I sent this to my US Senator.
Dear Senator Boxer,
I’m a constituent and a network engineer. I invented key parts of the communications systems used by your personal computers (Ethernet and WiFi,) and I’m against net neutrality regulations.
Net neutrality sounds like it’s as wonderful as motherhood and organically-grown, fair-trade apple pie. We all want our Internet to be free, fair, and open, and nobody wants any heinous tollbooths in the Info Super Highway, gatekeepers in Cyberspace, or Big Brothers deciding what blogs we can read. If the issue were that simple there wouldn’t be a debate, net neutrality laws would pass both houses by acclamation and be signed by our President with flourish and fanfare and peace would rule the planet, etc.
But public policy is not so simple, especially in a case that involves the regulation of the single most complicated machine ever built, the Internet.
The Snowe-Dorgan and Markey Amendments contain a poison pill that will stifle the evolution of the Internet, in the form of a prohibition against a Quality of Service surcharge:
If a broadband network provider prioritizes or offers enhanced quality of service to data of a particular type, it must prioritize or offer enhanced quality of service to all data of that type (regardless of the origin or ownership of such data) without imposing a surcharge or other consideration for such prioritization or enhanced quality of service.
The argument in favor of this provision says that it’s needed in order to prevent the formation of a two-tier Internet, where one tier has Quality of Service and the other doesn’t, and this is somehow bad for Daily Kos and Google.
This is a false claim, because the engineering math behind Quality of Service says it can’t be applied to every stream from every user. In Lake Woebegon all the children can be above average, but on the Internet all packets can’t.
We have a two-tier Internet today where commercial customers have a full range of service plans available to them, but consumers have a very limited menu. The provision would guarantee that the consumer menu will always be severely limited. We don’t need a single- or dual-tier Internet but a multi-tiered one where every new application can get the services it needs from the network and nobody has to pay for services they don’t want. As Tim Berners-Lee, inventor of the World Wide Web said:
We pay for connection to the Net as though it were a cloud which magically delivers our packets. We may pay for a higher or a lower quality of service. We may pay for a service which has the characteristics of being good for video, or quality audio.
That view cannot be reconciled with the provision cited.
Perhaps ironically, the old monopolists – AT&T, Verizon, Bell South – advocate expansion of consumer choice while the new monopolists – Google, et. al. – advocate for restriction of choice.
Consumers benefit from the choice that the Stevens Bill offers them, and I hope you will reverse your public position and reject the special-interest-benefiting amendments offered by the agents of the New Monopolists.
The Internet has never been regulated in the manner of the cited provision, and anyone who tells you otherwise is trying to bamboozle you. Don’t be misled.
In 2002, Internet framer Vint Cerf said: …we must dedicate ourselves to keeping the network unrestricted, unfettered and unregulated. That’s as true today as it was then.
Respectfully,
Richard Bennett
Richard, there is one issue your letter to Senator Boxer does not address. Why should ISPs be the ones who prioritize traffic? Shouldn’t the model be that content providers and content consumers get to set QoS standards between each other without interference from the ISP?
As long as their accounts are in good standing, sure.
A realistic scenario could play out this way: I pay Verizon $39.99 a month for 20mbps access to the public internet and through a deal with Microsoft Verizon gives FiOS subscribers an additional 5mbps VPN over their fiber for MSN services. I could try to have a QoS agreement with Yahoo using the public portion of Verizon’s fiber but Yahoo wouldn’t be able to guarantee the same level of service that Microsoft could offer on the VPN.
How is that good for competition? Even if Yahoo offers a better service than MSN, the segregated bandwidth available to MSN will make it more attractive to consumers. This implicitly reduces competition and reduces innovation.
Richard, continue your fight against legislation, continue advocating a better architecture but I think you should consider that ISP’s partitioning off chunks of bandwidth is bad for competition and bad for innovation.
I think the scenario plays out like this: You can pay extra for QoS or not.
if you don’t pay for it and MSN does, you can use QoS with MSN but not with anybody who doesn’t pay for it. (Why you would want to is a different question, of course.)
If you pay for QoS you can use it with anybody, anywhere, any time. And the same goes for the services.
Is that evil?
Given that QoS is extra work for the ISP, it seems like a reasonable model to me.
