I was experimenting with tumblr today. Tumblr is a service for publishing Tumblelogs -- weblogs optimised for frequent short posts that are just snippets, quotes or quick links. Given that this is how Dave Winer has been blogging since year dot, one could be mistaken for wondering what the fuss is about.
What makes Tumblr interesting is that it gives you almost nothing that existing blog software doesn't offer. The entire value of the service seems wrapped up in the toolbar which is presented to you when you want to post.
The toolbar tells you what sort of things Tumblr wants to publish for you. There's no "thousand-word essay" option. The toolbar tells you explicitly that this service is just for notes, quotes and random YouTube finds. Which is perfect for me because my regular blog is explicitly designed for thousand-word essays. Tumblr-style quick posts just don't fit into its format.
I could redesign my blog to aggregate my short posts in the same way John Gruber does with his linked list on Daring Fireball, but frankly the pile of maintenance tasks I have outstanding for this blog has reached a height at which I just can't be arsed doing any of them.
And then there's the interface. A while back Gruber linked to a Tantek Çelik essay describing interfaces in terms of the cognitive load imposed by each decision. My blog software presents me with a highly flexible generic interface for putting content online that could be used to write lots of small posts instead of the occasional long one. Tumblr asks me which of six different types of post I want to write, a decision I've already made, then gives me the minimum number of decisions I have to make to get that particular kind of information online.
The fact that all both tools do is take a chunk of HTML and arrange it in reverse chronological order is irrelevant.
Function follows form.
I was also thinking today about how diffuse all my "stuff" has become online. I blog on my own server. I put photos on Flickr. I post links to del.icio.us. And now, it seems, I tumble onto Tumblr.
I'm spreading my eggs across too many baskets. The more of these services that have my data, the more I'm bringing forward the day that I'll have to waste a weekend migrating from one to another because the service has started to suck, or something better has come along, or they've just gone out of business. Worse, I'm increasing the likelihood that some of my data just vanishes altogether.
And, of course, if I want to switch services for some important but non-critical reason, it's too much effort.
To get out of this trap I either have to to move all my data back onto my own server, where I lose the cool interfaces and social functionality, and where have to spend too much time being a sysadmin. This would be even more of a hassle in the long run.
All these services are basically social content management systems, they just manage different types of content and thus need different interfaces. It's not hard to imagine an abstract content model that could encompass the data I keep on all of those services. (What is Atom, if not the XML serialization of just such a model?)
What if all these services had the option to store their data in standard formats, elsewhere in the cloud? My photos don't need to live on Flickr (although they might need to cache copies), Flickr just needs to know where to find them. Meanwhile all my data is in one place. I can back it all up in one operation. And if I wanted to move my del.icio.us data into Google Bookmarks, I would just need to tell the latter where to look. If I wanted to access the same data from both services, that should work too.
Even better, give me a system where I can manage all that data on my own PC, where I can choose which parts of my digital life go into the cloud, and which subset of that is visible to Flickr, or Zoomr, or del.icio.us, or YouTube, or Google Video, or Tumblr.
Apparently there's glimmers of this starting to surface: WebFS: a Web of Data, Time for an Internet Filesystem?.