by publius
Over the past few days, I’ve been reading the net neutrality comments filed at the FCC last week (from the major parties, that is). There’s some good stuff in there (Susan Crawford has more). As background, the FCC (perhaps in an effort to avoid dealing with the issue) released a “Notice of Inquiry” this spring, essentially asking for comments on the state of the broadband access market (i.e., whether net neutrality is needed). Google’s comments (pdf) in particular are both interesting and illustrate some of the high-level themes and questions in the current net neutrality debate.
To back up, when you cut through all the rhetoric, the net neutrality debate is really about “access tiering,” which is the practice of charging content or service providers (e.g., Amazon, Google, Vonage) additional fees for a guaranteed level of service (or “quality of service” — “QoS”). People raise the specter that the Verizons of the world will block websites and services, but I’m less worried about that. And frankly, that’s not really what the industry fight is about. It’s about access tiering.
To be grossly simple, here’s how it works. The Googles of the world pay access providers (Verizon, Comcast, etc.) for the bandwidth they use. But they don’t pay additional fees to access you the customer. When you go to the Google website, its servers ultimately send the information through the wires (or “pipes”) running into your house. Because access providers own these “last-mile” pipes, the fear is that they will impose an additional charge for reaching you — i.e., they will put bouncers at the door and charge a fee.
But these companies don’t necessarily want to block sites. What they really want to do is to set up premium “lanes,” with guaranteed QoS. For instance, to ensure that you get Google really fast, Google would have to pay up and sign on for QoS guarantees or get relegated to the slow lane. The fear, among others, is that this system (1) will raise the cost of entry for new providers/services, especially start-ups; and (2) will lead to a “second class” Internet as access providers ignore the non-premium lane.
The access providers’ arguments — which aren’t necessarily absurd — is that they need additional money to improve the Internet and, in particular, to increase bandwidth. The future of the Internet is high-data video and p2p services, so we need more bandwidth they say (and they’re right). Access tiering addresses this need by both subsidizing improvements and allowing for efficient management of congested networks.
Thus, the policy debate turns on whether net neutrality will prevent (or de-incent) network improvements. Access providers spend a long time saying “yes, it will,” while others say “no, it won’t.” Regardless, the focus is often on the effect that net neutrality regulations would have. Net neutrality advocates are often on defense, having to explain why net neutrality won’t harm the future of the Net.
Google puts an interesting twist on this — and goes on the offensive. They argue that a QoS access tiering regime would actually retard network build-out and create fewer incentives to expand bandwidth. The basic idea is that access tiering relies on — and creates incentives to maintain — limited capacity. Here’s what they say:
QoS robs broadband providers of their incentives to build out greater broadband capacity.
QoS originally was conceived as a software-based technical response to limited capacity. Once QoS becomes a profit center for the broadband provider, however, that provider no longer has an incentive to remove the bottlenecks that generate QoS revenues. Instead, the provider has every incentive to maintain capacity constraints in order to justify the QoS fees to customers.If the broadband providers are able to prioritize packets flowing over their network to the
benefit of themselves and their chosen few, they will come to rely on QoS as a revenue-generating crutch that deters them from building bigger, faster broadband pipes that actually serve everyone.
In other words, regulatory policy should encourage a system of abundance rather than freezing a system of scarcity in place. Once this system is in place, Google argues, inertia and reliance will set in, limiting future reforms.
Access tiering isn’t an insane idea, but the broader argument against it is that it would have harmful effects (externalities) on the public at large by locking the broadband market into a “scarcity” model. Net neutrality, by contrast, would force access providers to expand their pipes, or so the argument goes.
In a sense, this seems unfair. We’re essentially asking the access providers to incur private costs for public benefits. Two responses – first, it may not be a public cost if the improved, net-neutrality-inspired pipes sharply increase demand and grow the market. Second, the providers networks’ are themselves the product of earlier public subsidies.
