Thing is, if they have backups, even editing data doesn't do anything. Or they could even just have it set up to only display the most recent version but still keep each edit on the db. Wouldn't even be hard to implement. Hell, it wouldn't even be that hard to implement a historical series of diffs so they don't have to store the full comments for each edit if the edit is a small one.

Like if I wanted to run a service that made it easier to find interesting data, part of that would be to flag deletes and edits as "whatever was there before has a higher chance of being interesting".

Once something is posted, IMO just assume that it can't be unposted and trying to unpost it might work similarly to the Streisand effect.

Even here. Sure, the source is open and I'd bet looking at the delete and edit functions would make it look like everything is fine. But other federated servers don't have to run the same code and can react to delete and edit directives from other servers however they want. The main difference between this platform and Reddit in regards to control over posted information is the fediverse can't prevent entities from accessing the data for free (albeit with less user metadata like IP and email).