What do you think about communications data collection and storage?

[Roland Perry]

In article 
<a2b6592c0904301251j630441e2m28491bdbee2d7b8c@mail.gmail.com>, Igor 
Mozolevsky <igor@hybrid-lab.co.uk> writes
>>> So this "communication data", is that the link layer data or the
>>> application layer data?
>>
>> It's any data that's necessary to get the information from A to B, plus a
>> few things that aren't, like the date and time (modulo if the date and time
>> are wildly wrong some communications might be rejected).
>
>So by that definition, you not only have to log source and destination
>addresses of each packet, but also what router it came from and what
>router it went out on? What if it were a GRE tunnel- guilt by
>association?

One of the fallacies (which I've been pointing out for at least the last 
eight years) is that the communications data for a particular 
communication is a static thing. Of course, some of it is (the ultimate 
destination, for example, and the time it was sent if we can agree for a 
minute that it's part of the dataset) but there's no reason why it 
shouldn't have *other* communications data temporarily associated with 
it for one leg of its journey, which is then discarded later in its 
travels[1].

A bit like relativity, what you see depends on where the observer is.

To return to a postal/courier analogy - the better tracking sites tell 
you which depot the parcel is currently in, whether it's on a lorry and 
if so which depot that lorry is heading for. But unless someone is 
gathering it centrally (which is much easier for a closed system like a 
courier company) then it would be invisible to you when you inspected 
the parcel on the doorstep.

[1] Or in an email header, "added" to the top and perhaps even form part 
of the 'content' as observed later in its journey.

-- 
Roland Perry