I am providing here a summary of the comments received. Thank you for the feedback, which has been of a very high standard and very helpful. In this summary I have edited comments and in some cases combined them with others and précised. I take full responsibility for any misrepresented or ignored and mangled points that result and further comments and corrections are welcome. I have been asked for more information, especially about the background to the Australian position. My apologies for not giving this before. The ICAIS Task Force web site is at http://www.apii.or.kr/telwg/ICAIS/ICAIS-frame.html Included on the site is a paper outlining Australian objectives for the project. Unfortunately there is little 'hard data' on the site other then the reports of the study group itself. There is a paper on Internet traffic measurement by Nevil Brownlee. My summary of feedback follows. The draft comment was circulated to the New Zealand Network Operators emailing list (nznog(a)list.waikato.ac.nz) and to the members of the Internet Society of New Zealand (ISOCNZ, isocnz-members(a)isocnz.org.nz). These two groups encompass most, if not all, of the technically knowledgeable people in New Zealand dealing with TCP/IP issues and many others who are concerned with policy development in the Internet area. The draft was also passed by subscribers to one or other list to Australian commentators. A total of 11 people have returned comments so far, some of whom are experts in TCP/IP matters and measurement of Internet traffic flows. Some comments were privately to me, others were broadcast to the NZNOG list and elicited further online comment and discussion. Overall there has been support for the draft, with some specific criticism of para 11 which needs redrafting and clarification. In general the New Zealand position statement seems to have the support of at least the technical community. Points made supportive of the draft include: (i) International traffic is carried over symmetric pipes but traffic flows are usually in fact highly asymmetric. Pipes are usually sized to accommodate flows in one direction (in New Zealand's case, inwards) so that the much smaller back-flows are often essentially free to the circuit provider. (ii) Much traffic between non-US countries crosses the US networks and the cost is negotiated as part of inter-provider agreements. We are smaller than the US providers and need them more than they need us. There's nothing in principle stopping an NZ ISP negotiating with a US ISP an arrangement where the US ISP paid for US traffic crossing the international link. (iii) We do not want governments imposing a regime which would force ISPs to do PTT-style charging for voice calls over the Internet - that would severely limit the development of voice, video, multimedia, etc. technologies over the Internet. (iv) The number of services running across the PSTN is limited so it is relatively straightforward to make an agreement on interconnect. By comparison, on the Internet subscribers of the network are free to invent their own services without the interconnecting networks knowing anything about them. Hence any simplistic (and non-zero-settlement) charging regime imposed by network operators and based on aggregate traffic measurement is doomed to failure. For example, a charging scheme that attributed value to volume of traffic would place all the value in the content consumer - content supplier relationship with the content supplier (since there would be a net traffic volume imbalance in the direction supplier -> consumer). (v) The current Southern Cross project is an example of how various telecommunication companies will invest, regardless of (potentially) unfavourable interconnection arrangements. Over five years, Telecom's international Internet traffic carriage has fallen from 98% to 22% highlighting that significant other Service Providers have chosen to implement large bandwidth provisions regardless of the supposed "unfair" interconnection agreements. NZ's position should be to actively argue that the current arrangements should not be changed rather than passively opt for the status quo. (vi) Unless interconnect agreements mandate a flat 50% share of the costs, the asymmetry of the traffic means that non-US connectors would still pay the lions share. (vii) Since 90% of the end user cost is attributed to local access cost issues can be addressed locally without recourse to expensive international quango's. There were no comments strongly opposed to the draft but there were some cautionary comments and a number of questions. My interpolated comments are in parentheses. (viii) There was some surprise expressed at the high proportion of the cost to the end user being attributed to the local loop quoted in my initial commentary. (It should be noted that this is over the APEC area and in NZ's case is likely to be lower but the actual figure is not known for NZ). It was also pointed out that the proportion of the international traffic cost to the ISP well be much higher than for an end user since they are paying a smaller part of the cost of the local loop. Very late in the preparation of this commentary, Neil James provided the following figures for Otago University: LAN (local loop) 57%; National 4%; International 39%. (These figures are probably a reasonable attribution of the costs faced by a NZ-based ISP rather than an end user, if so, costs to end users in NZ for international connectivity are substantially higher than the 3% measured for the APEC area as a whole. This reflects the relatively lower overall local loop cost in NZ than is the case for the APEC area as a whole.) (ix) Whether the current situation has limited growth or not is surely only part of the equation. The other part of the equation must be 'are we paying a fair proportion of the costs?'. (x) How much hosting business is lost to overseas hosting companies because of the international traffic costs? Some government departments even host sites overseas. And a similar point made by another commentator: NZ ISPs are effectively subsidising anyone outside NZ who views a website hosted by them. There is anecdotal evidence that NZ ISPs are reluctant to host popular websites for this reason. Ihug, for example, finds it cheaper to host their clients' websites on a server in the US. (xi) From a national perspective, we are subsidising consumers in another nation. It would be interesting to compare the costs of hosting an identical website in NZ and the US and see whether the traffic charges are significant in the NZ case. (xii) There were a number of comments to the effect that measurement of traffic flows is both possible and useful, noting that para 11 of the draft is misleading on this point. (Clearly this paragraph needs redrafting to ensure accuracy and clarity.) (xiii) The statement in the draft that value of Internet use is not measurable just in terms of traffic flows is a subjective, value-laden statement, which makes it harder to defend. It should be amended to the effect that the Internet allows much more varied styles of remote interaction than voice telephony, which greatly enhances the ability for groups of people to work together despite being located in different economies. (xiv) VOIP certainly raises new issues. If NZ typically provides the gateway into the an IP pipe used for VoIP from US to NZ for US callers clearly we would want the US to pay a portion. Despite the problem of measurement one has now to wonder just how we can agree on appropriate sharing of costs. Frank March Specialist Advisor, IT Policy Group Ministry of Economic Development, PO Box 1473, Wellington, NZ Ph: (+64 4) 474 2908; Fax: (+64 4) 471 2658 --------- To unsubscribe from nznog, send email to majordomo(a)list.waikato.ac.nz where the body of your message reads: unsubscribe nznog