Traffic flowing nicely
From the OECD, a useful paper about interconnection online. It turns out (surprise!) that inter-networking is working fine without intervention. There are zillion networks out there, and as long as the local telecommunications environment is sufficiently open (all the way to opening up incumbent facilities to competitors), these networks are finding ways to connect on their own:
The greatest cost barriers to any country connecting to global networks are not traffic exchange relationships, in competitive environments, but monopolists charging high prices in the absence of such competition.
Also — where there's facilities-based competition, broadband prices can plummet and services to rural areas can be profitable. Global Broadband Battles makes the same point: Reform to telecommunications regulation (opening up facilities to competitors) is the key to stimulating growth in access.
Thanks to Milton Mueller for the pointer to the OECD paper.
Visibility
Tom Evslin recently spoke about the need for volunteer monitoring of broadband networks. There's no question but that making network operation visible would help — we'd be able to tell whether non-minimal discrimination targeting particular applications was going on.
But it's a tough logistical problem, requiring adept coordination of decentralized data and people:
Each volunteer would download software that triggers their computer to send out test packets called pings to various websites. Because pings automatically trigger a return packet, they can be used to measure the speed of a connection between two computers. Each probe PC reports its results to a central server that can then work out from all the ping times whether packets from certain websites are being deprioritised, and if so by which broadband providers, says Evslin.
The providers will have all kinds of sensible reasons why they're managing their networks in particular ways, and plenty of room for arguments about why any data gathered is inherently inaccurate. But I like the SETI approach, and Tom's idealism is inspiring.
It would also be good to involve actual network researchers to do controlled experiments with the cooperation of the broadband providers. That's the kind of thing that CAIDA does well. But it needs funding and support from the right players. It's a big step to require “private” network operators to support research about the operational nuances of their networks.
We make progress when we can see what we're doing, but on this particular front we'll probably be operating in the dark for a long time.