¹Bohn, Braun and Wolff (1994) also address these issues and suggest a voluntary priority scheme.

²Some other derivations of congestion tolls are provided by Mendelson and Whang (1990), MacKie-Mason and Varian (1994), and Shenker (1995).

³delay within the backbone can be easily modeled as in Gupta, Stahl and Whinston (1995a).

⁴From the users' perspective, in reality, the bottleneck is either the server's pipeline or the slowest data communication link in their path to the server.

⁵Note that some users might decide not to get the service because of excessive delays, however, users with negligible delay costs will try to obtain the service regardless of the delays. Thus, with no pricing mechanism the services can potentially be accessed by only the users who value it the least.

⁶Very little is known about the performance and/or design of feedback mechanisms to control the processes in real-time. To our knowledge, our effort is the first systematic, theoretically supported, approach to such real-time problem solving. We have gained valuable insights and have developed several mechanisms for consistent predictions in real-time. The details of these approaches are beyond the scope of this paper and will be presented elsewhere.

⁷Realistically, this work would be done by a smart agent executing on the users machine.

⁸Note that the perfect waiting time information scenario case is the "best-case" scenario for our implementation of the free access policy because users first check where can they get the fastest service and the information they get is exact.

⁹Note that this figure for delay cost is probably too low by an order of magnitude. However, recalibrating delay cost to 10% of the mean value of a job (or $0.277 per hour) leaves the net benefits and reported in Table 1 essentially unchanged. On the other hand, a uniform rescaling of the value of a job and the delay cost would, of course, simply rescale all the numbers in Table 1 proportionally. Thus, Table 1 is conservative.

¹⁰Conservatively, as of October 1994, there were 2.5 million servers on the Internet (source: MATRIX News); it is safe to assume that collectively 10% of those servers have a capacity equivalent to our server capacity.

¹¹Witness the phone company wars.

¹²Friedman, 1971; Fudenberg and Maskin, 1986; Abreu, 1988; and Dutta, 1991.

¹³For example, see van Damme, 1991; Mailath, 1992; and Stahl, 1993.

¹⁴For example, see Holland, 1976; Arthur, 1990; and Stahl, 1995