IP addresses and personal information (was Re: [LINK] Fwd: On Line Opinion - 16 February 2007)

[Craig Sanders]

On Sat, Feb 24, 2007 at 10:31:55AM +1000, Irene Graham wrote:
> What I do not understand is how can they classify web site visitors
> into those or any other categories -if- they are only collecting
> completely anonymised information from ISPs?

guesswork.

web stats programs make the (often erroneous) assumption that multiple
page requests from the same IP address within a short time of each other
equate to one person (often called a "visitor").

(this is often wrong for many reasons including: because someone on a
dynamic IP may log out and another person log in seconds later and get
the same IP allocated to them, and it also fails to take into account
proxy servers which service requests for many - perhaps thousands - of
users simultaneously. of course, this latter caveat doesn't apply when
you're analysing proxy logs)

so, one way of doing it is for the script to notice that IP address
a.a.a.a visits a site about, e.g., pop stars popular with teenage
girls. it then records that as "IP address a.a.a.a at such-and-such a
time is called 'Person A'", and flags it as being in the teenage girl
demographic. the next time it sees that IP address in the logs, if it is
within a short time of the previous request then it assumes that it is
the same person and records that "'Person A' visited, e.g., a chatroom".


at no point is any individual ever identified. there's just an
assumption that a record of a particular IP address visiting a
particular site a) equates to a single (unidentified) individual, and b)
places that individual in a particular demographic group.

all very fuzzy, and leading to some pretty graphs that may or may not
have any proximate relationship with reality.

in short: it's just web stats.  a steaming pile of bovine excrement.

craig

-- 
craig sanders <cas at taz.net.au>

Currently listening to: Vibrasphere - Nowhere

LSD melts in your mind, not in your hand.