How to Move A Blog from TypePad to Tumblr (Spoiler Alert: It Sucks)

I spent an embarrassing amount of time in the last few months porting Roger’s blog (http://informationarbitrage.com) from TypePad to Tumblr.  As much as I love Tumblr, this was not an easy process.  My eventual workflow was:

Blog Content

• Export from TypePad
• Convert Typepad file to Blogger format (note: although this app claims a 1MB file size limit, it handled my 6MB file just fine)
• Create a Blogger account and upload the converted file
• Import from Blogger to Tumblr via a 3rd party site (thank you Terry Hung!)
• Use some nifty scripting hacks cobbled together by the team at IAV/Kinetic to crawl informationarbitrage.com, collecting a txt file of every link on the site
• Convert the txt file to csv
• Create a job on Mechanical Turk that asks Turks to go to a given link (the old TypePad link) and note the post title and date. Then, using Google or the Tumblr archive feature, search for the post on our Tumblr domain with the same date and title, and then feed us the corresponding URL.  We paid $0.10/link, which given the uncanny speed of the MT workforce amounted to an effective hourly rate of $18/hour.  It took all of 3 days to map ~600 links.  For those who are interested, here’s what the HIT (human intelligence task, in MT-speak) looked like:
• Next, do some data munging to make sure the newly created linkmap (TypePad > Tumblr) is clean
• Make friends with the (current and former) Tumblr team. This was by far the easiest step. those guys/girls are all sorts of awesome.  On that note, try GroupMe (like BBM, but cross-platform) and Instapaper (take your unread web pages with you)
• Ask said friends to bulk upload a file consisting of some 614 redirects
• Breathe out

Blog Comments

Disqus was easy enough. Luckily, they had just released some migration tools right around the time I was doing this (although the documentation was not at all clear enough to a non-programmer like myself).  Of course, I ran into problems, but Daniel Ha and his team swooped in to save the day.  But what about the older, pre-Disqus comments?  Well, if you have a way to do that, please, please, for the love of god, let me know.

Parting Thoughts

• While I think it’s short-sighted to choose your blogging platform by ease of import (sorry Posterous), I wouldn’t fault someone for not wanting to go through this hellish process.  I love Tumblr.  I think Roger’s enjoying it too.  But man, this just sucked.
• It’s possible that Tumblr consciously de-prioritized building a tool to import old blog content. Their platform is decidedly different from traditional blogging, and encouraging people to pipe in their old content might also encourage them to continue blogging in their traditional manner.  If that’s the case, I sympathize with the motivation, but I think they’d get desired behavior and customer acquisition if they offered the tool wrapped in fun documentation explaining how Tumblr is different and why that makes it so engaging.
• In adding legacy TypePad comments to older posts, I came across an incredibly frustrating bug in Tumblr.  There is an undocumented (so far as I can tell) size limit on text posts.  When you try to post an over-length post, you don’t get a warning, you get a heart attack, because Tumblr will simply delete your body text and save the post as empty.  Hitting the back button will not recover your text. It’s just gone.  This isn’t a CSS/HTML issue; it occurs on the default Tumblr theme.  Nor is it a browser issue; it occurs on Safari, Chrome and Firefox alike. Here are some screenshots to demonstrate: