Apportioning the Internet

The control of the infrastructure of the Internet is a controversial issue. Recently in the news was the refusal of ICANN, the governing organisation of the domain name system, to approve an application for the creation of a “.xxx” domain — aimed at sites providing pornographic material. The issue is deciding just who actually made the decision. But it’s definitely not the only issue of global fairness in Internet infrastructure.

The problem is that ICANN makes domain name decisions which are then effectively binding on the world (though China may be thinking of breaking ranks). But in general, the rest of the world has to follow suit if the Internet is to continue to work smoothly.

ICANN, in turn, operates under a mandate from the US government — which requires regular renewal. So essentially the US government is in ultimate control of a vital piece of global infrastructure, as a result of the Internet’s history. There has been intensive lobbying at the UN to globalise control more — without progress so far. There is, however, an obvious problem of how such control would be realised — since such a global decision-making authority would need to be to make, well, decisions — and decisions which are technically optimal. But I think the case is pretty clear that more global control is ultimately the “correct” solution, once an effective model can be designed.

The debate has been turned up a notch with the decision of ICANN to refuse the .xxx domain — since it’s pretty clear that it was largely motivated by US government interference. This turns ICANN control from largely a moral question into something that is clearly a practical issue.

There is another issue in global internet “fairness”, however: IP address allocation (the other half of the addressing system). For historic reasons, the US has a hugely disproportionate share of IP address in the current system (IPv4). This is why South Africa, say, has so many more subnets, as IP address are that much scarcer and expensive. Currently growing in implementation, however, is IPv6, which will expand the length of addresses from 32 to 128 bits — or over 10^28 per person. So clearly address shortages won’t be a problem for quite some time, once it gets fully adopted.

But I was very surprised to see earlier today a breakdown of how IP addresses in the new system are currently apportioned. Surely a new system is a great time to make a truly equal distribution of addresses? But no, while the US-EU inequality has been ably corrected, the system is HUGELY skewed in favour of addresses for the developed world — I can’t even see the African portion on the breakdown! Indeed, address allocations are FAR less equal even than GDP differences between countries.

While the size of the address space means this isn’t a practical issue, at least for a long time to come, it’s really disappointing to me that the decision-makers apparently can’t be bothered to even make the Internet look like an equally-shared resource. For shame.

EDIT: I’m now a little confused. Further reading has led me to two conclusions: firstly, I don’t understand IPv6 routing at all; and secondly that the graph MAY in fact be reflecting currently active IPv6 connections — in which case it largely reflects big backbone routers getting ready for the conversion. Or it may indeed reflect actual assignments. Anyone know?