impossible to provide a browser-client cloud data service without also creating a massive collection of people's private information
Well instead of just hoovering up every photo you take with Glass into G+ [1], they could implement even a fig-leaf's worth of client-side encryption. This is far from impossible. Just saying.
[1] Not just by default, but as I understand it, without even the option of turning that off.
Client-side encryption doesn't help much with privacy concerns because the client and the server are both wholly produced by Google, so you have to assume that Google can pierce that encryption if they want to. That's exactly what happened with DropBox.
It helps protect against unauthorized access, but doesn't really address privacy concerns with Google itself.
I did say "fig leaf". But some level of client-side would at least demonstrate good will, or some level of "giving a fsck".
There is a real difference though, which is the difference between code on your client which perhaps you could at least check if you're stubborn enough[1], and plaintext in a datacenter. This can mean the difference between targeted surveillance and mass surveillance.
Dropbox is a separate case - it was always "trust us" (we have your keys) encryption, and not what I would refer to as "client side" at all.
[1] This is a hard problem if you don't trust the OS
Comments
Well instead of just hoovering up every photo you take with Glass into G+ [1], they could implement even a fig-leaf's worth of client-side encryption. This is far from impossible. Just saying.
[1] Not just by default, but as I understand it, without even the option of turning that off.
Client-side encryption doesn't help much with privacy concerns because the client and the server are both wholly produced by Google, so you have to assume that Google can pierce that encryption if they want to. That's exactly what happened with DropBox.
It helps protect against unauthorized access, but doesn't really address privacy concerns with Google itself.
I did say "fig leaf". But some level of client-side would at least demonstrate good will, or some level of "giving a fsck".
There is a real difference though, which is the difference between code on your client which perhaps you could at least check if you're stubborn enough[1], and plaintext in a datacenter. This can mean the difference between targeted surveillance and mass surveillance.
Dropbox is a separate case - it was always "trust us" (we have your keys) encryption, and not what I would refer to as "client side" at all.
[1] This is a hard problem if you don't trust the OS