It all depends on what you define as the "cloud" and where your data actually lives. Would I worry about all my data living in one enormous datacenter in the U

It all depends on what you define as the "cloud" and where your data actually lives. Would I worry about all my data living in one enormous datacenter in the USA right next to everybody elses, all owned by one gigantic corporation? Sure. Would I worry less if it was hosted locally by a company in my home town with a datacenter that I can actually go and visit and some agreement in place so that I can have take a copy of the raw data myself and have their copy deleted? Certainly, not to mention that the latency would be better. I think fast upload is the answer to this and I believe that google fiber plans on providing this. In that case you can simply have a home server which is basically a NAS+Router on steroids with a static internet IP (thanks to IPV6) , this server could provide a pretty web interface for "normal" users and a unix shell for l33t h4x0rs. Since upstream is fast enough the service is going to be indistinguishable from something like dropbox in terms of speed and I will have full control. The issues here are data and power reliability , what happens if the disk in my home server dies? There is only so much redundancy you can cram into a small low cost box. I guess the solution would be third party backup services that take advantage of fast upstream to sync the contents of your local box and perhaps can also perform as a "failover" box if your local node fails. The difference here is that you would get a choice of third party backup suppliers, though I imagine most ISPs would provide this as part of the broadband service. One thing that does concern me somewhat on HN and other development oriented websites is that people do a lot of talking about stuff like APIs and Web services rather than protocols and standards. Web APIs provide the illusion of data portability since you have some way of interacting with it, but the reality is that you need a way to dump all of your data and you also need 100% compatible software to move it to. I guess it's difficult to make a business case for designing Email 2.0 unless you know that you can monetise users through some kind of lock-in. Otherwise you could design a fantastic open service only to have someone copy it , undercut you on price and steal all of your users. For example if you could set up your own third party "facebook server" that allows users to interact with FB users without agreeing to facebook TOS or seeing facebook ads.