The biggest problem I see here is forwarding someone else's unknown <i>and</i> unencrypted content.<p>Imagine you're running a relay on your machine, and somebody posts some CSAM, or stolen CC data, or hate speech, etc through your relay.<p>Unlike an email client, your relay actually makes that message available to anyone.<p>Unlike an email relay, you are not forwarding the message to someone else like a dumb pipe; there's no final destination.<p>Unlike a torrent / DHT node, you do not know beforehand what you are going to make available.<p>Unlike a Tor exit node, you do not deal in encrypted fragments which you don't store locally.<p>Unlike an IPFS node, you do not store fragments of files while not even having a full set of such fragments for a given file.<p>Even though the message is signed, you can't reasonably prove that you do not possess the private key used to sign it, or have not possessed.<p>So, if you are an anti-censorship activist, and your machine gets searched (which may be even easier to do in the case of a VPS), you, the runner of the relay, may have some pretty unpleasant time. Even if you manage to convince everyone that you only forward these messages and do not inspect or endorse them, and the authorities will not even fine you, it will cost you time and trouble.<p>Worse, it becomes easy to send something incriminating from a throwaway or stolen client <i>through</i> your relay, and report you to the police. This can of course be done with email or any IM, but these will immediately show you what you've got, so you can e.g. immediately delete it. They do not allow perfect strangers quietly put random stuff onto your disk.<p><i>Of course</i> the point of the protocol is to resist censorship, so any content deemed criminal by someone cannot be entirely blocked. But the publishers and the receivers of such content make their choice, while those running a relay seem to remain unprotected from from consequences of that choice which they did not make.