"Warrants authorising phone taps treble"
Tom Thomson
ukcrypto at chiark.greenend.org.uk
Sat, 2 Feb 2008 20:15:02 -0000
Peter Fairbrother wrote:
> I'm not entirely sure that wording works for search strings.
>
> First let me say I have no idea how Google actually works, except that
> they have lots and lots of computers, but it could be like this:
>
> Suppose they have one computer with all the hits for "illegal" on it,
> and another with the hits for "interception". When someone searches for
> "illegal interception" a request-receiving computer routes the request
> to the two computers which compare files, or file numbers, and one of
> these outputs results which match, either directly or through the
> receiving computer.#
>
> Now access to the files in both computers has been obtained, so maybe
> Google have to identify both computers, and maybe thereby the search
> string, uniquely so if each computer only contains the hits for one word
> - which is quite possible considering the number of computers Google has.
>
> A bit fanciful perhaps, but it shows the difficulty.
I don't see any difficulty at all. The user issues a request to an
individul computer (identified by ip address, maybe also by domain name) -
that's the only communication the user takes part in. Yes, that computer
may in turn communicate with other computers is irrelevant, the user does
not originate any pakets addressed to those computers so a request for comms
data for that user's communications isn't going to get any information about
those computers. If on the other hand there were a request for information
as to how the request was dealt with by the computer that received it, a
reasonable response to which would identify those other computers, that
would not be a request for communications data because it would require the
programme to be identified regardless of whether it was the only programme
stored on that apparatus, which is explicitly not comms data.
M.