#activitypub wizards, what happens if you set up a redundant relay between a gaggle of #Mastodon servers?

#activitypub wizards, what happens if you set up a redundant relay between a gaggle of #Mastodon servers? Nothing, I assume, apart from wasting traffic and CPU cycles?

@saustrup@mstdn.dk I would assume the relays themselves are smart enough to not relay content they've already seen before

@julian My thought was to set up at least two relays for relaying messages between, say, 10 Mastodon servers. Without knowing exactly how relays work, I would assume that they would pull down at least the unique identifiers of each post and keep track of what they've relayed where. That means each server will probably get at least two notifications for each foreign post, and will then reject the second. My fear was that it would be the entire message being pulled and pushed, which could end up being a bit of a waste.

@saustrup @julian Old(ish) fart here ... I've sometimes thought that NNTP's old IHAVE/SENDME subprotocol might work well for relays. It would involve a bit of redundant transmission of object IDs, but the object data itself would be sent pretty efficiently, certainly more so than with an ongoing stream of HTTP POST requests.