- From: Lorrie Cranor <lorrie@cs.cmu.edu>
- Date: Thu, 19 Feb 2004 15:24:45 -0500
- To: 'public-p3p-spec' <public-p3p-spec@w3.org>
OK, responding to Giles and Eric's concerns... how about: <p>A piece of data X is said to be <i>linked</i> to a cookie Y if at least one of the following activities may take place as a result of cookie Y being replayed:</p> <ul> <li>X is retrieved from a persistent data store or archival media.</li> <li>Information identifiable with the user -- including but not limited to data entered into forms, IP address, clickstream data, and client events -- is added to a record, data structure, or file in which X is stored. </li> </ul> On Feb 17, 2004, at 8:10 AM, Giles Hogben wrote: > > >> **On Feb 16, 2004, at 4:59 AM, Giles Hogben wrote: >> ** >> **> >> **> Some comments: >> **> 1. I don't think the requirement that it be stored as a particular >> **> database >> **> record is valid. I think that linkability should be described >> **> independently >> **> of the technical architecture used. This is why I tried to >> **describe it >> **> in >> **> terms of the intentions and proportionality. >> ** >> **This actually goes to the heart of what I was trying to do... >> **I wanted >> **to define "linkable" independently of technical architecture >> **but define >> **"linked" more narrowly. So far I haven't come up with an >> **example of an >> **architecture in which we would want to say that data is >> **linked and does >> **not involve either triggering a database retrieval or >> **storage. Perhaps >> **you have an example? >> ** > > Cookies and files used in forensics are not linked to a database. > Server > logs are not really databases? > >> **> 2. You do not mention the use of referers to link cookies >> together. >> ** >> **I will add that. >> ** >> **> 3. I think the examples given are simpler than those I gave. >> **> >> ** >> **Is that a good thing or a bad thing? >> ** > That is a good thing. > >> ** >> **Lorrie >> ** >> ** >
Received on Thursday, 19 February 2004 15:24:16 UTC