Diaspora is not as good social network as facebook and can not be very good at privacy because personnal user data stored in a pod are not encrypted and can be read by the pod owner. Encrypt/decrypt should be performed only at the browser level. I think diaspora should first start building a good cryptographic framework. Once the foundations are ok, they will find people to help for pretty features.
The problem is that a social network that respect privacy would be a paradise for bad guys.
> Diaspora is not as good social network as facebook and can not be very good at privacy because personnal user data stored in a pod are not encrypted and can be read by the pod owner. Encrypt/decrypt should be performed only at the browser level.
That would probably kill Diaspora's most realistic (obvious) option for scaling up to the size of a major social network. Right now it's small enough that it's reasonable for individuals to run pods altruistically.
But if it gets big, that's going to get expensive. Somebody will have to pay for all that hardware, electricity and bandwidth. Nobody wants to pay to use social networks, so it sure won't be the users. At least not directly. A much more feasible option is to aggregate all the data you're privy to as a pod operator and use it for advertising purposes.
Well, encryption doesn't increase the cost much and is only welcome. The way to deal with increased user base is to even out the federation - i.e. in ideal case the pod will run on user's own computer, and thus will reduce the cost of hosting to virtually the cost of Internet connectivity, which is already paid by the user. Of course in practice it's not always the case, but quite a number of such pods already exist.
Unless there's a clever scheme I don't know of, encryption that prevents pod operators from reading content would increase the cost by making the system significantly more complicated.
If the data you post to your network is encrypted, then that means anyone who wants to see it needs to have a key to decrypt it. Which might not be terrible as long as posting stuff to the network is all-or-nothing. But if the network needs to have support for allowing different acquaintances access to different content depending on your personal desires then it would become morass. You'd need to encrypt everything differently for each of the different combinations of people or groups of people that you might want to share with, and every time you friend someone, unfriend someone, or rearrange your sharing settings for a person, you'd potentially have to generate and distribute new keys.
As far as self-hosting pods, that's fine as long as you want to limit your social network to people who are in a position to host their own pods. But that's got to be a fraction of a percent of all Internet users.
At present installing your own pod is indeed a difficult task, and is too involved for an average user. However Diaspora is working on simplifying this.
Yes, encryption with such access scope is complex, but I was referring to the cost of running it, not to the cost of developing the architecture. Development can indeed be hard and costly, but many things are, and open source still tackles hard problems successfully.
The problem is that a social network that respect privacy would be a paradise for bad guys.