4 ms·

> champion decentralized sharing systems which would make this sort of content uncensorable. Decentralization does not make illegal content uncensorable, in fa

> champion decentralized sharing systems which would make this sort of content uncensorable. Decentralization does not make illegal content uncensorable, in fact the opposite is true, because regular citizens and small organizations hosting individual nodes are exposed to a lot more risk and legal liability than a massive tech giant. These gigantic centralized networks allow this type of content to hide in plain sight because something like Facebook is "too big to fail" and won't ever be shutdown regardless of how many times illegal content is re-uploaded to the platform.

Y_Y 5y ago
Cunningham's Law dictates that I must mention, there are many decentralized systems in which content cannot simply said to be hosted on one node or another. If I use the Chinese Remainder theorem to distribute data to five hosts, such that you need to get the data from three of them to reconstruct the file, which one is hosting it? And what if it's encrypted so that the host couldn't read the data anyway?
- root_axis 5y ago
  > If I use the Chinese Remainder theorem to distribute data to five hosts, such that you need to get the data from three of them to reconstruct the file, which one is hosting it? This kind of technical obfuscation won't stymie a judge, the owners of all five nodes would be held liable if they were instrumental in facilitating access to the illegal material. > And what if it's encrypted so that the host couldn't read the data anyway? This isn't really a practical setup for a social media website. If the host can't decrypt the data then the social media users won't be able to do so either. People could share illegal content in some kind of private E2EE group chat like Signal, but in such a scenario decentralization is of no relevance since the buck stops at E2EE either way.
  - Y_Y 5y ago
    But they aren't instrumental! Maybe only four actually sent the file, and even a judge can be made to understand that one of those was superfluous. If you want to make it even less clear, then imagine that the client is told they can download chunks from ten peers, of which five have relevant data, if which any three are sufficient. So someone comes along and downloads eight random chunks and reconstructs the bad file. If you are required to prove that a particular peer infringed a copyright by distributing a protected work, or a work derived therefrom I think you could have a hard time. I'm not sure I understand your second objection. Suppose I publish a "status" including some copyrighted work and I encrypt it such that you need a private key corresponding to the private key of anyone on my friends list to decrypt. Then it's distributed among a thousand peers (who are not my friends), such that one of my friends can download from a random twenty, receive garbage from ten, and they'll have a 30% change of being able to reconstruct the encrypted file. Can you now prove that on the balance of probabilities a peer has violated the copyright?

Y_Y 5y ago

Cunningham's Law dictates that I must mention, there are many decentralized systems in which content cannot simply said to be hosted on one node or another. If I use the Chinese Remainder theorem to distribute data to five hosts, such that you need to get the data from three of them to reconstruct the file, which one is hosting it? And what if it's encrypted so that the host couldn't read the data anyway?

root_axis 5y ago
> If I use the Chinese Remainder theorem to distribute data to five hosts, such that you need to get the data from three of them to reconstruct the file, which one is hosting it? This kind of technical obfuscation won't stymie a judge, the owners of all five nodes would be held liable if they were instrumental in facilitating access to the illegal material. > And what if it's encrypted so that the host couldn't read the data anyway? This isn't really a practical setup for a social media website. If the host can't decrypt the data then the social media users won't be able to do so either. People could share illegal content in some kind of private E2EE group chat like Signal, but in such a scenario decentralization is of no relevance since the buck stops at E2EE either way.
- Y_Y 5y ago
  But they aren't instrumental! Maybe only four actually sent the file, and even a judge can be made to understand that one of those was superfluous. If you want to make it even less clear, then imagine that the client is told they can download chunks from ten peers, of which five have relevant data, if which any three are sufficient. So someone comes along and downloads eight random chunks and reconstructs the bad file. If you are required to prove that a particular peer infringed a copyright by distributing a protected work, or a work derived therefrom I think you could have a hard time. I'm not sure I understand your second objection. Suppose I publish a "status" including some copyrighted work and I encrypt it such that you need a private key corresponding to the private key of anyone on my friends list to decrypt. Then it's distributed among a thousand peers (who are not my friends), such that one of my friends can download from a random twenty, receive garbage from ten, and they'll have a 30% change of being able to reconstruct the encrypted file. Can you now prove that on the balance of probabilities a peer has violated the copyright?