HTTP doesn't have any redundancy at this level. Redundancy for websites means using multiple servers on the same IP address usually through network routing tricks. DNS round-robin for example doesn't quite work.

And yet HTTP servers seem to be doing just fine. Relays will become as reliable as websites over time, there is no reason they shoudln't be.

Plus then we have our redundancy of specifying multiple inboxes or multiple outboxes.

So I don't think there is a long-term problem with reliability of relays.

Currently though, sometimes all of somebody's relays are not working properly because nostr is still a work in progress. 

 Yeah I also think that those who run relays on here not for fun but to support network will be doing so. So, it should be more stable over the time. Regarding malicious stuff I’m not sure though. How to deal with that ?