Vadim is right, in that peering and transit restrictions are
*address-based* and hence enforceable by routers, whilst AUPs are
*content-based* and so absolutely unenforceable by technical means.
In this regard, AUPs are like *most* laws in the non-virtual world - e.g.,
laws against speeding, unleashed dogs, homicide, fraud, etc. - in that the
behavior which they proscribe cannot in even the most elementally free
society be prevented; they rely instead for their observance on a (largely
unwritten) social contract between netizens and network providers, and
infractions can only be punished not prevented.
And just as only a fraction of all crimes are noticed and reported, and
only a fraction of them in the end punished, so it is with AUP violations.
Corporate intranets generally have exceedingly strict AUPs, and some
infractions can be punished by dismissal; AUPs are in fact very common.
The remarkable feature of the old NSFNET AUP was NOT the number of times it
was violated, but rather the lengths to which an extraordinary number of
decent corporate and private netizens went to ensure its observance. I
promulgated it and I know.
*Whenever* there is public funding of *any* activity, an AUP is attached to
oulaw the misuse of taxpayer money. Internet II is not exempt.
Some say that acceptance of an AUP is too high a price to pay in
circumscription of personal liberty for the benefits of public funding. I
understand and respect that view, but I don't agree.
Stephen Wolff Business Development
ciscoSystems 202 337 1213