You are making the assumption that Verizon would sell you the extra bandwidth to establish the QoS with Yahoo; I don’t think they would. Their business model is to freeze out competitors, just take a look at what they’ve done with VCAST on the cellular side.
Is it evil? That’s not the right question. The right question is: Is it anti-competition?
I assume they want QoS revenues from anybody they can charge, you assume they don’t.
There doesn’t seem to be any way to resolve the differences in our assumptions, given that the feature in question is not currently available. Therefore, I propose we watch what they do and prepare to regulate if they do as you suggest.
I want to avoid pre-emptive regulation because it’s hard to narrowly tailor it.
Richard thanks for your replies. Regulation does not seem like the way to go, nor does legislating from the bench; which is what would happen if we sit back and wait for court cases, I am looking for other ways to convince the telcos to stay away from Access Tiering.
I’ll continue to be a reader; I find your posts to be insightful and entertaining, but I’ve been posting here too often and it’s time to let others into the conversation. Please feel free to stop by my blog any time.
– Tom
Tom @ redbanktv dot org
Several ISPs have announced that they intend to charge QoS fees to content providers. The fear, however, is whether they will offer QoS to their competitors in video and voice services (or other services that the ISPs enter), and even if they do on what terms. It seems likely that ISPs would not offer QoS to their competitors or offer it only for very high fees. Charging those high fees to competitors could sqeeze them out of the market, thus reducing competition. Over time, even the very large competitors could be squeezed out (because they cannot compete with the ISPs who do not have to pay the QoS fees). An extreme but plausible result is that we have only the ISPs providing certain services (video and voice for starters).
The price squeeze problem has a long history in telecommunications, and various methods have been used to combat it, with varying success. The most recent method is NN legislation. As far as I can tell, most of the NN legislation is designed to permit QoS tiering without permitting QoS fees.
Richard, you have argued here (and elsewhere) that this system would require that the ISPs give QoS to anyone who wanted it–which is everyone–so that the ISPs could not feasibly provide QoS to anyone. I am not sure I understand entirely.
Let me explain. Why couldn’t an ISP decide to provide QoS only to all video service providers? My understanding is that each video service provider could “paint” the IP header of its data packets to indicate the appropriate Class of Service, and that voice data packets would then be sent to the appropriate queue and prioritized according to the ISP’s algorythm.
It seems to me that there are too few video service providers to flood the queues with video data packets such that the ISPs could not effectively provide QoS to all video service providers. And even as the number of video service providers grow, the amount of video services purchased, and therefore running through the queues, will increase but not by that much. Each consumer can watch only so much video and only so much video at a single time.
And maybe the ISP could also expand its QoS for voice services, too, the same way. That wouldn’t mean that all content providers would get QoS because their data packets would be painted as a different Class of Service.
And it seems likely that the ISPs would be willing to provide QoS to all video and voice competitors given the ISPs’ interest in providing video and voice with enhanced QoS.
And I think NN legislation could be written, if it has not been already, to allow the ISPs to offer QoS to individual consumers who wanted to pay for it.
What am I missing?
Do you have any support for this:
The solution to QoS overload is to build faster pipes. Fine, now who pays for them and why?
Except in rare circumstances, I can download video now without enhanced QoS or at least without ISPs imposing QoS fees. For now, the pipes seem fast enough to handle video, which only becomes truer as fiber gets deployed. (Note: According to the telecoms, deployment of fiber is more dependent on cable franchising reform than net neutrality). In the future, when the pipes might become inadequate for the services demanded, more Internet platforms will be commercially viable. Then, NN proponents argue, competition (rather than regulation) can prevent the price squeeze.
Is there evidence that the current pipes are not fast enough for now, especially with the deployment of fiber?
QoS doesn’t have anything to do with web sites or downloading files.
I should have been more careful with my choice of words. I can get a live video stream during teleconference calls in a non-VPN environment.
Yes, you can, at a bandwidth and jitter level that precludes any sense of “presence”.
I worked on the videoconferencing-over-the-Internet issue ten years ago on a DARPA contract, and that work convinced me that the Internet needs QoS-on-demand to fully support this application.
What we want is the ability to multicast HDTV-quality audio and video without going bankrupt, where deep buffering isn’t feasible.