But from a more realistic perspective, it’s going to be hard to get anything passed if the uber-powerful access providers are passionately opposed to it. Thus, if the politics of net neutrality are hopeless mired on this question of fairness, maybe the ultimate answer is some sort of public subsidy for private fiber buildout, not unlike the great infrastructure projects of the past (railroads, canals). Instead of the stick, the government would rely on the carrot. For instance, we might say “if you agree to net neutrality regulation, we will drown you in money for new fancy-pants fiber.” A few months of Iraq spending would go a long way in that respect.
We’re essentially asking the access providers to incur private costs for public benefits.
Except that we used tax breaks to fund the construction of a lot of those datapipes. Net neutrality, so far as I’m concerned, was part of the deal we made with these corporations.
Here’s a novel idea: why not just charge consumers more for higher bandwidth speeds? You, know, like they do now. What’s really going on is that AT&T wants to get a cut of Google’s profits – which is like FedEx or the post office getting a percentage of Amazon’s profit every time they deliver a package ordered online.
If the popularity of overnight delivery services starts putting strains on FedEx and forces them to buy more planes, they boost their fees. AT&T just doesn’t want to extract the money visibly – we might switch to cable.
Google is right – it’s the same reason why we’re stuck with limited refinery capacity for gasoline. The gas companies stand to make far more from a restricted supply than they would be increasing overall volume.
If the intention is to increase bandwidth and ensure bandwidth to those who’ll pay for it, then why not require that all the income from QoS goes to increasing bandwidth? Say, set QoS prices so that 40% of the money will provided a defined level of service for QoS buyers, and then spend the other 60% to increase non-QoS bandwidth.
Or possibly make that 20:80. If you want fast service, then you pay for it and your money goes for faster service for everybody. If you don’t want to pay then you depend on what the paying customers provide for you.
But I don’t like the idea of spammers using up that free bandwidth to send me spam.
“We’re essentially asking the access providers to incur private costs for public benefits.”
I don’t see how we’re doing that. We are asking them not to charge discriminating rates. It looks to me like a classic common carrier question. Furthermore they are already getting paid for access to the their broadband by the end user. A taxi gets to charge the rider, not the rider and the destination.
I wish i could find the URL, but didn’t the feds already give the Baby Bells a huge chunk of change in the late 90s to encourage them to build out fiber? And for our trouble, we get <1 Mbps down and 128K up from AT&T for $10 a month. Our broadband speeds are humiliatingly awful compared to the rest of the world, and that's before you consider that I would rather trade with Al-Qaeda than Comcast - at least they acknowledge they want to make my life awful.
QoS is a very important topic and will be important in the future. Especially as more and more data gets pushed on to the Internet. Most people will want certain sectors of data prioritized. National defense, your VOIP (Voice over IP) phone calls and banking transactions should all have priority over many other types of traffic such as adult entertainment, and many many personal websites.
Laying fiber can be an extremely expensive affair. Use of the Internet is increasing every year and saturation of existing lines in the next many years would not be outlandish.
To me as a networking professional the question is not should there be Internet wide QoS but how should priority be established. The data communication links that criss cross the country have been built with enormous amounts of public money because having those lines is a net public benefit. I will agree that it would then be improper for private cooperations be able to determine who has access to the best speeds on those networks.
This is the tricky part. I favor market forces for the allotment of scare resources but don’t always trust that it will equate to the most public good. At the same time government bureaucracy is not the most efficient at completing tasks such as this. Additionally government control of transmission mediums is an open door to censorship (ask Howard Stern about that).
Honestly the problem with net neutrality is not should QoS systems be widely implemented based on the type of traffic, but who should be able to set priorities and who should pay for the scarce resource.
wiz – i think everyone agrees that latency-sensitive packets for voip, video, etc. should get priority. in fact, as you know, they already do. the question is whether the network operator can strike a deal to make, say, Vonage’s packets prioritized over some other VoIP packets.
that’s the type of individualized QoS that the big boys are contemplating.
so yes, you have identified the tough questions.
Let me repeat my suggestion. If priority is fairly auctioned, and all the proceeds go to expand bandwidth, then isn’t that a start at providing the feedback loop we want?
