I’ve always built my personal blog’s content and reputation at its own domain, completely under my control, despite being hosted on many different platforms and serving different roles over the years. It has never been a subdomain of any particular publishing platform or host.
Tumblr respects this. From day one, David and I gave it free custom-domain support, full HTML control, and no forced branding or advertising. But Tumblr is a hybrid of a blog-publishing platform and a social network that seems truly unique — the “pure” social networks aren’t nearly as willing to allow you to own your identity there.
Yes, that is a cool Tumblr feature — to permit you to use a custom domain, that even cloud application platforms like Google App Engine or Heroku do not permit totally (custom domains there are limited to subdomains).
But if I cannot get my content out of your platform, it really is not any different than those decried closed, proprietary networks. And even Facebook offers an downloadable export dump of your data (though admittedly flawed).
Tumblr offers two methods by which you can export blog data:
Tumblr API — (first I’ve seen of the new version now using OAuth, the old API is what I had been tinkering with). Write a script to traverse your post collection in 50 post chunks. Or if not coder fluent, then look to a third party to provide such a service (which many might exist, as I may just be ignorant of their existence).
An “official” beta Mac OS X backup/export app - accessed via the “Goodies” page
Attempts to apply either of these methods have not proven successful, at least for this blog (other blogs with less than 500 posts, both of the above procedures were fruitful). At a total post count exceeding 25K, perhaps this specific Tumblr is too gnarly a beast for the API or backup app.
It is vexing as the primary reason I desire a backup is so I can simply search my posts and be confident of the results. As it stands now, the Tumblr search is riddled with blindness. For awhile, I believe I rectified my dilemma by piping in the RSS here to Google Reader. And that worked fairly well, letting me execute snappy searches on my post collection here. But just like old fashioned Google search, Google Reader now fails, only bringing back recent results (a phenomenon I note that is true of other subscribed feeds also).