On Thu, Jul 1, 2010 at 8:52 AM, Richard L. Barnes <rbarnes@xxxxxxx> wrote: > There's a difference, however, between ticking a box and having individual > user-attributable credentials. The two techniques are focused on different > goals, generically binding users to an AUP, without caring who they are, > versus being able to identify individual users on the network (with more > detail than a MAC address). > > The proposal here is the latter, which would seem to raise the question of > why individual user attribution is necessary, i.e., why anonymity in the > IETF network unacceptable -- even within the pool of IETF participants. > I agree with Richard's view here, and I suggest the following modifications to the proposed admission control: 1) Use only paper-provided slips to provide authentication credentials. There is no stated reason for associating specific registration data with the network authentication method and it is trivial to provide the slips of paper to anyone with a proper badge. Let the individual getting a slip shuffle the pile, get multiple slips every day, or do whatever else they would like to increase randomness. But start from the presumption that the admission control is to limit access to "registered attendees only" not to provide an association to registration data. 2) Favor anonymous MAC registration over portal methods. Set up a terminal or group of terminals which allow individuals to register their MAC addresses for access. Allow anyone with a badge access to those terminals, and do not collect information on which individual entered which MAC address. (The portal mechanism relies on a specific ordering of application protocol activity at best; at worst it provides a full-on monkey-in-the-middle. That should be a last resort) 3) For the portal, there is no reason to have the MAC-based permissions created to be time limited. If proper credentials from a slip of paper are entered, there is no reason not to treat this as equivalent to registration of the MAC address for the duration of the meeting. My personal preference is that this requirement from the host be politely declined as contrary to the usual operation of the IETF network. But if it is not going to be declined, then the admission control should not further the ability to associate specific credentials to individuals. Just two cents, Ted Hardie _______________________________________________ Ietf mailing list Ietf@xxxxxxxx https://www.ietf.org/mailman/listinfo/ietf