[tesin]

While I agree with this as a general rule, it ignores an entire class of problem. I work in a data preservation space - if we let people delete things every time they wanted to, it'd be a disaster. Instead we 'tombstone' it (hide from everyone but admin), then let someone with some space from the process confirm it actually should be deleted. We've averted dozens of disasters this way.

I'd argue the user must feel in control, as you say, but not necessarily _be_ in control.

[temporallobe]

Same here. We implemented a “soft delete” in a system I worked on recently, as actually deleting something could have legal implications. We just have a table column where we mark it as deleted and then exclude those from any retrieval queries. If we ever need to undelete, someone can manually update the database.

[prox]

Well, it ties in with another rule I have, don’t allow for stupidity. If a user can do some stupid things, it’s not designed well. (Say your deletion problem)

But yes, as long as these decisions are made consciously by a team, it’s get the attention it deserves.

[ryandrake]

I don’t know—this kind of deception is how the industry attracts more regulation. Delete needs to go back to meaning delete, and the user should be in actual control.

[renewiltord]

No. I frequently undelete or unsend. Others do too. Removing the Windows Recycle Bin or its equivalent would not make my life better by "giving me agency".