A British perspective on a very American process

As a new member of the the “Tech Elders”, I was invited to join yesterday’s hearing [PDF] in Washington, DC on the reclassification of broadband Internet access services. The US Federal Communications Commission (FCC) has decreed that Internet access should switch from being lightly regulated as an ‘information service’ (Title I) to a more heavily regulated as ‘telecommunications service’ (Title II).

I’d first like to say that the process and content was a credit to the rule of law in the United States. The efficiency with which the event was conducted was exemplary, the quality of the arguments was high, and the substance was, well, substantial. My respect for American democracy was given a big boost.

The hearing was packed out. I got to the courthouse around 7am and stood in line for two hours, which meant I got one of the last few seats in the main chamber. It was quite an atmosphere! Here are a highlights from the copious notes that I took.

* * *

Issue: The hybridisation of ‘information’ and ‘telecoms’ services

There was extensive debate about whether activities like DNS and caching mean ISP services cannot be ‘telecommunications’, since they use computing functions that non-ISPs also perform as stand-alone unregulated services.

My take? On this, the FCC had the stronger case. These functions were effectively ancilliary and supporting. There is a considerable history of common carriage having computational elements. However, the idea of common carriage is poorly suited to distributed computing, which is what ISPs now enable.

Issue: Defining what it means to ‘offer’ an ISP service (per Brand X)

There is a legal ambiguity around whether a telecoms service is defined by the components by which it is enabled, or its overall function. Is “access to the Web” a telecoms service (as its mechanism is to “take you to the website”), or is it an information service (as its outcome is to offer “Webby info goodness”)?

The arguments centred on the idea of information copying at a distance being like “efficient movement” of physical goods, with “edge providers” being “shippers”. There is a boundary between the “access to the information service” and “the information service itself” whose position is unclear (if it can even be defined).

My take? There is a basic problem in regulating broadband ISP services, which is defining what the service actually is. What is the “promise”, and how can you know if it was delivered? Where does it take you, and with what kind of experience?

The FCC was arguing that a distributed computing service is a duck because it is aquatic, even if it also lays eggs like a snake and swims like a platypus. Their position was unconvincing. Reasoning by metaphor between physical and virtual systems is unhelpful.

Issue: Are ‘specialised services’ harmful?

The existence of ‘fast lanes’ must mean everything else becomes a ‘slow(er) lane’. Is this a good or bad thing?

A judge pointed out that the FCC was making a bogeyman out of paid priority (which is currently rare), whilst we already have ubiquitous and uncontroversial paid peering. The railroads were at liberty to charge for refrigerated containers for goods that needed special handling, so it seems “utterly reasonable” that ISP should be able to do the same. Indeed, “users who create a cost should bear that cost”.

The FCC’s representative waffled around the “virtuous circle” (which is deeply flawed), and had no good response when challenged that competition law was perfectly capable of taking care of any issues.

Furthermore, former senior FCC economists had decried the ban on ‘paid prioritisation’, saying there were simpler and less damaging ways of achieving the same protections. The FCC had offered no response whatsoever to this criticism.

There was a golden moment when the FCC was proposing the way to help ISPs grow, is by constraining their ability to monetise their services so that the “edge might blossom”. It became necessary to destroy the industry to save it…

My take? All other transport businesses have tiered services that align price and cost to timeliness of delivery. The FCC is pushing a hypothentical “dread” which has absolutely no factual substance behind it, and has lost tremendous credibility as a regulator as a result.

The FCC’s position is that it’s OK to push the problem from the wholesale interconnect side onto the retail users. Yet the idea that end retail users are going to pick up the mantle of disputes over interconnection pricing is ridiculous.

Banning a market for quality is an anti-innovation policy. It creates a distortion by preventing rational resource pricing through market mechanisms. A simple general rule on equal access to paid priority is plenty enough.

Issue: Does the FCC have the authority to act?

If “proof by repeated assertion” was a valid form of logic, then the FCC clearly has authority to act! Over and over, we were reassured that reclassification was fully within their discretion and aligned with statute.

The reality is that the FCC is working to regulate the wholesale side (interconnection) by proxy through control over the retail service. The separability of the “interconnection market” from the underlying service cannot be sustained.

The central issue is this interconnection, as the statute refers to THE Public Switched Network (PSN), i.e. a singular thing. The law also says that to be a telecommunications service, you have to be able to reach substantively ALL end points of that PSN. National policy is clear that the Internet is by default unregulated.

