What's in your threat model that's simultaneously a non-targeted attack, but also sophisticated enough to get through basic security (eg. strong password or public key)?
I commented extensively on the other thread (cliff notes: nonstandard SSH ports induce their own security risks [if on port 1024 or above], add complexity with minimal benefit, and generally indicate you should be focused your effort elsewhere - like setting up a wireguard VPN.), but they're not entirely wrong here. Just misguided. If you are guarding against OpenSSH 0 days as an attack vector, one, there are better security measures - see VPNs - to take that provide your greater protection in general, and two, port scans are trivially cheap, so once they infect all the things listening on port 22 they'll use that compute and network capacity to scan the rest of the internet to find things listening on nonstandard ports, because botnets = money, so increasing the size is a worthwhile investment. It gains you a few hours, or you could throw up Wireguard on another host where someone then needs both a Wireguard 0 day and an SSH 0 day at the same time to compromise your system.
