I technically disagree with the premise. Relay operators *are* hosting the notes, but they're hosts in the same way that twitter is a host, not responsible for what end users actually post. The idea (as I saw it) was that relay operators in different jurisdictions would be subject to different laws. So lets say hypothetically that Germany wanted to take one of bot's notes down they could force German relay operators to remove the notes, but not say a relay in the Canary Islands so that would provide some resiliency for the end user in the case of censorship.

However for an end user, if bot is hypothetically in the US and a US court directs them to remove a post they would generally expect to have that functionality. It's understood that this may not scrub the post from existence due to on or offline archiving of said post, but it would at least be removed from the primary source where possible.

Now if we look at deletes on nostr. I tested from other apps today creating a note on iris.to and then deleting it on snort.social. All of the relays listed on NADAR accepted that delete including damus.io. Unless damus is specifically flagging notes sent from the app differently it should be possible to delete those notes from other apps. As such you aren't making nostr delete proof, you're merely inconveniencing your users, forcing them to either migrate to another option or switch between damus and another tool if they wish to or are compelled to delete note(s).