Interesting that this meme seems to be doing the rounds now in lots of contexts. People seem to have more faith in their own hard drives (or institutional support) than I do I have to say. - Cameron Neylon

I guess there are two issues here - if it is "on your own servers" (bearing in mind that means very different things to different people) it is relatively unlikely to be cut off without notice but I would rate the chances of catastrophic loss of data on both my own disks and (some) institutional ones as being higher than it would be on, for instance, things running on amazon servers - Cameron Neylon

I think a good solution would be open exchange formats. If you can easily transfer the data from one service to another, then it is also easy to maintain redundancy by having your data on multiple services. This would protect you against data loss in case one of them ceases to be. - Lars Juhl Jensen

It also ties into what I think institutional repositories should be doing - which is aggregating all this kind of material back and holding it independently. But data interchange formats are a key first step. You have to be able to choose to get it out and put it somewhere else (I'm sure we're all doing that with FF [cough...cough]) - Cameron Neylon

A few thoughts: 1) With big firms like google or amazon, I think it's less an issue of hard drives crashing, and more an issue of lock-in (and potential lock-out). 2) I'm looking forward to more services like birdfeeder (http://dsandler.org/wp...) that allow local control, but still sync to the cloud. (Not up yet, but looks promising) - Chris Miller

Deepak is NOT going to like this. The issue about lock-in is relevant, though. - Mr. Gunn