Is this more or is it less effective than a large /etc/hosts blacklist? It doesn't seem to be doing anything special like the Privacy Badger. The only relevant sections in the paper I saw were
We implement an API based on Google Safe Browsing, a
mechanism for efficient URL-based blocklist updates and
lookups [9]. We use a subset of approximately 1500 domains
from Disconnect’s privacy-oriented blocklist to identify
these unsafe origins [10]. We update the blocklist every 45
minutes to minimize the effects of incorrect blocklist
entries.
Another open challenge is applying Tracking Protection only
to third-party content. We can avoiding cross-site tracking
by blocking content from high-volume sites such as
facebook.com without breaking them when visited directly.
Heuristics such as the Public Suffix List4 can help better
determine the set of domains that are considered first-
party.