California Dreaming…about the Internet

All state legislatures are pretty nutty, but I doubt any is as peculiar as California’s. While it was once a model of the modern, professional, policy making body, a series of unfortunate historical events – mainly term limits, massively powerful public employee unions, and a runaway initiative process – have reduced it to a shadow.

I used to spend a lot of time at the capital in Sacramento in the mid-’90s, just as term limits was kicking in. I got to see the changing of the guard from well-informed, long-term legislators to newbies who knew they weren’t going to be around for long, and it wasn’t pretty.

Perhaps the highlight was a reaction by the then Speaker of the Assembly (the lower house) to a suggestion by the Senate’s pro tem to contact the governor before finalizing a bill. The speaker asked: “How do I do that, just call him on the phone?” Long-term speakers like Willie Brown never asked that question in public. This former speaker is now running for governor.

Even Supporters Aren’t Impressed

Nobody’s impressed by California’s attempt to legislate net neutrality; even fervent supporters of the Wheeler FCC’s Title II order, such as EFF, are critical:

States are constrained, however, because federal policy can override, or “pre-empt,” state regulation in many circumstances. State law that doesn’t take this into account can be invalidated by the federal law. It’s a waste to pass a bill that is vulnerable to legal challenge by ISPs when strong alternatives are available.

EFF’s alternatives aren’t very compelling – a series of actions that would ossify today’s technology – but if you can’t win EFF over to your side, your net neutrality bill isn’t very strong. Few have actually read the bill, so let’s do that just for fun.

Reading the Bill

The California legislature puts all its business online (thanks, Jim Warren!) so it’s easy to see what’s afoot. We can look at bill as it moves through the process, gets amended, and is analyzed by committee consultants: go right here.

The operative parts of the bill start with section 1, where the legislature explains what it’s trying to do:

(a) The Legislature finds and declares that the Federal Communications Commission has repealed net neutrality rules intended to protect consumers and to ensure fair and reasonable access to the Internet.

(b) It is the intent of this act to ensure that corporations do not impede competition or engage in deceptive consumer practices, and that they offer service to residential broadband Internet customers on a nondiscriminatory basis.

Of course, paragraph (b) collides with federal consumer protection law enforced by the FTC. So California’s beef isn’t with the FCC (as paragraph (a) suggests) as much as with the FTC. That’s pretty weird. And the bill does nothing to ensure broadband is actually offered on a nondiscriminatory basis around the state.

This raises a question: if California believes federal law is deficient with respect to protecting consumers from ISPs that “impede competition or engage in deceptive consumer practices”, does it also believe that the feds are on top of these practices when other firms – such as, I dunno, Internet ad networks, social networks, retailers, or video streaming services (just for example, doncha know) – are doing them?

This is a plausible interpretation of California’s beef. Some advocates of state-by-state regulation of ISPs have argued that the FTC lacks the awesome technical expertise that led the FCC to opine that DNS is a routing function. But the focus on regulating one part of the Internet to ensure that it won’t do things it’s never really done while ignoring another part of the Internet that actually does those things (Google, Amazon, and Facebook discriminate like mad) is pretty much what this beef is about.

But I digress. The key point is that California says it wants to play the part of the FTC, which can only mean it believes it’s better at it than the real FTC. OK, let’s see how it intends to do this job.

What’s Wrong

The bill makes two errors in its definitions. First, it defines “Broadband Internet access service” as a “mass-market retail service by wire or radio in California that provides the capability to transmit data to and receive data from all or substantially all Internet endpoints” as well as all the features “incidental to and enabl[ing] the operation of the communications service…” This is meant to distinguish the out-of-state ISPs from the in-state monopolists, but it fails.

Gmail, Twitter, and Facebook enable communication among substantially all Internet endpoints. The companies that provide these services are the largest users of Internet transit services and such infrastructure as undersea cable. Their whole business depends on communication networks that they themselves own and operate.

The bill also tries to give itself a hedge against this overlap between ISPs and “edge services” by declaring ISP information processing services that support the communication function as non-separable from communications. So this is the old communications vs. content dichotomy from the telecom law of the ’60s. But the trouble with reducing services like DNS that are provided by both ISPs and “edge services” to nothing more than communications is that it works wherever you find DNS. If Comcast employs DNS to make its network more efficient, then Google must be employing it make the Google network more efficient as well. And what does that characterization do to the edge monopolies?

What’s the Product?

It implies that Google’s offering, whatever it is, has a big component of communication. We might argue that the Google product is chiefly the processing of certain kinds of information – the location and relevance of web pages, the location and interest of consumers – to which communication is ancillary.

