3 ms·

> The best means of interplanetary communication one way messages and a delayed confirmation. I mean, yeah - my whole point is that's not far from what twitter

> The best means of interplanetary communication one way messages and a delayed confirmation. I mean, yeah - my whole point is that's not far from what twitter is already. Musk is getting a social media platform full of users that's tailor-made for interplanetary latency.

noobermin 5y ago
Twitter is RESTful (I think, not web expert) so they are just get requests, but the low latency makes it appear as live communication because you can just keep sending gets and update the client. Over an interplanetary scale with minutes to hours of delay between sending and recieving a signal, it would just be infeasible to create the same experience. It would be much better to just use one way messages, like that of video recordings or sets of texts that are recieved in full, with whatever losses would have to be simply accepted as part of the cost of communicating in space. Confirming receipt and communicating errors to the sender (like that of TCP) would just mean now you have to wait 2x times more to get the next chuck of the message again and hope that that too isn't degraded, so it would be pretty lossy, pretty "slow" level of communication, very much unlike the internet here on earth. Social media in general would be infeasible on an interplanetary scale.
- usrn 5y ago
  Nah, social media is plenty feasible and the Twitter UI/abstraction is probably ideal like GP says (although I think the implementation would be different than you're imagining.) They'd just colocate on the other planets and replicate the database. Stuff like that is a solved problem (in the case of TCP, just increase the window size. Although you'd probably want to make sure transaction boundaries and packet boundaries line up to minimize latency/jitter. If they wanted to minimize the latency they'd make large media available as a merkle tree and make blocks available as they arrived correctly a la Bittorent or IPFS (heh.)) The delay this introduced would have a social effect similar to time zones: "Oh the martians are caught up to event <x> and their reaction is <y>" instead of "Oh the Australians are waking up after event <x> happened today and their reaction is <y>."
- generalizations 5y ago
  I think the entire architecture would have to be rewritten. Talking about the protocols involved is getting lost in the weeds. I'm looking at this from the perspective of the user experience and the physics limitations, and I'm observing that they fit very well together. The point is that a non technical user sees Twitter, fundamentally, as a place to send one way messages out, and to consume one-way incoming broadcast messages. And, crucially, the exact time of sending or receiving isn't a significant aspect of the UX. You just log in and see what showed up in the queue over the last n minutes. If you want to say something, just send a message; you don't really care when it's received, because Twitter is fundamentally all about checking your feed whenever you feel like it, and 23 minutes latency is nothing. The user experience fits the physics limitations. Who cares what the current implementation looks like.

noobermin 5y ago

Twitter is RESTful (I think, not web expert) so they are just get requests, but the low latency makes it appear as live communication because you can just keep sending gets and update the client. Over an interplanetary scale with minutes to hours of delay between sending and recieving a signal, it would just be infeasible to create the same experience. It would be much better to just use one way messages, like that of video recordings or sets of texts that are recieved in full, with whatever losses would have to be simply accepted as part of the cost of communicating in space. Confirming receipt and communicating errors to the sender (like that of TCP) would just mean now you have to wait 2x times more to get the next chuck of the message again and hope that that too isn't degraded, so it would be pretty lossy, pretty "slow" level of communication, very much unlike the internet here on earth. Social media in general would be infeasible on an interplanetary scale.

usrn 5y ago
Nah, social media is plenty feasible and the Twitter UI/abstraction is probably ideal like GP says (although I think the implementation would be different than you're imagining.) They'd just colocate on the other planets and replicate the database. Stuff like that is a solved problem (in the case of TCP, just increase the window size. Although you'd probably want to make sure transaction boundaries and packet boundaries line up to minimize latency/jitter. If they wanted to minimize the latency they'd make large media available as a merkle tree and make blocks available as they arrived correctly a la Bittorent or IPFS (heh.)) The delay this introduced would have a social effect similar to time zones: "Oh the martians are caught up to event <x> and their reaction is <y>" instead of "Oh the Australians are waking up after event <x> happened today and their reaction is <y>."
generalizations 5y ago
I think the entire architecture would have to be rewritten. Talking about the protocols involved is getting lost in the weeds. I'm looking at this from the perspective of the user experience and the physics limitations, and I'm observing that they fit very well together. The point is that a non technical user sees Twitter, fundamentally, as a place to send one way messages out, and to consume one-way incoming broadcast messages. And, crucially, the exact time of sending or receiving isn't a significant aspect of the UX. You just log in and see what showed up in the queue over the last n minutes. If you want to say something, just send a message; you don't really care when it's received, because Twitter is fundamentally all about checking your feed whenever you feel like it, and 23 minutes latency is nothing. The user experience fits the physics limitations. Who cares what the current implementation looks like.