No, not distributed system like “beowulf cluster” or “seti@home”, I’m talking a distributed system for each and every person on the web. I’m trying my best to get there myself.

By distributed, I mean both “worlwide accessible” and “no single point of failure that’s not being taken care of by a professional”. Note that if you can satisfy the second, then you’ve probably satisfied the first. Primarily, I care about my data. I can re-image my iBook, and I can get from 0 to full Debian install in about 2 hours, but what I can’t do easily is regenerate my data. Wait – that’s not true. I can’t regenerate the data that I created. The data that I imported is still out there waiting for importation. Thus, I feel no need to back up my mp3 collection. All of my legal mp3s have a physical backup already, and if I lose the unauthorized copies (yarr) then my conscience will be a little cleaner. The only thing I really need to save in iTunes is my ratings. Which I see no way to do.

Freaking iTunes.

But! Back to my larger point. I want my own personal data saved on other systems, and I want those systems to be maintained by professionals. I’ve created a relatively small amount of data – bookmarks, text files, and lots of digital pictures. If people are willing to host these small (but important to me) things, then I would be more than happy to allow them to mine for correlations (within reason).

If all of my bookmarks could be on del.icio.us with a nice interface that waded through my sea of tags effectively, then I would solely use foxylicious. I’m thinking of backing up all of my photos by uploading them the flickr. I’m fine if they show unintrusive ads or have some preferred vendor when I make prints from those pictures – they are performing a valuable service for me by making sure that my data is in the hands of professionals.

Hey! iPhoto! Howzabout a nice flickr export feature?

Hey! Flickr! Howzabout a gallery exporter that will read in my galleries and save them to flickr with the same comments and by making the directories into tags? I’ll write it myself if I have to, but I won’t like it!

I’ve already moved away from having a local addressbook with things like Orkut and the like. Unfortunately, coverage there is spotty at best. We need (read: I want) some nice central server that can keep track of my contacts, allow them to update their own addresses, and it needs to have an open API. I predict that the first orkut/friendster/etc-alike that has an open API for people to make clients will go really far.

Pleasantly, people are starting to realize that open API does not mean “unprofitable dot com loser idea”. Flickr is making money, del.icio.us takes very little cash to run and he has a really nice data set for text ads, and Google and Amazon are both firmly in the black. I’m hoping that the next one will be some nice calendaring/event scheduling software. Ideally, this software would combine with open-API-orkut-replacement, each using a really simple interface based on easy standards, and then I could invite to events people who claim to be my friend.

Now the only things that I need to worry about are email and my writings and programs. Email is on the department server, and we pay Lauradel &co. a lot to make sure that it doesn’t disappear on us (hopefully – they definitely deserve it). If I was more comfortable with webmail, I would be all about the gmail as a Pine replacement. I still hate editing text online, but AJAX web apps are becoming more and more compelling replacements. If I can get some sort of pico/vim/emacs-alike as my textfield editor by default, then I might be forced to switch. But I do like being able to grep my email, so I probably never will.

So email is (mostly) taken care of. What about my writings? Well, they are both the most important thing, and the biggest weak spot. If soy crashes, I have everything in /home backed up on pboothe.dyndns.org (my research machine), and I should also use rsync to get it all on slick (my iBook). But that’s a lame way of doing things. For a while I thought that e2 might serve as my repository, but they really don’t want everything, and many of my documents are not HTML. So what’s a guy to do?

Well?

Any ideas?

My current theory is that I need to submit all of my writings of any quality to various publications of low repute and get these publications to print my writings, thus preserving them both on those servers and findable at scholar.google.com. But that, gains, is a weakass solution. It only preserves my high quality writings. What about my low quality writings? My code snippets? The applet I made for Mark E Phair?

Well, as long as I can get it onto the web, my problems are being partially solved by Brewster Kahle. But that, again, is somewhat weak. He’s not hosting my backup because he wants to hold my data, and I can’t tell him to update his server’s copy to the most recent version. So what I want now is some way of saving random text files to remote servers. Perhaps a free internet-wide DARCS/Arch/Subversion server that would allow me to upload as much as I wanted as long as it was ASCII text and each file was under 250k. I wonder if there’s money to be made there?

With this system and the email system, security starts being important. I want to backup my writings, but some of them I don’t want to be world readable – student’s grades and code submissions end up on the department managed servers, so that’s okay, but I don’t want the internet to read my email, and my bbgp work involves a lot of data that is kind of proprietary. Which sucks, but the alternative was not getting experimental data from those points of the internet.

So I want to be able to save my documents to a remote server, preferably versioned, and exert some control over who is allowed to read them and exert that control in a non-sucky way.

Hey! Anybody who wants to start a .com that does that, please let me beta test! Remember that open APIs are important here!

This is not what people generally refer to when they talk about distributed systems, but it seems to be where the future is heading. My journal is livejournal, my bookmarks are starting to be del.icio.us, my photo album may become flickr, and my email is being very carefully maintained by a professional. The only problem now is my original content that defies categorization – articles, code, makefiles, and random crap.