That argument is a nice way to distinguish the roles firms play in the Internet. ISPs supply simple communication, with information processing as an ancillary feature. Edge companies chiefly provide information processing, with communication as an ancillary. All we have to do to properly classify Internet businesses is to find their chief activities and take it from there. The fact that all Internet businesses are, by their nature, blends of communication and processing needn’t trouble us.

But maybe it should. Is there really a rational basis for regulating Firm A under one law because its business is 51% communication and regulating Firm B under a different law because the communication portion of its service is only 49%? This is the absurd heritage of the telecom law. At some point, the activities of global enterprises may consist of a larger portion of communication and those of national ISPs may consist of more content and information processing. And there’s no real difference between processing information and transporting it.

You Can’t Always Tell Them Apart

The fundamental similarity of the firms regulated under net neutrality regulations and the big Internet players that are not becomes especially problematic around net neutrality’s key proposition, the idea that “paid prioritization” must be banned because it’s an anti-competitive anathema. California defines the bogeyman simply and broadly: “the management of an Internet service provider’s network to directly or indirectly favor some traffic over other traffic, including through the use of techniques such as traffic shaping, prioritization, resource reservation, or other forms of preferential traffic management, either (1) in exchange for consideration, monetary or otherwise, from a third party, or (2) to benefit an affiliated entity.” [emphasis added]

It’s necessary to limit the ban to ISP networks because processing information in such a way as to speed some things up and slow other things down is a universal practice in the information technology industry. When I write a computer to make something happen quickly, I divide the task into subtasks and start the most time-consuming subtask first; or at least as close to first as possible, given the constraints of the the subtask’s need for data that must be produced by other subtasks.

The execution time of a program depends on the data it has to process in a given iteration; it take less time to compute the square roots of small numbers than big ones. So I’ll also optimize my program for the most typical data it will encounter in the real world. That sucks for the atypical data, but it keeps costs low and makes most people happy most of the time.

And I will also separate the task I’m most interested in from other tasks that are additional parts of the package of functions the marketing department wants me to build. Firefox, for example, processes web page content at higher priority than web page ads. It does this because people actually want to see the content. Is anyone complaining about the Firefox fast lane?

The Internet has Many Fast Lanes

California is bright enough to realize it can’t ban all forms of acceleration, but just barely. Its laundry list is comprehensive: “traffic shaping, prioritization, resource reservation, or other forms of preferential traffic management” get the ax. Were it not for the focus on ISPs, content delivery networks would have to go because they too are in the acceleration game.

Applying net neutrality rules to public CDNs would be especially troublesome because it would exaggerate the imbalance between the performance of big Internet properties with private CDNs (each of the five largest US companies has its own CDN) and ordinary-sized companies that depend on Akamai. So why ban acceleration by ISPs when acceleration by non-ISPs is hunky dory?

This question is key to the net neutrality debate because generic antitrust law would never do it. All it takes to make the economists who’ve developed our antitrust law happy (in this context) is equal access to acceleration. It’s not any harder to write a law that says any party offering acceleration services to Internet users must do so on a commercially reasonable basis. If they do this, the whole net neutrality debate goes away.

So why doesn’t California do that instead of screwing around with this crazy attempt to become FTC Junior?

What’s Missing

It appears that California’s Senate pro tem Kevin de León approached this bill in a political manner. He decided he had to have a net neutrality bill and simply asked his staff to find an expert to draft one.

So the question of why he should do something like this never came up; nobody asked: “what harms are we trying to avoid and how can we avoid them with a minimum of damage to innovation?” It often appears that the goal of net neutralists is to inflict maximum harm on ISPs, just to settle old scores. And some in DC policy circles have scores to settle going all the way back to the ‘6os.

Consumers were easy to get wound up about their ISPs ten years ago, when the Internet was new to them and their access to it was gated by a high-priced broadband plan. But I’m not so sure consumer rage is so easily channeled today. As Babette Boliek has pointed out, there’s plenty of evidence that consumers are actually more wound-up by violations of privacy and trust on the part of California’s “edge service” monopolies.

So Senator de León may just be fighting yesterday’s war. We’ll see.


It’s notable that this bill explicitly excludes the telephone network from its scope:

(a) “Broadband Internet access service” means a mass-market retail service by wire or radio in California that provides the capability to transmit data to and receive data from all or substantially all Internet endpoints, including any capabilities that are incidental to and enable the operation of the communications service, but excluding dial-up Internet access service. [emphasis added]

This is important because the ban on paid prioritization would make the telephone network unlawful. This is because it bans resource reservation for a fee, the very heart of telephone service. I leave it to the reader to think about why a measure that bans common telephone network practice – indeed, the fundamental design of the telephone network – is good for the Internet.