Responding to Felten (& Co.), squared

My response to Ed Felten’s policy recommendations (Felten suggests we wait and see before imposing network neutrality mandates), has garnered its own response. Tim Lee of TechLiberation has posted two responses (first) (second), the first of which becomes part of Felten’s sensible follow-up, and the second of which references this thoughtful post by EFF Chairman Brad Templeton.

The very short summary of the dispute goes like this: We network neutrality supporters are so scared of the perils of broadband discrimination that we are willing to accept the perils of imperfect regulation. Opponents are so scared of the perils of regulation that they are willing to accept the perils of discrimination, at least for now.

The other three guys are very intelligent people, one of whom has been dragged into this (reluctantly, I’m sure), and all of whom have thought long and hard about this. As in any decent debate, both sides have some merit.

First, consider the large degree to which we agree: End-to-end architecture is generally good. In a perfect world, broadband markets would be filled with competitors. They would generally build neutral networks, but experimentation with non-neutral networks would be rewarded or punished in the competitive market. Most of the US in the actual world is characterized by a broadband duopoly (at best). A viable and competitive third or fourth broadband service provider (BSP) might be headed your way, maybe providing service via wireless transmission, but in the vast majority of the country, that day is years down the road.

We all agree that discrimination by broadband providers, which would erode the end-to-end architecture that has made the internet such a wonderful thing for years, is generally a bad thing. Further, we all agree that the muck of Congressional and regulatory politics, which has made the US Federal Government such a mixed bag of good-and-bad for many more years, is also bad.

The point of disagreement is on the relative likelihood and probable impact of each bad thing. I have condensed what has been said elsewhere into a few key points, and I will respond to each in turn.

Opponents’ Summary Point 1: Don’t mandate a specific architecture.

Response: Who’s advocating that?

For instance, Templeton argues:

I think we make an error by deciding this is about mandating an architecture for routing internet traffic. This is really a question of monopoly power. We don’t want the duopoly of telcos and cable companies using their franchised position of strength to defeat other players. We don’t want phone companies blocking Vonage, or cable companies blocking or slowing internet video services.

He contends that, if we must pass a law, we should use this law “to clarify just what the monopolies shouldn’t be doing.” In other words, laws should focus on business practices rather than internet architecture.

I agree with his recommendation to prohibit discriminatory business practices rather than mandate specific internet architecture decisions. What I disagree with, however, is his claim that all draft proposals thus far “have ranged from being quite constraining on the net to being trivial to bypass for a determined telco or cable company.” Obviously, Rep. Joe Barton and Sen. Ted Stevens have the “trivial” end of the spectrum covered with HR 5252 and S. 2686 (pdf; the latest link I could find is this mid-June draft).

I cover this problem below. But I don’t see tech mandates in the draft proposals I’ve seen.

For instance, what in the Snowe-Dorgan proposal, S 2917, mandates a specific internet architecture? The text is remarkably free of techno jargon. It forbids the blocking or degrading of legal net traffic, but it specifically authorizes companies to prioritize packets. If VoIP and streaming video need a smarter network, companies can build that smarter network. They just cannot charge extra for delivery of those specific services. They can even charge more for faster pipes, so they still have the economic incentive to build them out.

(In fact, they will build them out under either scenario, at least in wealthy neighborhoods. The working classes won’t pay enough in fees, and few push media companies are going to pay to have their content delivered to the working classes. But that is a whole different topic.)

Opponents’ Summary Point 2: Broadband companies will do everything in their power to a) skip out of neutrality mandates, and b) impose them in perverse ways on newcomers.

Response: The fight is worthwhile and imminently winnable.

Sure, AT&T will fight to twist the law. And the mafia tries to evade law enforcement and to turn “their” cops on their enemies. Is the answer to wait and see how bad the mafia problem gets? Do we fret that the cure may be worse than the disease?

Lee in particular shows surprising willingness to give in to an industry that he describes as deliberately unregulable. To whit:

What happens once network neutrality is the law of the land, Public Knowledge has moved onto its next legislative issue, and the only guys in the room at FCC hearings on network neutrality implementation are telco lawyers and lobbyists? The FCC will interpret the statute in a way that’s friendly to the telecom industry, for precisely the reasons Herman identifies.

