With email, I re-read drafts twice before sending. But with social media I'm used to proper delete handling by almost the entire fediverse (which is all HTTP), and it's difficult to override that muscle memory for Nostr.
Discussion
HTTP doesn't inherently have delete though. If you visit my web page, I can't change the code that was sent to your machine. I can write code to update content on the fly, but that's not specific to the protocol.
I see the "delete this post" feature as a fear-driven action that encourages irrisponsible behavior. In some environments, like Telegram, it changes the context and meaning of conversations that are read by others later.
I know that I make mistakes, and I have no problem apologizing for them when I realize I'm in the wrong. It seems like other people have gotten used to hiding, and not owning up to their posts/actions.
You can choose a client that only indicates that an author wants to retract a post, and they can also provide a reason for the deletion. This is not about owning up to posts or mistakes, but about fixing typos, removing tracking params in links, and stuff like that. Since, as you say, you cannot actually delete a post that has been published and synced outside of your control, the fear about people hiding things is unwarranted really.
Also very simple, and implemented like this in many XMPP clients: only allow edits if they happen shortly after the original message. Later, just show it as a separate new message, and discourage doing it in the first place with client-side UI preventing you from editing too late.
I'm not sure what environment you're referring to.
The idea of deletion changing context and/or meaning is warranted, since it actually happens.
Since you can't choose other people's clients, or what relays your note will hit, the concept of editing/deleting is really a facade.
Again, it's only about deletion *requests*, i.e. showing intent. Why are you painting it as if anyone claims proper deletions are possible? My very post that you replied to explicitly says "requests", and I haven't seen anyone saying anything else about the feature.
It already works like this on other decentralized protocols, with no demonstrable bad side effects that would outweigh the usefulness. If you want to insist that it's such a big problem, then please point us to good examples of that on other decentralized protocols, or concede that it's pure speculation. Also, you should not use any clients implementing NIP-09 (or not letting you turn it off), since that's your personal choice anyway.
I see now. If you don't mind that it's not deleted/edited for others, and only care about the display of desire, then I don't see anything out of place.
Solution would be offering users a way to link an updated version of a post to the original