Simple Access Control

At the IETF meeting, I raised the possibility of adopting
a really simple access control model -- for example

1. No principal hierarchy (i.e. user/group)
2. Principals identified only by digital certificates
3. Small fixed set of rights
4. No ACL inheritance (either static or dynamic)
5. Only creator of resource can set its ACL

As I understood it, a number of people seemed to assert,
that independent of the exact details, there were inherent
unresolvable problems in adopting something simple.

Would someone be kind enough to elucidate those
reasons?

  -- Ellis