If (a) telcos and cable cos are impossible to regulate, or (b) the FCC is fundamentally corrupt and/or incompetent, then we have bigger problems on our hands the net neutrality, and we certainly should not passively accept this state of affairs. But if this is the case, then there’s nothing we can do, but there’s nothing we can do to make it worse.

Thankfully, neither of those categorical, Manichean statements is true. Broadband providers are difficult, but not impossible to regulate. The FCC is often on the wrong side of things, but not always, and as the Prometheus decision illustrates, the FCC can also be checked when they stray too far a field from the legislative text. Prometheus is also a great example of a tiny, under-funded group taking on major companies and a powerful Federal agency and scoring a sizable victory. Regulatory and judicial victories are orders of magnitude more feasible for small groups and single citizens.

Not only can network neutrality regulation work, it already has. Consider why network neutrality is the norm now: common carrier regulatory legacy from the dial-up era. In 1996, when the telephone lines were being congested by dial-up ISPs, a totally unregulable telco industry would have found a way to start dropping calls to competing ISPs. Not consistently, of course, but let’s say a random 20%. That behavior might be hard to prove, and it might have been in the telcos’ best interests (indirectly, through lowered system usage and somewhat greater long distance business), but it just didn’t happen. Why?

Dropping calls on purpose was and is explicitly illegal, because landline telephone companies are regulated as common carriers. Bells knew that even a cozy FCC could not ignore explicitly illegal behavior. Further, the inability to charge extra for dial-up calls greatly reduced the incentive to drop those calls. They would have angered more customers than it would have been worth, considering the threat of regulatory action.

In the network neutrality discussion, we have the same companies, regulated by the same agency, forced to comply on essentially the same issue. It worked before. Even if it works much less effectively this time, some degree of credible threat of enforcement (by Federal court if necessary) will at least prevent the explicit pay-to-play business model. Then broadband companies will cool their discriminatory heels, because if they can’t directly cash in on discrimination, the indirect benefit (e.g., VoIP customers giving them a little more long distance business) won’t be worth the infrastructure investment.

Regulating the right way will be difficult, but possible. Wrestling broadband companies off their brand-new discriminatory hubs and the huge revenue streams they will have built in the next 5 or 10 years? When the public has moved on, SaveTheInternet.com is parked, and PublicKnowledge.org loads at 1/10th the speed of DontRegulate.org? Now that’s impossible.

Opponents’ Summary Point 3: If Congress passes toothless regulations, this will make it even harder to fix the problem later after discrimination becomes a problem.

Response: If we pass telecom reform this year, that’s your other option.

Felten, for instance, says, “I worry that Congress will enact toothless rules or vague statements of principle, and then declare that the issue has been taken care of. That’s not what I’m advocating; but I’m afraid it’s what we’ll get if [we] insist that Congress pass a net neutrality bill this year.”

I also fear this scenario, but note that it will almost certainly happen if either telecom bill passes. (The audio recording of Stevens’ famous speech even begins with his reference to the “Internet Consumers’ Bill of Rights.”) I urge you to read HR 5252, which the House has already passed, and the relevant section of the Stevens bill, S. 2686.

Both contain net neutrality rules that probably wouldn’t do much and certainly wouldn’t prohibit the Ed Whitacre business model, aka extortion.

Congress is, of course, very good at passing toothless laws to regulate communications industries that profit handily from government-granted monopoly licenses. It takes some degree of specific public pressure to get anything better. And, as I mentioned in my first response to Prof. Felten, that window will pass quickly.

Opponents’ Summary Point 4: We might pass something that’s even worse than Whitacre’s extortion, or what we pass might become worse through regulatory perversion, and then it will be just as hard to undo it later.

Response: Regulation makes good scenarios more likely, and bad scenarios less likely.

Felten is worried “we’ll adopt well-intentioned neutrality regulations that we’ll regret later — and if the issue is frozen later it will be even harder to undo our mistakes.” Lee is worried that, “if the telcos figure out how to use the rules to their advantage, they’ll lobby just as hard against repealing them. (just look at the legal fight to liberalize cable franchises).”

