Re: help me!!!

Rigo Wenning <[email protected]> Sat, 15 Apr 2006 15:06:07 +0200
Newsgroups gmane.comp.web.p3p.devel
Organization W3C
Message-ID <200604151506.14467@rigo>
--nextPart1854114.2CW7dY1zAx
Content-Type: text/plain;
  charset="utf-8"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

Dear Nguyen,=20

Am Friday 14 April 2006 14:41, sprach Nguyen Viet Ha:
> I have been trying to understand what is involved with P3P and I was
> dubious about whether a single line of code in a header would solve any
> problem, and even worse, if its insertion would open a whole can of
> compliance worms.

P3P is a protocol and a data format to tell a user's computer what personal=
=20
data is collected about him and how this information is processed and=20
transferred to third parties, where the user can complain about things etc.=
=2E=20
Read the introduction of the specification to get  a feeling. Often, for=20
cookies and other protocol information or forms, users don't really know=20
what's happening to the information they are giving away. P3P is giving the=
m=20
the necessary information. User agents are displaying the information to th=
e=20
user in many ways. E.g. Privacy Bird is an icon that turns red once the=20
privacy policy does not correspond to the user's preset preferences.=20
>
> >From what I've read regarding P3P there are 3 levels of policy:
>
> 1.	Compact policy - that can be inserted into a header for example
> - I think this is what eGS have suggested.

Note that the compact policy is only an abbreviated form of the full XML=20
policy. There MUST NOT BE a mismatch. But compact policies are less=20
expressive, so normally the compact policy overstates a bit.

> 2.	XML policy for machine reading (which can be referenced on each
> page, modified for each page, or modified for sections)

This is the normative one for P3P. It MUST BE also present if you implement=
=20
compact policies to be conformant to P3P.

> 3.	Text policy for human readability

Yes, this is the normal privacy policy that explains the use of personal da=
ta,=20
retention etc in your system and company/organization. The P3P specificatio=
n=20
is a good checklist of questions that should be answered in the human=20
readable policy. But there exist also software that translates P3P back to=
=20
human readable policies. There should be no mismatch between the human=20
readable policy and the XML Policy.
>
>
>
> I understand that P3P is good practice and not a technical nor legal
> requirement, but,

This is not really a true statement. One should NOT lie in those policies a=
s=20
this might have legal consequences, especially when confronted with=20
consumers.
>
> What needs clarification is
>
> -          I understand that even though our system is JSP we can still
> include the required HTTP declaration in a header. - is that right?

In fact, in the HTTP-header, you convey not only the compact tokens, but al=
so=20
the information on where to find the policy reference file. A user agent wi=
ll=20
analyze the header, recognize and use the tokens, look for the policy=20
reference file. The policy reference file contains a link to the policy and=
=20
the user will fetch that policy and analyze it. Normal caching is 24 hours,=
=20
but you can tune to longer. The W3C-site has caching for one week.
>
> -          Is there a genuine technical requirement to have a P3P
> policy, compact or otherwise, ie: will it be a significant benefit to
> our system?

No, there is no technical requirement. The benefit is greater trust from th=
e=20
users that know what their data is used for. Some browsers also handle=20
cookies depending on the presence of a P3P Policy (compact and full). If a=
=20
thirdparty cookie has no P3P policy, it will be blocked in those browsers.
>
> -          Can a compact policy statement code in a header stand alone
> as the privacy policy in an application, or it will need the other XML
> and text policies to reference to?

No, this is not conformant. A performance improvement can be done if the P3=
P=20
Policy is in the same file as the Policy reference file. This is possible f=
or=20
simple policies.
>
> -          If a line of code can stand alone in the headers - what
> should that code be? (verify the code Gareth Boden has suggested)

The description of the code is dependend on your usage of personal=20
information. There MUST be a link to a policy reference file and optionally=
,=20
there can be the compact tokens. (For cookies you should also have the=20
tokens). There are Privacy Policy editors that help you to write your polic=
y=20
and the compact tokens. See: http://www.w3.org/P3P/implementations

>
> -          Will the line of code impact in others ways - new
> accessibility issues for eg: other browser problems / user agents.  How
> much back testing will be involved?

A user agent that does not know anything about P3P will just ignore all tha=
t=20
data. User agents/browsers that are P3P enabled will behave dependent on th=
e=20
privacy policy that they find. If the policy announces lots of data=20
collection and unlimited transfer to third parties, the browser might block=
=20
the site. If the policy is privacy friendly, the browser might open up more=
=20
than it would without P3P policy.=20

Best,=20
=2D-=20
Rigo Wenning            W3C/ERCIM
Staff Counsel           Privacy Activity Lead
mail:[email protected]        2004, Routes des Lucioles
http://www.w3.org/      F-06902 Sophia Antipolis



--nextPart1854114.2CW7dY1zAx
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)

iD8DBQBEQO/GCRlGO2Cxp/0RAu4/AJ41hIMYuigv5qdjadvIVR06cXujnwCeMVHE
Z9JT5SmKRZI5ZkRtxj3udlA=
=GDms
-----END PGP SIGNATURE-----

--nextPart1854114.2CW7dY1zAx--