Hi

Well, if you have a system capable of identifying similar files (in
case of media files, this would be based on fingerprinting), it means
that you are capable of querying a legitimate media database, which
opens the gates to legitimate filtered peer to peer. The press release
seems to indicate a focus on media files.

I guess the system relies on fingerprinting for achieving:
* encoding-independant media file identification; using a transcoding
layer, one could theorically download from multiple sources of
different encoding; this would mean global downsampling though
* name/metadata-independant media file identification
* media tastes aggregation: somebody that listens to abstract hip hop
has probably much in common with the people downloading the same
thing, who probably download the same type of medias; somebody that
watches Prison break has the former episodes... Using this
similarities, you can reduce the files and peers discovery problems

In another topic, what is the statistical probability that two chunks
from two different files (in a virtually infinite files database) can
have the same binary content? If the chunks are small enough, why not
? For instance, two zipped files could have the same binary encoding
with different huffman dictionaries... We can think of pure redundancy
too: how many programs share the same installer software? A lot. How
many programs reuse code? A lot too...
Of course doing a query on every chunk would be overkill...

Cheers

Florent
_______________________________________________
p2p-hackers mailing list
[EMAIL PROTECTED]
http://lists.zooko.com/mailman/listinfo/p2p-hackers

Reply via email to