They get their list of dropping names from Verisign, NSI and the other sources of such lists. Almost all are public, except Verisign, which only gives them to registrars, but you can get it from almost any registrar.
They then scrub the data sources like Overture, Alexa and the like, often using a large number of IP-diverse machines to avoid being shut down for slamming the sources.
The rest is simple database work, albeit with a LOT of data.
It's really not that hard, if you have the understanding of the lists, enough machines to get the data without being shut off, and are a decent database/web programmer.
Any analysis you can do above and beyond building your database is gravy. For DropShark, I do quite a lot, for example.
