[AusNOG] BSA Act part 3 (Filtering)

[Matthew Moyle-Croft]

Bevan Slattery wrote:
>
> Yes Bevan.   I said before you incorrectly asserted that because P2P 
> DPI and caching is doable that it suddenly is easy to classify the 
> CONTENT according moral standards.
> P2P DPI and caching doesn't have to do nearly the same work.
>
> Again MMC -- you are putting words in my mouth.  I said that it would 
> be quite capable of intercepting a 'flow' and assess whether that 
> portion of flow is on the 'list'.  I wasn't saying you should 
> intercept the flow, reconstruct it get some intelligent software to 
> consider it's classification based on the number of flesh colored 
> pixels and moaning.  Are you suggesting that with the current P2P 
> caching systems out there they can't determine the file (or portion 
> thereof) from the flows so it can provide a locally cached copy of 
> it?  Seriously?  Then how does a P2P caching system work?  It just 
> inserts some locally cached content randomly in the hope that it's the 
> same as being sought?  Of course not.  Based upon flows it can 
> identify what is actually being transmitted and then it substitute 
> it's local copy.
>
(1) The DPI identification works statistically - doesn't need to look at 
content just flows and then pass it off to the caching side of things.
(2) The analysis (identifying a file) only works on a subset of flows 
(ie.  not on HTTP).
>
>  
>
> The difference here MMC is that it doesn't CLASSIFY the CONTENT, it 
> IDENTIFIES the FILE and checks it against a list.
>
Yep.  Much easier than having to look at each flow and classify content 
by morals.   It's just a list of hashes you've cached.

As you said before:

"It is PRACTICALLY impossible to guarantee a clean feed. If you can't 
guarantee a clean

feed, then you are providing parents with a false sense of security
thinking they have."

Why?  Because no one's developed a good, line rate, classify all content by morals system.  Because it's HARD.

MMC




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ausnog.net/pipermail/ausnog/attachments/20080102/8429bf94/attachment.html>