Haystack Blog

Over the summer I had the sometime pleasure of building a research application on the Facebook platform.  In retrospect I’m not sure why more people aren’t doing this — we were able to get thousands of users, lots of data, and roll out the biggest (and highest-stakes) user study of my life.

There are three approaches I would advocate for others thinking of doing research on Facebook:

1. Play by the rules. If your app can fit within the traditional bounds Facebook provides to apps, then definitely do it.  You will get a much larger user base, and it’s easier for your app to go viral.  And this being flu season, virality is all the rage.
2. Break the rules. As Dan Olsen says, feel free to reinvent the wheel in your quest for great research. Specifically I’m thinking using tools like Greasemonkey, Chickenfoot scripts and Firefox extensions to rewrite the Facebook UI to meet your needs. This has the downside of requiring a higher install cost for users.  If your innovation is in the technology and interface anyway, and you’re content to run a lightweight evaluation on a small group, then no big deal.
3. Take your ball and go to your own handball court. I haven’t tried Facebook Connect yet, but I think it will give you many of the upsides of Facebook integration by integrating Facebook services into your website intead of the other way around.  The downside here is that users need to remember to log in to your site whereas everyone ends up at Facebook eventually.  But again, if this helps you execute your research, the tradeoff may be worth it.

Fair warning: developing for the Facebook API is still a moving target. During the development period Facebook released its new profile redesign, changing tons of API calls in the process. Since I was using an open-source wrapper to the PHP library — and one which I have now discovered is devoid of developers — this caused considerable annoyance and breakage of code.  (”Oooh, so we’re posting to the News Feed of folks on the old profile but not the new profile?”)  This was not exactly welcome as the CHI deadline loomed — buyer beware.