The more money anybody is willing to pay for bandwidth, the faster bandwidth expands. If there’s anough bandwidth that nobody wants to pay for priority, then there’s enough bandwidth. The more that priority is worth, the faster bandwidth increases.
Don’t give the people who provide bandwidth extra profit for having shortages. Give them the money to expand when there are shortages.
How to optimise that — how to prevent special deals and kickbacks and so on, is left as an exercise for the regulators.
While the telcos are right to whine about the massive costs they bear in providing the “last mile” access, it’s a problem born entirely from regulatory greed. I could probably make a modest living providing internet, cable & phone service to my neighbors here in Manhattan, just on one block. But as long as it costs a fortune in fees to the City, along with stacks of paperwork to put a couple cables in the street, it’s only going to be the big telcos in this game.
I question the wisdom of allowing private companies who control what has become a necessary resource to arbitrarily restrict access to that resource. Any sort of tiered access is going to have to be closely monitored to prevent abuse. It seems far more efficient to impose net neutrality ‘regulation’ that prevents such things entirely than to deal with a web of actual regulation in the wake of the first big scandal that results from QoS.
And forgive me if I don’t have a lot of sympathy for access providers because they have to invest a lot of money in infrastructure to stay competitive so they can continue to make money. They’re sitting on a cash cow, and they’re pissed that they have to buy the feed. “Oh, those poor, good-hearted dears! Won’t somebody think of the giant corporations?”
I would be much more inclined to consider the genuine technical issues that tiered access might address if I had the slightest confidence that the media corporations wouldn’t immediately politicize the whole thing. But just as we’ve seen repeatedly with (for instance) net-nanny type services censoring criticism of their sponsers, I see lots of reason to expect that tiered access would be used against corporate and political enemies, and no reason at all to trust in the emergence of a system in which only technical concerns dominate.
At a very minimum, I’d want the tiering to be completely transparent, subject to review, and publicly documented, and I can’t see that happening.
Publius: “People raise the specter that the Verizons of the world will block websites and services, but I’m less worried about that. And frankly, that’s not really what the industry fight is about.”
Why are you so quick to dismiss that concern?
Verizon may not have any interest in blocking websites based on content, but they have very good reason to discriminate against services that directly compete with one of their other businesses, i.e. Internet telephony. They wouldn’t have to block them, they could just not allow them to work very well. This has been mentioned frequently as a major factor in the telcos’ anti-neutrality efforts; it seems odd for you not to mention it here.
“National defense, your VOIP (Voice over IP) phone calls and banking transactions should all have priority over many other types of traffic such as adult entertainment, and many many personal websites.”
Possibly, but I suspect in practice the adult entertainment people would pay for priority, and national defense wouldn’t.
As far as I can tell, QoS measures are mostly being suggested by telcos who want to stop competing voip services from working, or other schemes for hosing big websites which make a lot of money.
Hob – it’s not b/c i think telcos are such great people. It’s b/c I think they fear the outcry — one of hte benefits of the rise of the NN debate is that everyone is on the lookout for things like this. Blocking VoIP could easily justify a new regulatory regime.
there are, however, other ways to skin a cat. Verizon is attempting to bankrupt Vonage through patent litigation. And they could easily charge VoIP a hefty fee for access to the fast lane, which would be less obvious discrimination.
Jonas,
Can you please provide a cite for your claim that the major cost to providing data service in Manhattan is regulatory compliance and paperwork?
I really don’t believe that is true. In any event, I believe that Manhattan is irrelevant when discussing last-mile pricing issues: the population density is sufficiently high that anyone can make money. The real cost is wiring millions of miles of cable to people in rural, suburban and exurban communities: the truth is, the cost of providing fiber to everyone in Wyoming is probably significantly greater than the cost of providing fiber to everyone in Manhattan.
National defense, your VOIP (Voice over IP) phone calls and banking transactions should all have priority over many other types of traffic such as adult entertainment,
Speak for yourself, buddy.
