Jeremy Spilman [ARCHIVE] on Nostr: 📅 Original date posted:2015-03-27 📝 Original message:> On Mar 27, 2015, at 8:16 ...
📅 Original date posted:2015-03-27
📝 Original message:> On Mar 27, 2015, at 8:16 AM, Matt Whitlock <bip at mattwhitlock.name> wrote:
>
> Isn't the goal of this exercise to ensure more full nodes on the network?
Basically we're talking about a form of Sybil defense and better quantifying true blockchain resiliency by proof of storage.
In this case the goal is to see if we can prove the number of distinct digital copies of the blockchain. This is actually a tricky problem because it will (always?) devolve to inferences from response timing, and we are running over a heterogenous network with heterogeneous machines.
It would be extremely impressive to achieve a reliable mechanism for discerning a local copy exists under these constraints, particularly without false positives and false negatives, and without imposing very substantial one-time encoding costs, e.g. on par with doubling the verification cost.
I think while its a difficult cost-benefit analysis, even code complexity aside, it's interesting to discuss all the same!
Simply having many unique IP addresses possibly accessing the same unique copy provides a different (if any) benefit. E.g. Tor uses IPs as a cost factor, but (until recently?) didn't even factor in things like them all being the same Class C.
📝 Original message:> On Mar 27, 2015, at 8:16 AM, Matt Whitlock <bip at mattwhitlock.name> wrote:
>
> Isn't the goal of this exercise to ensure more full nodes on the network?
Basically we're talking about a form of Sybil defense and better quantifying true blockchain resiliency by proof of storage.
In this case the goal is to see if we can prove the number of distinct digital copies of the blockchain. This is actually a tricky problem because it will (always?) devolve to inferences from response timing, and we are running over a heterogenous network with heterogeneous machines.
It would be extremely impressive to achieve a reliable mechanism for discerning a local copy exists under these constraints, particularly without false positives and false negatives, and without imposing very substantial one-time encoding costs, e.g. on par with doubling the verification cost.
I think while its a difficult cost-benefit analysis, even code complexity aside, it's interesting to discuss all the same!
Simply having many unique IP addresses possibly accessing the same unique copy provides a different (if any) benefit. E.g. Tor uses IPs as a cost factor, but (until recently?) didn't even factor in things like them all being the same Class C.