[palata]

This.

We need to think about the data we need to store before we store it, only store the data that we need to store, and only store it for as long as needed.

It reminds me of CIs. It's now so easy to throw 40 jobs on GitHub actions that people don't think about them. I have been in a startup where people would debug in CI: they wouldn't have e.g. Windows on their machine (maybe they should have, given that their product was supposed to run there) and were fixing compilation issues by sending patches and patches to the CI. Every single time it would trigger the 40 jobs. Sometimes you could see a patch sent every 5 min for 3 days (where reproducing the issue locally would actually take 3s and not 5min). They did not even bother disabling the 39 uninteresting jobs.

For open source projects, it's just wasted energy, for private repos it was costing the company a lot. This was just malpractice. But nobody cared. The finance person would say "GitHub is expensive", the CEO that "well we need it" and the engineers that "I don't want that Windows crap on my computer", I suppose.