Following last week's column about Baxter, my idea for a distributed kinda sorta peer-to-peer Internet data back-up scheme, I expected this week to write about all the problems readers found with the idea, and all the existing Baxter-like services none of us had heard about. Well, things change, and I'll be doing that column next week

I described a similar way to do it a couple of days ago, with links a company that wants to sell such a system for intranets.

I like this idea. It sounds a bit like a miniature Freenet. I'd be a little concerned, though, about losing whatever indexes the server is retaining and being unable to recover the data. The indexes need to be backed up as well, but how do you back them up? It's a bit chicken-and-egg.

It might also be interesting to have proxies which act like clients but which hand off their storage to other clients, thus merging a bunch of different datasets together. The use I have in mind for this is (say) a family all individually backing up to each other, but also handing off to people outside the family as one big lump where it's not obvious who created what. Of course, more degrees of separation between you and your data increases the possibility that you won't be able to get it back from that source again later. A hybrid, though, where a client essentially backs up its own backups could cause your files to end up on the disks of people you've never even met.