cyberspacemapThere are worries that Internet Service Providers (ISPs) may inject intelligence into their networks to try and unfairly differentiate their services from competitors’. Time Warner’s recently reformed End User Licensing Agreement (EULA) may be the most recent demonstration of this kind of differentiation. The EULA recognizes a difference between third-party video streaming, and streaming content from Time Warner’s own network spaces, and authorizes Time Warner to:

…monitor my bandwidth usage patterns to facilitate the provision of the HSD Service and to ensure my compliance with the Terms of Use and to efficiently manage their networks and their provision of services. TWC or ISP may take such steps as each may determine appropriate in the event my usage of the HSD Service does not comply with the Terms of Use. I acknowledge that HSD Service does not include other services managed by TWC and delivered over TWC’s shared infrastructure, including Video Service and Digital Phone Service. (Source)

This is arguably a ‘bad’ integration of smart networks and ISP pipes because the intelligence being located in the network discriminates on the basis of data sources and destinations, as opposed to discriminating on the basis of application-types. Time Warner’s EULA facilitates discriminated billing processes by establishing a competitive advantage for the ISP’s own services at the expense of other video and VOIP providers. Given that Time Warner does have a cap-and-bill based billing system, their discriminatory bandwidth monitoring practices should lead us to question who, exactly, should be responsible for monitoring consumers’ actual bandwidth usage; whereas gas pumps are certified as being accurate, who should certify ISPs’ own monitoring systems? How exactly can consumers be alerted that there is, indeed, valid third-party certification processes that accurately monitor bandwidth usage? Should other bandwidth uses (e.g. operating system and application updates) similarly be exempt from using up a consumer’s bandwidth quota?

If we assume that ‘bad’ competitive practices are abandoned, then in a truly byte-based system we could ask why ISPs should be allowed to categorize what are permissible/impermissible appliances to attach to the network. In a utility-based situation, I don’t need permission from the utility provider to hook up  a new phone; why should I be prevented from setting up a file-server? DSLreports nicely correlates byte-based monitoring against what occurs in the the electricity market:

What devices and appliances you use is out of the of control of the utility provider. . . In a post metered bandwidth billing world, the ISPs will no longer have any right to dictate what physical devices are making use of the data after passing through the meter, as described in the FCC’s long delayed yet to be implemented “Plug and Play” law. (Source)

While we might question the need to use byte-based bandwidth (Anderson has a damning piece that demonstrates that American ISPs’ revenues are up, whereas capital costs are way down), we can bet that after introducing this kind of metering consumer expect that their ISP won’t meddle with the packets they generate or receive (though we may get into discussions of authenticating packet streams and guaranteeing their integrity to avoid incorrect billing practices). Just as my electricity company doesn’t cut me off when I use a lot of power, there should be no reason (or, at best, only a few) for an ISP to cut individuals off from the ‘net should they be using excessive amounts of bandwidth. Now, the issues that arise most quickly are:

  • Who determines what is ‘excessive’?
  • What is done to notify consumers of their ‘excessive’ use?
  • How are throttling practices employed to limit a subset of consumers from upsetting a service’s integrity?
  • Does byte-based billing threaten to undermine emerging services, should costs for byte-billing be particularly high?

Each question is important and deserves careful consideration – it is particularly important to actually, empirically, prove that this kind of a billing structure is necessary.

At the same time that there are discussions of (re)introducing byte-based systems, network neutrality discussions are getting back underway in the US and Canada. Senator Ron Wyden recently stated that “[c]ontinued growth of the ‘Net right now is being hampered by the lack of clear enforceable standards on net neutrality. I don’t think the country can afford that” (Source) and Charlie Angus (Digital Affairs Critic for the NDP in Canada) has tabled Bill C-398, “which will ensure the future development of the internet is not impeded by unfair throttling or interference by telecom giants” (Source). What would it mean to (sensibly) institute network neutrality practices in a byte-billing digital environment?

I would, first and foremost, suggest that packet prioritization systems would almost necessarily remain. I haven’t found a reasonable position that justifies driving all packet prioritization schemes from ISPs’ networks, but instead a large number of positions meant to limit ISPs from unfairly applying their schemes in ways that work against consumer interests. In Canada, the Canadian Association of Internet Providers (CAIP) is asserting that the CRTC’s willingness to let Bell Canada apply DPI-enabled throttling against CAIP’s consumers was unfair because the decision was rife with factual and legal misunderstandings; as a result, Bell’s current throttling practices can be read as detrimentally affecting CAIP’s customers (note: I’ll be pouring through CAIP application, as well as their 2008 proposal, in the fairly near future with a post or two on it to follow). Rather than deploying DPI ‘solutions’ unilaterally and simultaneously targeting consumers and wholesalers, from a network neutrality perspective DPI could be deployed to limit viral outbreaks, manage traffic under particular (preferably publicly stated and revealed) conditions, and so forth. Intelligence can lie in the network so long as it is applied in accordance with neutrality principles and, in the US, CALEA effectively demands that (at least residual) intelligence be located at certain parts of ISPs’ pipes.

Should we arrive at byte-based billing, perhaps customers could customize what kinds of throttling are applied to their traffic; maybe parents want to throttle peer-to-peer traffic so that their kids don’t consume excessive amounts of bandwidth, or instead set this kind of traffic to the lowest priority level possible. Perhaps consumers actively want (for some reason) to limit an ISP from locally caching their data. In essence, intelligent networks aren’t necessarily bad (and in fact are very helpful in balancing loads/limiting genuinely harmful dissemination of some data) and could be used to enhance consumer choice without leading to a contestation with network neutrality principles. Consumers themselves might insist that heightened (though regulated) modes of intelligence should reside in networks to better control their byte-based utility bills. Technologies such as deep packet inspection could very well thrive in such an economic system; ‘dumb billing’ does not necessarily mean that ISPs will be motivated to return to ‘dumb pipes.’