A sample P3P file (Cont.) - Page 3
August 12, 2002
If you're not familiar with XML, just think of it as an
anally-retentive version of HTML with an infinite number of tags.
It's unlikely that the original developers ever imagined it would
be described that way, and it's not even 100% accurate, but it's
a useful description for the real world. The two important things
to remember about XML are that it always requires closing tags
and it absolutely will not tolerate errors.
The file starts with the usual introductions - the XML version and
P3P version that apply. This is quite similar to the Doctype
declaration included just before the start of an HTML file. Then
comes the URL and name of the equivalent human-readable privacy
policy.
Now we get the first block of data, highlighted in the example in
blue. This is the description and details of the company or
individual making the privacy statement, otherwise known as the
Entity. In this case the details include a full postal address as
well as email.
Our next block of data is the small Access section, highlighted
in red. This informs users how much of the information they are
allowed to access from the records the company holds about them.
Entries here vary from "all" to "none". In
theory it's one of the most important elements of file, but of
course if it's set to "none" it's no great help to users.
Third in our breakdown is the Disputes-group, highlighted in
green. Here the company (entity) explains what users can do if
they have a dispute over the privacy policy of the site. There
are two main elements to this - the "service" section
and "independent" section. The service section gives
information for settling disputes directly between the user and
the site without third parties becoming involved. The independent
section gives details of third parties that can become involved
in the dispute if the user wishes.
In the service section a company email address is usually
specified, that can be used to contact the company in the event of
a dispute. Often it's a repeat of the email address given earlier
in the Entity section. Some sites merely refer to a relevant HTML
page.
The independent section gives contact details for the independent
privacy organisation that may be contacted in the event of a
dispute. Often it also includes a URL and a long description of
the independent organisation's aims. Two relevant organisations
are BBBOnLine and TRUSTe. You might like to note that privacy
campaigners have sometimes disputed the independence of
corporate-funded privacy organisations.
Now we come to the final and most difficult section of the file,
dealing with the site's collection and use of data, including
from cookies. In our example, this is presented in two sections,
a purple section and final blue section, but you'll find many
variations in how real sites deal with this element of the file.
Sometimes there's just one section covering almost everything.
Sometimes there's an additional section covering the information
collected from registered users, and many times cookies are given
their own section rather than being combined with anything else.
It can be difficult to work out what each section is for, and
either the P3P specification itself or the sites using it seem to
get in a bit of a muddle when it comes to differentiating between
information collected from all users and information collected
from registered users.
Let's investigate our example in more detail.
The first (purple) section deals with information collected from
all users, and the second (blue) section deals with users who go
on to make a transaction with the site. Each section begins with
the <STATEMENT> tag and ends with </STATEMENT>.
The first section starts off with the <CONSEQUENCE> which
isn't really a consequence, it's actually the reason why the
information is collected. Then we have a <PURPOSE> which
is a more quantitative assessment of the reasons for collecting
the information and its uses. The entries here are selections
from multiple choice options, so are more strictly defined than
the Consequence, which can easily lapse into meaningless marketing
talk.
In the <RECIPIENT> area we get to learn who receives the
information, and under <RETENTION> we find out how long
it's kept for. Under <DATA-GROUP> is the list of collected
data. <CATEGORIES> tells us the broad categories of
information collected.
The second (blue) section follows a similar pattern, but here we
have two sets of data in the <DATA-GROUP>. The first
includes clickstream data, in other words the sequence of pages
the visitor clicks through, and cookies. It doesn't exactly say
what information is collected, but it does say what it's used for,
under <CATEGORIES>. Then a final set of data is mentioned
close to the end, covering information collected from HTTP headers,
and any search-strings the users enters into the internal search
mechanism.
That's enough of an introduction to the content, now let's look
at where the file goes.
Policy Reference File
The main file analysed on the previous page can be slotted
anywhere on a site, and in practice these files are found in
many places under different filenames. Continuity and
predictability are provided by the Policy Reference File, which
is a small XML file that points the way to the big one. This is
always called p3p.xml and placed in a folder named w3c, directly
after root. So for www.zzzz.com the file address would be
www.zzzz.com/w3c/p3p.xml
Here's an example of a p3p.xml file:
<?xml version="1.0" ?>
- <META xmlns="http://www.w3.org/2000/12/P3Pv1">
- <POLICY-REFERENCES>
- <POLICY-REF about="http://www.tinhat.com/privacy/policy.xml">
<INCLUDE>/*</INCLUDE>
</POLICY-REF>
</POLICY-REFERENCES>
</META>
The important line is the one specifying the location of the big
XML file.
Under the full W3C specification, there are some alternatives to
this system, which is called a "well-known location,"
but in practice it's the method most often used.
If your site delivers cookies, there's one more stage to go - the creation of a Compact Policy.
A sample P3P file - Page 2
P3P - Platform for Privacy Preferences
Compact Policy - Page 4
|