained metadata for my videos. It was easy to find, because I had verified that this file contained metadata for my oldest post, and my newest post.

Well, apparently that level of verification was not enough. I quickly discovered that my data export did not contain any of the titles I have submitted alongside my videos between 2020-01-18 and 2022-01-02. I asked for my data, and all I got was a donut, with a gaping hole in the middle.

Data loss. Data I had handed over to Instagram, who presumably employs very capable engineers, whose abilities are more than enough to not fuck this up. Data that had turned the gears of the giant ad machine and generated profit.

I rushed to check my data export from Facebook, which I had generated a couple of years ago, when I deleted my account. It was missing a lot of data, and a mysterious file that has the .zip.enc extension, which, apparently, is encrypted and can't be opened by the data's owner. And my account is now long gone.

More data loss. Data that Meta is legally bound to give back to me. Data that was apparently not important enough to export correctly. Data that had turned the gears of the giant ad machine and generated profit.

So, the lesson for me was to accept the data loss, and augment it with a financial loss by getting a decent NAS and a couple of giant hard disks.

My advice -- if you're tech-savvy enough, host the source of truth or a mirror yourself. Write scripts to sync the data, check the integrity of every single byte of data exports, complain loudly on discrepancies.

If you're not tech-savvy enough, ask your friendly neighbourhood nerd for help. Hope it works out for you.