From: azeem@dial.pipex.com (Azeem Azhar) Absolutely correct: can I direct you to two of our articles on the subject (and a useful graphic): http://www.economist.com/issue/19-10-96/ld4401.html http://www.economist.com/issue/19-10-96/sf0774.html http://www.economist.com/issue/19-10-96/sf1.gif

I took a quick look at the articles, and see that you have confused rather a lot of technical details. For example, "intranets" are not "new" ways to bypass the Internet, and there is no such thing as a "private Internet". Intranets are just marketing-speak for the _existing_ local networks that the Internet joined together. Most local networks are already "private"; private networks have been with us for decades, and the "big I" Internet connects them to make them accessible outside their locality. Although you have identified some of the problems with currently priced phone systems, you were incorrect to state that the largest cost to the PSTN is the switches and wires. It is well documented here in the States that that the largest expense of any PSTN is the cost of administration, billing, and call management. It is clear that the primary edge enjoyed by the current Internet pricing models is the reduction in administrative overhead. It is inaccurate to say that "[Internet] users pay only for their cars and its fuel". The cost of the "street" is fixed, and well known to the providers, and is already included in the monthly cost of the "fuel". Similar to how we already fund roads here in the States (fuel tax)! And I wouldn't place too much credibility on the Varian/Mackie-Mason papers. As I have been watching them for some years, they published the same papers with minor modifications in multiple publications (a general no-no in the academic community for padding their Curriculum Vitae), and have had in the past models that bore little resemblance to Internet reality. As Mackie-Mason once admitted to me, (paraphrased off the top of my head) these concepts have been over-simplified for the consumption of economists.

That's true: in a conversation with Phil Lawlor (CEO of AGIS), he told me that one of his customers ISPs, who was also peering with AGIS at the Mae-E, indulged in "bandwidth stealing" by receiving their Usenet news feed across Mae-E rather than as an AGIS customer across the few T-1s they had to AGIS.

It seems you do not understand the concept of "peering", and little understand routing mechanisms. If the ISP is peering with AGIS at MAE-East, then AGIS has agreed to accept traffic _sent_ from the ISP at MAE-East for the rest of its network, at those locations included in AGIS BGP advertisements. Conversely, unless the AGIS routing sends to that ISP at MAE-East, then it would be impossible for the ISP to _receive_ their Usenet feed from AGIS. Therefore, AGIS itself _must_ have directed the flow at the ISP via MAE-East. The Internet routing is based on best-effort forwarding, not receiving. WSimpson@UMich.edu Key fingerprint = 17 40 5E 67 15 6F 31 26 DD 0D B9 9B 6A 15 2C 32 BSimpson@MorningStar.com Key fingerprint = 2E 07 23 03 C5 62 70 D3 59 B1 4F 5E 1D C2 C1 A2