I'll try to package it up into something useful then! In short, my system basically was just a highly parallelized graph crawler in node.js, which dumped into mongo for easy queries and analysis. I'll try to make it a bit more agnostic though.

One issue is that the Graph API for groups is in fact broken — it's request pagination system yields an inconsistent (and incomplete) data set. We've reported the bug[0], but I've been seeing reports of this for months.

I've actually been beginning to suspect some sort of deliberate crippling, though I haven't exactly been able to work out a motive :)

[0] https://developers.facebook.com/bugs/518983258224814/