The Politics of Systems

EDIT (23/01/2015): Changed some text to make clear that you can still run Netvizz by grabbing the source code, registering a new app, and running it in developer mode.

EDIT (25/01/2015): I have written a small install guide for the source code on github. I’m unfortunately unable to help with individual problems, if you’re unfamiliar with server administration, your department’s tech support team should be able to help.

EDIT (28/01/2015): Since Facebook has changed the way apps are created, you can apparently no longer run just scripts requiring extended permissions in newly created apps, even in developer mode (making my source code useless for you). I have therefore whipped up a version of Netvizz that can only do pages and groups without requiring extended permissions. Since this does not have to go through review, you can use the app directly here.

EDIT (29/01/2015): Facebook’s policy review has accepted the new version of Netvizz (with personal network functions removed) and the app is again accessible here. API v1.0 is still going to be retired in April and this may pose problems, but this is something for another day.

EDIT (02/05/2015): API v1.0 has now been retired, but a new version of Netvizz (v1.2) has survived the changes and should continue functioning in the foreseeable future. Personal and group friendship networks are gone for good.

Original Post:

Today Netvizz, an app that allows researchers to download data from the Facebook platform, was suspended by the company and I received a mail explaining why:

Your app is violating the following Platform Policies:

Platform Policy Section 1: Build a quality product.
Platform Policy 1.1: Build an app that is stable and easily navigable.

Platform Policy 3.3: Only use friend data (including friends list) in the person’s experience in your app.

To clarify, your app should be stable and easy to use and shouldn’t stall escessively. Additionally, you should not allow friend data export, even if that data is anonymized. You can access the full list of our Platform Policies here: https://developers.facebook.com/policy/.

Since Facebook has recently been very preoccupied with app privacy – for very good reasons actually – this does not come as a surprise. I have been anticipating API changes and the retirement of version 1.0 that comes with some very sensible changes in how data is delivered to platform apps for a while. Apps are clearly one of the biggest problems when it comes to Facebook’s privacy puzzle and most changes make a lot of sense. As Bernie Hogan wrote here, friendship connections are one of the casualties, as they will no longer be available to apps at all (v2.2 no longer makes them available). I was hoping to stall a little by moving to API v2.0, which still runs until April 2016, but this seems no longer viable after this morning’s news. As much as I agree with the general changes Facebook is making, I think it is a real shame that the analytical possibilities apps like Netvizz afford will no longer be available to researchers.

Over its roughly five year life span, what started as an inquiry into Facebook’s API, ultimately had over 60K unique users and analyzing their friendship network has been the start into graph analysis for many people. GetNet, a modified version of Netvizz, was used by Lada Adamic in her highly successful Coursera MOOC, allowing students to look at a network they are intimately familiar with, making network visualization much more tangible. GetNet actually still works, but will probably break in April 2015, if not shut down earlier.

For me personally, Netvizz has been a ambivalent project. On the one side, I enjoyed the tinkering with the API, but on the other, maintaining a complex tool in my spare time has often been a challenge. As anybody who offers software online for free will tell you, the mass of not always friendly emails can be daunting. I’m also not a computer scientist and I work in a humanities department, where technical work does not really count in performance reviews.

But the real problem with the current situation has little to do with me and much more with the many courses and research projects that have been relying on Netvizz. They are left out in the cold. So here are some elements that will hopefully help them deal with the situation:

• Despite my hesitation to make software public that can be used very easily to download large amounts of non-anonymized data, there is so much code already in the wild that another set of scripts is not going to make much of a difference. I’m therefore making Netvizz’ source code publicly available. This should allow research projects relying on Netvizz to take the source code, register their own app at developers.facebook.com and run it in developer mode (just to make this clear, since I am the developer, I can actually still run the app, but it is no longer publicly available), which should work until April 30, 2015, the day v1.0 of the API retires. I apologize for the crappy code quality, this is one of those projects that grow and grow and never get a real redesign.
• I will try to enter into further communication with Facebook to see what can be done, but I don’t expect much from that.
• If that does not work, I will submit a version of Netvizz version for review that excludes personal network features and focuses on pages and groups. It’s still going to “stall excessively”, though, since it gets a lot of data.

I have no idea how long any of this make take. In the meantime, check out this list for alternatives, most of which hopefully still work. But make no mistake: this may well be the beginning of the end for external Facebook research with digital methods.

Post filed under facebook, method, social networks, softwareproject, surveillance.