This presumes humanity is a space fairing or interplanetary civilization.
How would something like the fediverse, internet, cryptocurrency, etc function with major latency? As an example, a signal takes between 5 and 20 minutes to travel from earth to mars. A roundtrip response would take at best 10 minutes and at worst 40 minutes. Now lets say you live on mars and your home lemmy instance is mars.social. You want to see what news people are chatting about on earth and heard that [email protected] is a good community. If you put that into your instance search box on mars.social the absolute best you can hope for is a response in 10 minutes. I assume the request would totally fail anyway due to rtt being set to low and the packets expiring before they ever reached the destination. The internet we all know and love is totally intolerant of high latency. Just ask people who use satellite internet or tor.
Edit: i think, but am not certain, that ipv6 replaced rtt with hop count. If so this may not be an issue as the time it takes would not matter as long as the hop limit was not reached.
Caches and new protocols probably.
As far as I know Lemmy already caches content. So a post on earth.social would take 10-20 minutes to reach mars.social, but the browsing on your end would always feel responsive.
Centralized sites that want to be available on Mars will set up servers there as well and do a lot of caching and sync-ing between Earth and Mars to make the browsing feel responsive. For example if you’re using Facebook on Mars you will be using a Mars server and it will feel responsive as usual. But content from Earth will not show up immediately for you, and content you upload will not show up instantly on Earth. The delay will mostly be noticeable if you try to have a discussion with someone from Earth.
For crypto and other networks that expect faster responses, we will probably see a shift in architecture. For example maybe all blockchains will feature a form of sharding where you end up validating/mining blocks only with nodes in proximity and then sync back to a main chain at a large interval, where the latency is ok.