3 ms·

I have always been interested in imagining if Internet ever goes down (Or our computing power comes to a grinding halt due to an EMP strike) then how would bitc

I have always been interested in imagining if Internet ever goes down (Or our computing power comes to a grinding halt due to an EMP strike) then how would bitcoin networks work (again not from a realistic point of view, but from a Anathem (by Neal Stephenson) way. The difficulty rate would have to be lowered, low enough to be processed via some human effort, but not super easy[1]. Maybe there would be these monasteries where monks would process transactions by hand. Considering there would be synchronization issues so each region would mine the transactions using state channels and every month whenever ships carry updated blocks across the ocean, there would be a sync up of inter-continental trade. Or at that point we would all revert to using gold. 1. You can get 0.67 hash per day if you perform these transaction only using pen and paper, http://www.righto.com/2014/09/mining-bitcoin-with-pencil-and-paper.html http://www.righto.com/2014/09/mining-bitcoin-with-pencil-and...

rtkwe 9y ago
It wouldn't, at least not in a useful way. At that point there's two major issues: hashing becomes impossibly slow, network partition aka forking the chain, and validating transactions. Just distributing a new block becomes a huge task in and of itself. Also before a transaction can be included our hypothetical human miner has to check to ensure that the inputs/source address has a corresponding output/target address earlier in the chain. That step could require scanning the whole 15 GB blockchain to confirm that a transaction is valid.
d33 9y ago
Thanks for drawing my attention to Anathem! Sounds like a nice (but very long) read: https://en.wikipedia.org/wiki/Anathem https://en.wikipedia.org/wiki/Anathem
cyphar 9y ago
I reckon that at the point where you no longer have near-instantaneous network communication (and assuming we didn't lose all of the computer scientists) then humanity would revert to using a Paxos-like system for managing state between different areas (I remember seeing on HN not too long ago an interesting explanation of Paxos -- not too different to the paper -- about how you could create a style of legislation between neighbouring cities with hermits and achieving consensus with Paxos). I don't know if you could do proof-of-work with Paxos, maybe it'd be possible to do proof-of-stake or something like that?

rtkwe 9y ago

It wouldn't, at least not in a useful way. At that point there's two major issues: hashing becomes impossibly slow, network partition aka forking the chain, and validating transactions. Just distributing a new block becomes a huge task in and of itself. Also before a transaction can be included our hypothetical human miner has to check to ensure that the inputs/source address has a corresponding output/target address earlier in the chain. That step could require scanning the whole 15 GB blockchain to confirm that a transaction is valid.

d33 9y ago

Thanks for drawing my attention to Anathem! Sounds like a nice (but very long) read: https://en.wikipedia.org/wiki/Anathem https://en.wikipedia.org/wiki/Anathem

cyphar 9y ago

I reckon that at the point where you no longer have near-instantaneous network communication (and assuming we didn't lose all of the computer scientists) then humanity would revert to using a Paxos-like system for managing state between different areas (I remember seeing on HN not too long ago an interesting explanation of Paxos -- not too different to the paper -- about how you could create a style of legislation between neighbouring cities with hermits and achieving consensus with Paxos). I don't know if you could do proof-of-work with Paxos, maybe it'd be possible to do proof-of-stake or something like that?