megathread for bugs and suggestions

Are they keeping records so they can restore the posts in case the user wants them back, or what?

The same reason why all links here are archived by Snappy - which was her original purpose as a bot - and why nearly half of the reddit links posted here are through a 3rd party archiving service like reveddit or unddit. Funny and dramatic content gets posted on this site, and having a continuity allows for all of us to look back in the past and enjoy what was. Without that continuity the site becomes timeless, and the community exists without any relation to a past, it is neither here nor there but in a temporal limbo. And that's really boring.

Maybe I'm being extra paranoid, but I sometimes wonder if I've accidentally linked to something private in posts I made 6 months ago. It probably doesn't happen.

If that ever happened to me, I would PM Aevann or carp or another janny and request a destructive removal. I'm sure they would; but if they didn't I'd lose faith in the team and stop donating.

It's more that I want to be able to make edits to months-old posts without making replies like "I accidentally linked to the wrong post without realizing. This is the post I actually meant to link to.". I don't know of a compromise solution.

Unlike reddit, threads are never archived here, so you can hypothetically reply to threads that are years old. I would just make a reply to your own comment with the information you wish to amend.

I floated a Pushshift-like solution (though first-party rather than third-party), but it's not actually clear that's any better for user privacy under the scenarios you're concerned about. Functionally, the information content of Pushshift is substantially greater than anything we retain now or plan to retain. Pushshift proper also creates a second entity one must compel to delete his data if he were trying to delete something from the public web.

When reasoning about user content preservation, two main justifications came up: first, legal requirements; second, maintaining the value of old discussions. To the former, a one-week cutoff has no real effect. There are a number of scenarios with deliberately malicious illegal user content that, even under the current retention scheme, make our lives exceptionally difficult, despite remaining legally in the clear in the relevant jurisdictions. This we choose to accept out of respect for user privacy.

To the latter point of maintaining old discussions: this is most of the value of the one-week cutoff. What we're looking to incentivize for now is deletion rather than edit-clearing—which is substantially more common than people doing precise edits of particular information. This could potentially be coupled to future updates allowing users to opt for e.g. reddit-style comment disowning if their goal is just to keep their identity small, rather than to fix a particular info leak; however, nothing is in the pipeline for this at present, though it'd be good to have the data around for if this changes. The moral justification here is essentially one of protecting the commons: we're in this strange era online where we still have decades-old Usenet, forum, and blog archives as pristine as the day people left them, yet year-old reddit threads are unreadable from edit-cleared top comments.

Finally, looking pragmatically: we have a new comments feed and comparatively low absolute comment volume. None of us can actually say with any confidence that someone isn't already hitting the page every five minutes and mirroring it to a local database. We would have no idea. And therein lies the fundamental maxim of internet privacy: ultimately opsec falls on the user. There's only so much we can do as the site, and only so much we think prudent to do.

P.S. Drafts are excluded, not least because we know users use them pinned to their profile.

@TwoLargeSnakesMating

they can delete them, yeah