The FCC is claiming that there is still only one PSN, with the Internet being a subnetwork comparable to fixed, mobile or pager networks. The judges were having none of this, since there is a clear distinction between phone numbers and IP addresses, which cannot “reach” each other.

The FCC’s retort was that the ubiquity of the Internet’s ability for “people to reach each other” made it functionally equivalent to being part of the PSN, and hence this equivalence enables you to pass the test in law to be able to regulate ISPs as a telecoms service.

There has been a long-standing exception for mobile broadband, which is being sustained here. As a result (as one judge noted) you might get different regulatory outcomes depending where in your house you stand and which network (fixed or mobile) the device attaches to.

It was noted that the FCC has created a set of “vague standards” over ISP conduct which create great uncertainty. Every ISP has to guess whether any change in its technical or commercial model may be retroactively declared unlawful.

My take? Heavens, what a mess! How on earth will the FCC regulate upcoming hybrid access services that dynamically blend fixed and mobile? It’s clear that the law was intended to make only the traditional circuit business a Title II service.

No matter how much you think that margarine is buttery, you can’t unilaterally reinterpret the butter law to cover all greasy spreads. Suddenly even cream cheese becomes “butter-like substitute” due to its fat content and consistency.

The FCC is turning a law about the Public Switched Network into one about the Packet Switched Network. No level of sophistry can transmute multiple “connected” services into one “interconnected” network. The gap between the FCC’s actions and the statute cannot be bridged by inventing a new “ubiquity” test out of thin air.

Issue: Did the FCC follow due process in proposing new rules?

Let’s keep this one short. No, they manifestly didn’t. However, the resulting harm is open to debate.

Issue: Does Title II infringe upon first amendment speech rights?

The argument proposed was that ISPs are like cable services, who exercise editorial discretion over what to aggregate. The counter-argument was that if they do so, how can they claim immunity for responsibility for the content as mere conduits?

The government appears to have taken little consideration over the impact of this reclassification on speech issues.

My take? That ISPs services have hitherto offered connection to “the whole Internet” is not a reason to perpetually enforce this emergent phenomenon. In a future Internet of Things world, reaching only portions of the end points is likely to be normal. I don’t need to be able to connect to every electricity meter in China, as doing so has a real cost.

I felt this part was not well argued, since the “editorial discretion” over the allocation of performance was ignored in favour of total focus over blocking. There’s a good case to be made that an ISP should be able to declare itself “best for Skype!” should it wish to do so.

The interacting and competing free speech rights of users, ISPs and edge providers remains murky. Good luck to the judges in unpicking it.

The bottom line: Title II is a bad answer to a good question

How should we regulate abuse of power by infrastructure owners in the information age? The Title II reclassification is an attempt to constrain ISP power, but is a politically, technically and economically inept one.

My opinion is that the FCC has trashed due process, and chosen to largely ignore the law, too. That said, the reason it has done so is understandable. The law itself is no longer fit for purpose, storing up structural stresses in the regulatory and political system.

The conceptual model we are offered by the 1996 regulation is one of classic telecommunications vs “the rest”. This no longer matches the immutable external technical reality. Traditional circuits are corner case of a large possibility space of distributed computing services, offering contention-free transmission of ephemeral data.

We now have a system of virtualised compute, transmission and storage, all under software control. Both AT&T and Netflix are distributed computing platforms, just of different kinds. The distinction of “delivery” from “edge provider” fails to capture the complexity now involved.

In a sense, it doesn’t matter who wins this case, either at appeal or in any (seemingly inevitable) Supreme Court case. The law is not aligned to the virtualised computing world, and that world isn’t going to change to accommodate obsolete law. The only possible long-term resolution is for Congress to rewrite the law to be fit for the digital age, grounded in robust science.

My view is that neither side’s position seems tenable. These distributed computing systems are becoming critical infrastructure. There are serious bottlenecks to competition that create untrammeled power, and systemic issues and risks that the public interest requires addressing.

The idea there should be no constraint of arbitrary and capricious ISP discretion over resource allocation is silly. Conversely, the net neutrality fundamentalist attempts to have government define the possible services is outright stupid and irresponsible.

I believe that in the interim, the least harmful approach is de minimis regulation of a fast-evolving technology space. I therefore concur with Petitioner that the FCC has greatly exceeded its authority without due process, and that its Title II reclassification should be disallowed.