I don't know how you can implement AUP in the network in a reasonable way using the approach you propose. You move a centralized management problem into a much more distributed and harder to effect approach. Centralized? Where is the problem centralized? Unfortunately, it the Internet simply does not have a "center" any more, not in a global sense. I still don't understand the problem. You split your block into a non-AUP and AUP block, and advertise the ranges appropriately. This is consistent with the existing model of how routes are advertised and approved, and requires no changes in the system. We use the same techniques we are doing today. If we reduce specific advertisements from 500 to 2, instead of 500 to 1, we still win. Unfortunately, given the way the US agency AUP's are written, this is almost impossible. For starters, the AUP's talk about content of traffic, which has very little relationship to network numbers. Tell me, Milo, should I consider HP (including HP Labs) and Xerox (including PARC) as research or commercial? And even implementing (the rather lame) approximation by destination network number is very difficult in a CIDR world - it requires ESP on the part of a network providers to determine if a customer is and will always be AUP- compliant when assigning their network numbers. And, of course, this all has to be done at network number assignment time and is effectively cast in stone if the site doesn't want to renumber later if they change AUP's. We at BARRNet have tried very hard to do number allocations along AUP lines, and it has not been either easy or perfect. Saying: If we reduce specific advertisements from 500 to 2, instead of 500 to 1, we still win. is a nice idea, but it doesn't take into consideration those blocks which have been allocated in a mix-and-match way to both AUP-conscious and AUP-ignorant sites. And, of course, whose AUP are we talking about? NSF's? NSI's? What about places outside of the U.S. that may have their own AUP's? Does every provider in the world have to allocate a separate sub-block for every AUP in the world? Until we have something like SDRP or until the politicians are taught what is is not implementatble, NSFNET-like AUP's are going to be problematic. --Vince