What is better from users point of view? The network is only as good as the users think it is?
There is a nexus in dealing with information. This nexus involves the interaction of Human Needs, Social Science and Information Technology.
Think of users are the intelligent devices at the periphery of the network.
In engineering, there are objectives which feeds into an engineering control and performance loop. The performance of the control/performance loop then affects the objectives for refining the system. By adding user objectives and user control and user defaults to this system, the user can get better results from the environment.
The Internet is not going to collapse. We should not institute per-packet charges; however, we should probably charge more for better qualities of service.
Congestion is an intrinsic feature of shared-media facilities. Users that want to get uncongested services have no way express that need in the current Internet without building a special network for these purposes. Having a congestion charge will influence people use the network at less congested times.
Is charging necessary? Congestion is with us and we will not be able to resolve this problem quickly. Quality of service allocation is going to happen, like it or not. (MCI/BT has announced the creation of such a backbone and Internet II's requirements include this.)
What about resource allocations that are not "critical"? It's an open issue.
This is a happy opportunity, not a grim curse.
Including the user in the feedback and resource allocation loop is good engineering practice.
What else is scarce?
Some domain names....
IP addressing and route advertisements.
Using financial incentives would be a better approach.
Links between domains of commercial influence
Survivability
What is the value of the content?
Jeff suggests that regulation may occur to regulate Internet Telephone to level the playing field between circuit providers and Internet providers. The actual problem is really Regulatory Anachronism.
Bill Simpson believes that the real cost of telephone network is the overhead of the personnel to have the user in the loop, not the cost of the regulation.
Bill also suggests that adding the user to the Internet engineering model is impractical since the congestion in the network is generally highly transient and would require the user to make almost real-time decisions.
Jeff agrees with Bill on both points, but suggests that the user environment could be modified to help the user realize that the source of problems he is observing is congestion and not something else.
Jeff also suggests that some form of resource reservation or classing of services is needed, but these mechanisms don't need to be overly complex.