These are reasonable fears, so it is important to get it right the first time. But consider the relative likelihood of the following six scenarios, which brings us back to where we started disagreeing:

1a. We do nothing (or implement Stevens’ or Barton’s “solution”), and BSPs generally continue to deliver a nondiscriminatory broadband experience. Over 95% of end users’ demands are met to the best abilities of the network. Wireless providers enter and become viable. A few telcos roll out fiber to the richest neighborhoods.

1b. We do nothing (or implement Stevens’ or Barton’s “solution”), and BSPs begin discriminating to some degree, but most of them don’t go crazy with it. Cable companies make sure YouTube doesn’t stream very well, and they charge customers extra for their own streaming video service. BSPs demand fees for QoS guarantees but a small number of companies go along with it. BSPs further degrade service to begin driving new QoS customers, but reach an equilibrium at which point QoS fees are still a little more than they’ve lost in network investments and lost customers. Incumbent BSPs try but generally fail to use current telecom law to slow down new market entrants. A few telcos roll out fiber to the richest neighborhoods.

1c. We do nothing (or implement Stevens’ or Barton’s “solution”), and BSPs begin discriminating as a means of self-enrichment. They open a separate business office to take extortion money from sites that want to get through at full speed. If you the end user and/or Vonage want your VoIP calls not to be throttled, if you want your search engine to load quickly, pay up. Companies occasionally try to launch an antitrust case, but antitrust is an ineffective solution (after all, it worked so well for Netscape, who actually won in court), and a tiered internet takes hold. New broadband companies try to serve those who thirst for an open internet, but incumbent BSPs laughingly throw the entire Telecom Act at them, bankrupting each with nefarious FCC investigations. Regional duopolies reign throughout the US. A few telcos roll out fiber to the richest neighborhoods.

2a. We implement something like Snowe/Dorgan. It works to deter broadband discrimination well enough that over 95% of end users’ demands are met to the best abilities of the network. Wireless providers enter and become viable. A few telcos roll out fiber to the richest neighborhoods.

2b. We implement something like Snowe/Dorgan. BSPs wriggle around under the thumb of regulation, and they get away with some discrimination (a cable company degrades YouTube until they get called on it, a bell throttles VoIP calls from Skype but not Vonage and manages to escape punishment). BSPs try but generally fail to use the Telecom Act, including the neutrality rules, against newcomers. A few telcos roll out fiber to the richest neighborhoods.

2c. We implement something like Snowe/Dorgan. BSPs openly defy the law and open a separate business office to take extortion money from sites that want to get through at full speed. The FCC looks the other way. Those who are affected go to court, but the courts turn them away. Justice Scalia, in a famous opinion, writes, “Go away, or I shall taunt you a second time!” When newcomers enter the market, they comply with every letter of the law, but incumbent BSPs laughingly throw the entire Telecom Act (now including some unenforced blip relating to net neutrality) at them, bankrupting each with nefarious FCC investigations. Regional duopolies reign throughout the US. A few telcos roll out fiber to the richest neighborhoods.

The alert reader (all 5 of you who’ve made it this far) has surely noticed that, except for the policy solutions, 1a ~= 2a, and 1c ~= 2c. Each best-case scenario is roughly the same; we get a neutral internet and something closer to decent competition. Each worst-case scenario is also the same; we get severe discrimination, and incumbent BSPs successfully deploy regulatory strategy against viable competition.

Now, ask yourself which is more likely depending on which policy option we pick. If we do nothing, the first half of 1c is more than some crazy blogger’s bad dream. It’s Verizon’s business plan. It’s AT&T’s new revenue model. It’s Cisco’s fondest wish. (It’s also the FBI’s playground; more on this below.)

If we do nothing, 1b is the best we can hope for. Expect generally degraded service on some or all kinds of content, except the best-financed corporate stuff. Picture millions who would rather go back to dial-up. Expect dozens of countries to sprint past us in the broadband stakes.

If we do nothing, 1a is virtually inconceivable. Nobody spends $1 million/week opposing legislation and wins, only to behave as if it’s passed anyway.

If we adopt strong neutrality legislation, 2a is not exactly going to fall in our laps, but it’s reasonable. I’d say we have something like 1a/2a now, and even Felten acknowledges that the mere threat of regulation has been plenty to prevent nefarious behavior. This strongly suggests that strong laws could do a comparable job.