I’m being a little facetious, but seriously, I don’t want a corporation deciding what’s more important to me as a consumer. I want to be able to access what I want when I want it, and I dare say that boobs are as important to a number of people as finding out if their rent check has cleared yet.
Publius: It’s hard to argue these things without making it into a really technical discussion, but maybe you can just trust me on this one: ISPs already do charge Skype, Vonage, etc. for faster access. That is, the provider who is connecting Vonage to the Internet backbone will charge them for a lot of bandwidth – just as if a business has a call center with 3000 phones, they’ll pay their local phone company to make that work. This is a red herring that the telcos are constantly throwing in because most people just don’t know how the Internet works.
The other red herring is “we need special lanes for high-volume stuff.” Current NN rules don’t prevent that; it’s mostly a technological problem. Eventually the backbone will be higher-capacity and will be tiered to some degree, and Vonage will pay more to get onto the fast lane, and I’ll pay more if I choose to get the best possible connection from my house to the fast lane. However, Verizon doesn’t own the backbone, just the local connection, so that’s not as interesting to them.
What they’re after, the only reason they would need the law changed, is to be able to set up their network so that Verizon customers can get a great connection to Verizon’s own systems, a good connection to whichever third parties will pay extra to Verizon – even though Verizon is not the ISP on their end, i.e. it has nothing to do with “access to the fast lane” since they’re already on it – and a crappier connection to everything else. They wouldn’t need to “block” third-party VoIP to make it unattractive.
I have more to say, because I do know a little bit about this, but are we really suffering from a shortage of fiber cable?? I was under the impression that we still have a surplus from the dotcom boom and it’s just being underutilized. Perhaps someone can correct me here.
hob – aren’t you conflating backbone issues with last-mile issues though? i agree that the backbone is less of a concern (though is there really such a thing as a “fast lane”, or only “fastly marked packets”). though even the backbone is consolidating, and is increasingly held by retail ISPs like AT&T. in other words, fast backbone access is a necessary but not sufficient condition for vonage (b/c they can always get screwed at the last mile).
more generally though (and this is a point i’ll blog about one day), there’s a separate question of whether QoS is even possible. For instance, even if Vonage paid, how could Vz guarantee prioritized service (i.e., an individualized priority beyond best efforts) across the Net. It doesn’t own big chunks of the backbone, etc.,a nd would presumably have to enter separate negotiations with every link in the chain.
That’s why I wonder if the telcos and cable even really know what they want — they maybe are just fighting to maintain flexibility
adam – think of interstates and local “exit roads.” you need both to get to the gas station.
We have lots of fiber for the backbone (the “interstates” of the Net), but not enough for the last mile connections to your house (the “exit roads”). presumably, if everyone had a new fiber line running into their house, there would be no more bandwidth problems and the NN problem would basically go away
Yet somehow in the US we let the cell phone companies charge both the caller and the callee — something that doesn’t happen in other countries, as I understand it. That makes me less than hopeful that the telecom companies are going to lose this one.
I just got a notice in the mail from Verizon that they’ll be installing the required infrastructure for FIOS in my neighborhood. This apparently before NN is settled. I assume someone (not me) could comment on this as it relates to this post.
The question comes down to basically can the phone companies be trusted?
No.
They did the same sort of thing with DSL competition. Legally, they were supposed to offer the same service over their lines to everyone, even other companies offering competing DSL.
Did it happen? No. Funny how that worked out that the phone company’s DSL was always better quality and less random problems than competitors.
What would keep them from doing the same thing? Nothing.
“A taxi gets to charge the rider, not the rider and the destination.”
I know I’m quoting myself, but now that I think about it, I have a vague recollection that it might actually be illegal for the taxi to charge the destination (or that it used to be a common method of extortion). Does anyone remember that?
Arthur_Hendy
Baron Wigram Dave Bergamini Northop Hall John T. Flynn Swedish general election, 1998 Yitzchok Dovid Groner Jyarimon Cave of the oilbirds Oscar Rennebohm Marty Markowitz