eng, Dai

LIP: A Lifetime and Popularity Based Ranking Approach to Filter out Fake Files in P2P File Sharing Systems

P2P systems have incentive mechanisms to discourage free-riding.  People who contribute
get added benefits.  But when you discourage free-riding people tend to create fake
files in order to increase their value in the network.

Primary idea: Re-name file to name of popular file.  

Designed two detectors based on file lifetime (avg. retention time) and popularity.

Identify files based on hash of name and file contents.  Attempt to detect fake files
before and after legitimate file's release.

They had data from Maze, picked 5 files and analyzed their ability to detect fake versions
of those files.

Ideas/comments:
---------------
What about worms that, on download, replicate and share themselves in p2p systems.

Don't address differences between structured and unstructured overlays.  Need to collect
lots of log data.  What if people lie about log data and collusion?

Can we infer when files are deleted from the ISP level?

Can we do streaming media to determine that we're downloading the right file?

Does this provide extra evidence of "illegal" sharing?