If we adopt strong neutrality legislation, 2b is also reasonable, but not as likely as Lee and others suggest. I think 2a is actually more likely, at least for the vast majority of types of traffic, but I will concede some likelihood for 2b. VoIP faces real dangers (which Felten describes), as do other time-sensitive services that compete directly with other revenue streams that BSPs now enjoy.

Yet these dangers will generally be limited to just such services. Unless Verizon starts selling books, Amazon is okay, because my ability to reach Amazon.com increases what I’ll pay for DSL. Ditto music, pizza, personal networking sites, etc. Further, even vulnerable applications won’t be very vulnerable. If Comcast can’t cash in directly, it will hardly be worth networking investments plus the threat of a big regulatory fight. (See above; dial-up calls weren’t randomly dropped so that you’d still have to call your brother in Topeka.)

If we adopt strong neutrality legislation, 2c is extremely unlikely, but it’s not too much of a straw person to suggest that Lee (at least) seems to take it far too seriously. For the discrimination to be as bad as in 1c, the BSPs in 2c would have to explicitly and deliberately flout the express letter of the law, charge money for not doing something (throttling specific traffic) that they’re forbidden from doing, and get away with it. If it were going to happen so blatantly, it would have already happened in the dial-up era.

The “regulatory ammo” half of scenarios 1c and 2c is, thankfully, unlikely in both cases. Further, telecom regulation is already insanely complex, so the scenario that net neutrality regs will uniquely lead to anticompetitive rules gaming is so unlikely as to be not worth fearing. Verizon already knows the rules better than anybody (except maybe AT&T), and “you’re not being neutral” is certainly not going to be their best shot at newcomers.

Really, Lee is arguing that the first half of c will be just as likely under 1 or 2, and we should act to avoid the second half of 2c. Considering that it’s very, very unlikely, it’s a risk I’m willing to take in order to avoid the much more likely 1c. After all, several BSP executives have explicitly stated that 1c is their plan. If they think 2c is easy to get to, why are they spending $1 million/week to stop strong neutrality legislation?

The best- and worst-case scenarios are essentially the same. The middling cases favor regulation; discrimination will be less and less perverse under Snowe/Dorgan or similar legislation. The odds of a good or at least acceptable outcome are much more likely if we regulate, and the odds of a very bad outcome are much more likely if we do nothing.

If you’re still reading, I have two additional thoughts.

1. Don’t underestimate the need for a larger vehicle on which to carry a law mandating neutrality, and if it’s not video franchising, I don’t know what it is. (USF reform? I somehow doubt it.) Once telcos have won the state-by-state franchising game (and they’ve already won a few), they’ll blow off the need for a federal bill, and then we’re fighting in 50 different trench wars to restore the end-to-end internet.

2. Law enforcement writ large must be salivating at the idea of a non-neutral internet design. (Paul Kouroupas has witnessed the drooling first hand.) The privacy concerns alone suggest a neutrality mandate. Even if we do nothing and 1b happens, we the citizens still lose. Your BSP has built it, and law enforcement will come. (And they’re not played by Kevin Costner.)

The norm of relatively anonymous browsing? Bye-bye. Grey market traffic in, e.g., the Grey Album (not to mention Daily Show clips)? Bye-bye. Once BSPs roll out Cisco’s shiny new discrimina-tech, the law enforcement community will have the means to know what you do online. But we can trust the feds not to spy on us without a warrant, right? Granted, this might also happen under Snowe/Dorgan, but without Scenario 1’s direct incentives, they’ll probably keep building bigger dumb pipes.

Felten, Lee, and Templeton are all very smart guys who have obviously put a lot of thought into this issue. In my estimation, however, they are wrong about the characteristics and relative likelihoods of the policy scenarios above. Regulation is never the best-case scenario, but on this issue, it is the most likely means to get us the closest to what we want.

In short, if you agree that a neutral internet is very good and that discrimination is very bad, the inoculation is almost certainly better than the disease.

(Thanks to Tim Schneider for his valuable input on this book—um article—I mean, uh, blog post. This is also cross-posted on Shouting Loudly.)