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.
Presumably there would be a cache on Mars of [email protected] so that anybody who wants to view it would not have to wait 10 minutes… they would get the cached update - so they would immediately see the community as it was 10 minutes ago.
This cache would be continuously updating so to the user on Mars, there actually isn’t that much disruption. Every time they check, there would be updates.
10 minutes or even 40 minutes is not that long in the grand scheme of things. We start talking about lightyears is when I think it starts to break down.
TCP/IP stacks are going to need pretty large buffers if a packet needs resending and takes 20 minutes round trip to get it.
Link layer protocols are going to need to implement some kind of redundancy and parity scheme that accounts for the enormous latency (I’m sure NASA already has something like this)
Also the default 300 second HTTP timeout is going to need to be adjusted lol
I think lemmy already caches content. Viewing beehaw from mars.social should not feel unresponsive even though you’re looking at content that is 10-20 minutes old.
The only issue might be when Mars and Earth are on opposite sides of Sol. Then the cache get’s held for however many weeks it takes for a clear signal to go through.
https://en.wikipedia.org/wiki/Interplanetary_Internet
there has been some thought about this. imagine a series of interplanetary satellites that act as nodes. so you don’t need a clear signal from mars -> earth. you just need a clear signal to the next node, which would presumably be easier
obviously this is all sci-fi talk at this stage, but setting up internet on a mars colony is probably not gonna be the hardest part of colonization
It may not be the hardest part for sure, but could you imagine telling a kid they couldnt watch tiktok because they couldnt connect to the servers? People would never sign up for the mission.
I think that solar flares would be more of an issue as satellites get farther from earth. It’ll take a lot more resources to replace a damaged satellite orbiting mars than it would for one orbiting earth.
Not once microgravity manufacturing is implimented. Pull an asteroid into dock, break it down, and use the parts to make just the satelite. No rockets required! (Way oversimplifying it of course. But even mars-based satelite manufacturing would be much more energy inexpensive)
But how would the cached copy be started to begin with? Take a server to earth and plug it in to the net? Rsync (if it will establish the connection to begin with)?
You would obviously need to adjust a lot of software that has expectations for timeouts but there is no general technical issue that would prevent most protocols from working even with much higher latencies.
The ones most affected would be the ones where a user expects interactive feedback but the computers don’t care if they have to wait 3 ms or 30 minutes for a response as long as the timeouts are adjusted accordingly.
You would have to start adjusting for that on the lowest levels though since even TCP has a lot of assumptions for reasonable response times built in.