Cached pages, archive.org, and the limits of "delete".

To understand what "delete" can do, you have to know where a comment actually exists:

1. On Reddit - the live copy, the one you control.
2. In caches - search engines, Reddit mirrors like Reveddit, and scrapers that keep their own copies.
3. In archives - the Wayback Machine and similar, which snapshot pages and keep them more or less forever.

A delete acts on the first. The other two are the honest part of this conversation.

Deleting removes the item from Reddit: your profile, the thread, the API. Karmdit then runs its 7-day verification to confirm it's actually gone from public view, because Reddit sometimes lets a deleted item linger. For the live copy, delete works.

Delete is total on Reddit. Everywhere else, it's a probability.

If a scraper or archive captured your comment before you deleted it, that copy is theirs. Reddit-side deletion can't reach into a third party's database, and no honest tool can promise otherwise. Anyone who claims to "erase you from the internet" is selling you something.

The good news: most content is never archived at all. Archives and scrapers are selective - they favor popular threads and things that changed. The average old comment simply isn't worth anyone's storage.

This is where overwrite-before-delete earns its place. By replacing the body with a placeholder before deleting, Karmdit changes what a late or edit-triggered scraper can capture: the placeholder, not your words. Many scrapers snapshot on edit - so overwriting first turns their own trigger against them.

It doesn't rewrite history that's already archived. It does mean that from the overwrite onward, the recoverable version is meaningless.

Here's the honest standard to aim for: not "erased from all of human memory," which no one can deliver, but "not findable by a normal person doing a normal search." That bar is very achievable - delete the live copy, overwrite so caches degrade, verify it's gone.

If you're facing a determined, resourced adversary with pre-existing archives, that's a different and much harder problem, and you should know it. For everyone else, this is enough.