4 ms·

I think the faster we make companies liable for the content on their servers the better. Why? Because I want distributed/federated data to become an actual pr

I think the faster we make companies liable for the content on their servers the better. Why? Because I want distributed/federated data to become an actual priority to these companies. I want it to hit their bottom line when they get sued for content uploaded to their servers so they consider keeping the data owned by the user who generated it.

kodablah 8y ago
Surely you realize that making data a liability reduces the ability for you and I to host it, not increases it. I understand you are against centralization of data, but you can't magically commoditize and distribute it via punishments. You'll only strengthen those that can take a hit. While it seems mathematically logical to say "well, if we make it less financially beneficial to have data, they won't have it" what happens in practice is "well, if we make it less financially beneficial to have data, nobody will have it". There are alternatives to decentralization that have less effect on smaller orgs.
- Kalium 8y ago
  If anything, I think liability would be a forcing function that would strongly encourage centralization. Only those with deep pockets could afford it.
- wesleytodd 8y ago
  Yeah I don't want to be hosting data I have not vetted either. I think being aware of the content any device you own is serving on the net is a good thing, even if it reduces the amount of people willing to do it for others. TBQH if small companies cannot host user data that is good too. It should be very difficult for companies of any size to house data they are not willing to defend legally.
  - wesleytodd 8y ago
    Also, I am quite aware of how disruptive this ideal would be to most of our current large internet companies. I don't think it will realistically change any time soon, and am sure it would have side effects which would have to be balanced. But the direction we are headed now seems much worse to me than the alternatives.
  - kodablah 8y ago
    > TBQH if small companies cannot host user data that is good too. It should be very difficult for companies of any size to house data they are not willing to defend legally. We just disagree here. We know who can and who can't defend things legally. On a personal level, I don't like my options further reduced for who I can willingly give my data to. Legislate and determine what constitutes private data, data misuse, and/or surreptitious gathering if you want, but non-private information freely and consentually given should want to be freely proliferated.
    - wesleytodd 8y ago
      My argument is more that forcing tech companies to invest in distributed data ownership tech would push that forward in a way that would enable more competition of services because your data would be portable, not locked into vendors like it is now. In the short term it might result in less variety, but only until the tech proliferated and reached maturity. Then services could simply spin up on top of the data that you already host yourself. So you would have to take on the responsibility of finding a good way to host the data you want available (probably from your phone), but after that you would be able to use any service willing to operate on your data, but your data would be in public standards, not proprietary storage inside a facebook/google/amazon data center.
shawn 8y ago
I’ve been working on that. It’s a hard problem, but emule works quite well. I propose something based on that: when you start a program, you advertise all the sha256’s of all the files you share. Whenever you need a resource, you send a request to the dht for its sha256. Anyone who has the file will opportunistically try to send it to you. Since it’s a dht, there’s nothing to take down. And since it operates on sha256s of data, it’s implicitly secure. And it uses tor for the rendezvous, so it’s not possible to track who is requesting what, except by unique ID (which can just be a bitcoin wallet address you control). The hard part is, what do you do about abuse? What if someone spams the network with bots that try to fulfill every request with bad data? If anyone has research refs in this direction, I’d be grateful to read them. I’d also like to avoid a blockchain if possible, since it seems unnecessary for simple federated distributed data.
- wesleytodd 8y ago
  Check out Dat: https://datproject.org/ https://datproject.org/
  - shawn 8y ago
    Excellent: https://raw.githubusercontent.com/datprotocol/whitepaper/master/dat-paper.pdf https://raw.githubusercontent.com/datprotocol/whitepaper/mas... It seemed too simple of an idea not to already be done. Hopefully it conceals your IP. Arg. Nope. It does not. It leaves it up to the client to decide how best to request the content, and by default that almost certainly guarantees no privacy.
    - wesleytodd 8y ago
      It was not intended to provide privacy, the the distributed data store. If you want privacy you probably just want to have a private swarm, which is perfectly fine to setup. Just implement a client with a whitelist or another mechanism for the privacy you want.
      - shawn 8y ago

kodablah 8y ago

Surely you realize that making data a liability reduces the ability for you and I to host it, not increases it. I understand you are against centralization of data, but you can't magically commoditize and distribute it via punishments. You'll only strengthen those that can take a hit. While it seems mathematically logical to say "well, if we make it less financially beneficial to have data, they won't have it" what happens in practice is "well, if we make it less financially beneficial to have data, nobody will have it". There are alternatives to decentralization that have less effect on smaller orgs.

Kalium 8y ago
If anything, I think liability would be a forcing function that would strongly encourage centralization. Only those with deep pockets could afford it.
wesleytodd 8y ago
Yeah I don't want to be hosting data I have not vetted either. I think being aware of the content any device you own is serving on the net is a good thing, even if it reduces the amount of people willing to do it for others. TBQH if small companies cannot host user data that is good too. It should be very difficult for companies of any size to house data they are not willing to defend legally.
- wesleytodd 8y ago
  Also, I am quite aware of how disruptive this ideal would be to most of our current large internet companies. I don't think it will realistically change any time soon, and am sure it would have side effects which would have to be balanced. But the direction we are headed now seems much worse to me than the alternatives.
- kodablah 8y ago
  > TBQH if small companies cannot host user data that is good too. It should be very difficult for companies of any size to house data they are not willing to defend legally. We just disagree here. We know who can and who can't defend things legally. On a personal level, I don't like my options further reduced for who I can willingly give my data to. Legislate and determine what constitutes private data, data misuse, and/or surreptitious gathering if you want, but non-private information freely and consentually given should want to be freely proliferated.
  - wesleytodd 8y ago
    My argument is more that forcing tech companies to invest in distributed data ownership tech would push that forward in a way that would enable more competition of services because your data would be portable, not locked into vendors like it is now. In the short term it might result in less variety, but only until the tech proliferated and reached maturity. Then services could simply spin up on top of the data that you already host yourself. So you would have to take on the responsibility of finding a good way to host the data you want available (probably from your phone), but after that you would be able to use any service willing to operate on your data, but your data would be in public standards, not proprietary storage inside a facebook/google/amazon data center.

shawn 8y ago

I’ve been working on that. It’s a hard problem, but emule works quite well. I propose something based on that: when you start a program, you advertise all the sha256’s of all the files you share. Whenever you need a resource, you send a request to the dht for its sha256. Anyone who has the file will opportunistically try to send it to you. Since it’s a dht, there’s nothing to take down. And since it operates on sha256s of data, it’s implicitly secure. And it uses tor for the rendezvous, so it’s not possible to track who is requesting what, except by unique ID (which can just be a bitcoin wallet address you control). The hard part is, what do you do about abuse? What if someone spams the network with bots that try to fulfill every request with bad data? If anyone has research refs in this direction, I’d be grateful to read them. I’d also like to avoid a blockchain if possible, since it seems unnecessary for simple federated distributed data.

wesleytodd 8y ago
Check out Dat: https://datproject.org/ https://datproject.org/
- shawn 8y ago
  Excellent: https://raw.githubusercontent.com/datprotocol/whitepaper/master/dat-paper.pdf https://raw.githubusercontent.com/datprotocol/whitepaper/mas... It seemed too simple of an idea not to already be done. Hopefully it conceals your IP. Arg. Nope. It does not. It leaves it up to the client to decide how best to request the content, and by default that almost certainly guarantees no privacy.
  - wesleytodd 8y ago
    It was not intended to provide privacy, the the distributed data store. If you want privacy you probably just want to have a private swarm, which is perfectly fine to setup. Just implement a client with a whitelist or another mechanism for the privacy you want.
    - shawn 8y ago