4 ms·

I Might Be Spartacus: a differential privacy marketplace

59 points by mbaytas 8y ago

anonymousDan 8y ago
Interesting, but what's to prevent the person selling the data from completely fabricating it (as opposed to following noise according to some differential privacy algorithm)? Or is the assumption that enough data owners are truthful that you could filter out this behaviour?
- pythonaut_16 8y ago
  Could this be a useful model to apply to analytics and tracking? Selective sampling is already common (run X diagnostic 10% of the time). Could you say to a user of your application "Hey, we are tracking these X parameters, but applying Y degrees of uncertainty to your individual data." I'm not pro-tracking by any means, but would this be a reasonable improvement to users' privacy without first needing to convince companies to stop all tracking wholesale?
0xname 8y ago
While the idea of selling and controlling one's own data is enticing, would people actually be willing to set up the requisite hardware/software to manage it all? I can imagine a sort of "black box" that manages user preferences but for the average person who doesn't necessarily care about privacy, the proposition of earning a few dollars a month doesn't seem like a strong motivator.
- rexpop 8y ago
  No, and they shouldn't have to. The few dollars a month shouldn't be the primary the motivator, but it's hard to imagine--even, perhaps especially, for those of us pining for such a future--an internet in which our identities are not bartered for fractions of a penny.
- taurath 8y ago
  Everyone has a phone, and a private cloud to securely store data is not terribly insecure if done right. Life-Privacy as a first class feature of mobile OSs could be a huge opportunity to giving people more control over their data. I don’t see google doing this, but Apple could.
jka 8y ago
This is definitely a step forward in thinking regarding 'personal data control', but if you imagine 'data brokers' as the potential attacker here, and assume that their goal is to sell your information to the highest bidder, they still have some opportunity here. They could/would form 'data validation' companies who would investigate the plausibly-deniable facts you had been distributing. For example, the validator could buy your age from five different vendors and likely end up with a pretty good approximation of your real age. Ok, perhaps this creates a 'data marketplace' and there would be competition - but the situation is ultimately very similar to credit scoring agencies (Equifax et al), and clearly something is wrong in that industry. There'd be ample opportunity for these data validators to follow the same pattern as credit scoring agencies and charge third parties for access to the validated data. And possibly even charge people to have details corrected. Regulation might help, but regulatory capture happens. Some good ideas here and it'd be great to develop this further and make it more robust from attack.
eksemplar 8y ago
The article talks about it, but it seems too fantastical to me. I mean, what’s the business model really? Facebook earns around $2.5 a year per user, is enough people really going to bother selling their data for almost nothing?
kouohhashi 8y ago
Idea is very interesting and smart but I don't think people want to sell their data consciously. If Facebook get our data without our knowledge, it's kind of like acceptable but I don't think people push button to give away our data. It's a huge cognitive hinderance. And data can have value only when the volume is big. Since each data(small datasets) has almost zero value, i don't think it's expensive/attractive enough for us to push the button to sell it.

59 points by mbaytas 8y ago

Interesting, but what's to prevent the person selling the data from completely fabricating it (as opposed to following noise according to some differential privacy algorithm)? Or is the assumption that enough data owners are truthful that you could filter out this behaviour?

pythonaut_16 8y ago
Could this be a useful model to apply to analytics and tracking? Selective sampling is already common (run X diagnostic 10% of the time). Could you say to a user of your application "Hey, we are tracking these X parameters, but applying Y degrees of uncertainty to your individual data." I'm not pro-tracking by any means, but would this be a reasonable improvement to users' privacy without first needing to convince companies to stop all tracking wholesale?

0xname 8y ago

While the idea of selling and controlling one's own data is enticing, would people actually be willing to set up the requisite hardware/software to manage it all? I can imagine a sort of "black box" that manages user preferences but for the average person who doesn't necessarily care about privacy, the proposition of earning a few dollars a month doesn't seem like a strong motivator.

rexpop 8y ago
No, and they shouldn't have to. The few dollars a month shouldn't be the primary the motivator, but it's hard to imagine--even, perhaps especially, for those of us pining for such a future--an internet in which our identities are not bartered for fractions of a penny.
taurath 8y ago
Everyone has a phone, and a private cloud to securely store data is not terribly insecure if done right. Life-Privacy as a first class feature of mobile OSs could be a huge opportunity to giving people more control over their data. I don’t see google doing this, but Apple could.

jka 8y ago

This is definitely a step forward in thinking regarding 'personal data control', but if you imagine 'data brokers' as the potential attacker here, and assume that their goal is to sell your information to the highest bidder, they still have some opportunity here. They could/would form 'data validation' companies who would investigate the plausibly-deniable facts you had been distributing. For example, the validator could buy your age from five different vendors and likely end up with a pretty good approximation of your real age. Ok, perhaps this creates a 'data marketplace' and there would be competition - but the situation is ultimately very similar to credit scoring agencies (Equifax et al), and clearly something is wrong in that industry. There'd be ample opportunity for these data validators to follow the same pattern as credit scoring agencies and charge third parties for access to the validated data. And possibly even charge people to have details corrected. Regulation might help, but regulatory capture happens. Some good ideas here and it'd be great to develop this further and make it more robust from attack.

eksemplar 8y ago

The article talks about it, but it seems too fantastical to me. I mean, what’s the business model really? Facebook earns around $2.5 a year per user, is enough people really going to bother selling their data for almost nothing?

kouohhashi 8y ago

Idea is very interesting and smart but I don't think people want to sell their data consciously. If Facebook get our data without our knowledge, it's kind of like acceptable but I don't think people push button to give away our data. It's a huge cognitive hinderance. And data can have value only when the volume is big. Since each data(small datasets) has almost zero value, i don't think it's expensive/attractive enough for us to push the button to sell it.