[Tech] Some issues and considerations

Michael Rogers m.rogers at cs.ucl.ac.uk
Mon Jan 21 10:09:46 GMT 2008


Hi Stephen,

> In the UK, a new law has been brought in which would make 
> it a crime for a suspect who has encrypted data on his computer to fail 
> to reveal the password to the police.

The police can only issue a disclosure order if they believe "on reasonable 
grounds... that a key to the protected information is on the possession of" 
the person in question. I'm not a lawyer but that suggests a defence on the 
basis that you don't have, and have never had, the key in question.

http://www.opsi.gov.uk/acts/acts2000/ukpga_20000023_en_8#pt3-pb1-l1g49

> And in the USA, users with encrypted content are 
> curently protected by a constitutional right to privacy which prevents 
> police from compelling them to disclose their passwords. But right now 
> even that right is being put into question with an important test case 
> taking place (see link below)...

The test case relates to users who know a password but refuse to disclose 
it; it does not relate to users who don't know a decryption key (which 
would be too long for most people to memorise anyway).
 
> It is also important to point 
> out that at least in the USA the NSA avails itself to the use of advanced 
> programs that can carry out advanced 'dictionary analysis' to permute 
> nearly every possible combination of letters and numbers for a 'brute 
> force' attack to discover the password for an encrypted file - a process 
> that can take years.

Again, this is not strictly relevant - a password can be cracked using 
brute force, but a 256-bit encryption key can't.

> Secondly, there 
> are government installations in the UK (for instance a new MI6 building 
> on the London enbankment, which has the national internet traffic 
> channeled through it) which carry out surveillance of communications 
> including internet communications. This surveillance includes not just 
> keyword profiling but also several other different kinds of intelligent 
> and statistical analysis of the traffic itself, even where encrypted 
> files are involved, and an significant intelligence perspective can be 
> obtained in this way.

Yes, traffic analysis is a very important issue. Freenet does its best to 
frustrate traffic analysis by using a transport protocol with no 
unencrypted header fields, delaying and coalescing small packets to 
disguise timing patterns, and padding packets to disguise the size of the 
payload. Nevertheless I'm sure it's possible to design a rule for a deep 
packet inspection engine that will identify Freenet traffic.

A possible direction for future research would be hiding Freenet traffic 
inside other application-layer protocols (HTTP, BitTorrent, RTP etc).

Cheers,
Michael



More information about the Tech mailing list