There are only two ways to do this: 1) you can over-provision all the links like crazy; or 2) you can have QoS.
Given that the cost of over-provisioning is four times greater than that of QoS. it’s your basic no-brainer.
Let’s see if I understand now. Let’s assume that Congress passes NN legislation that permits enhanced QoS but not QoS fees. If the ISPs implement QoS, we get higher quality video conferences and other real-time video and voice services, which means more people will use them. The result is more demand. If the telecoms and cable companies expect demand to exceed pipe capacity, they might not implement enhanced QoS to begin with because there would be no point. An overcrowded QoS queue defeats the purpose of the QoS queue.
The key question then is, Will demand for real-time video and voice services increase (because of enhanced QoS or otherwise) so much that the pipes will not be big enough to handle them? I was speculating “no” in the short-term because of anecdotal evidence and the current fiber deployment, but you might know better. What is your basis for thinking we will run out of pipe in the short-run–despite the current fiber deploment–so that enahnced QoS won’t be possible?
QoS has value, and if you give things away that have value people take them. Post a note on Craig’s List for a free sofa and somebody will come get it even if it’s all ratty and falling apart. So the demand is there, and the supply is limited. As people become more accustomed to VoIP, IPTV, and IP Videoconferencing the demand will grow. Porn alone has to potential to push up the demand for IP Videoconferencing astronomically, so to speak.
And how do we balance supply and demand?
Yes, I recognize that demand will increase as quality increases. But supplying VoIP, IPTV and IP Videoconferencing will not be free. Thos providers have to pay access and bandwidth charges. Widespread use of these services will still cost a substantial amount. With NN legislation, enhanced QoS simply would not cost an additional fee. What I would like to know is whether you have any evidence that demand will exceed supply, especially as fiber is deployed, such that enhanced QoS will not be possible?
ChadB,
NN requires that providers not *charge* an additional fee… It doesn’t mean that the cost of providing QoS magically disappears.
Providing Enhanced QoS capability to people costs more than not providing said service… What you are suggesting would simply force ISPs and NSPs to shift their QoS costs on everyone, *even if they don’t use it or need it* Is it fair that users who don’t need QoS be forced to subsidize those few users (google Video for instance) that requires enhanced QoS?
Doesn’t sound fair to me….
Max, what is “fair” is a tough question.
You have suggested that QoS operating (and continued R&D) costs would be covered by bandwidth usage fees, which all users pay regardless of whether they provide or use services that need enhanced QoS. So, from one perspective, maybe website owners and users like my grandparents (who will never use anything but email) would be subsidizing Google Video a little bit if NN legislation is passed. But before you can condemn that as unfair, you have to consider the alternatives. Without NN legislation, small innovators needing that “subsidy” might not be able to enter the video market and compete with Google Video. Comcast might be able to squeeze those small innovators out with QoS fees and, indeed, eventually Google Video and everyone else, too. If that happened, we all would have to pay Comcast monopoly rates to get video service at all. And we, in theory, could be subject to Comcast controlling our video content. Would it be fair if we all have to pay much higher rates, lose innovation from start-ups for IPTV, and have access only to video that Comcast approves?
If its proponents are right, NN legislation would create a public commons of sort. It would be akin to the idea that I pay local property taxes, which go to supporting public schools, even though I do not have any kids in those schools. But smarter kids will benefit me indirectly, as will increased innovation on the Internet for my grandparents, so it is not necessarily unfair.
If we assume that NN legislation would work as its proponents predict (an assumption I do not know to be correct and one that Richard thinks is wrong), the choice seems to be (1) a loss in social welfare where almost everyone is made worse off in varying degrees (no NN legislation) or (2) maintenance of overall social welfare where some benefit more than others and in some cases a few people are made slightly worse off (w/ NN legislation).
With these kinds of trade-offs, what is fair depends on the theory of justice you are using. How do you want to evaluate this? Economic efficiency (utilitarianism)? Rawl’s theory of justice? Kantianism? Libertarianism?
Without QoS, large entrenched interests like Google and Yahoo have a huge advantage over mom-and-pop services. If the ISP can’t manage traffic, then whoever generates the largest amount of traffic controls the flow through the ISP. See the article on Google’s moonbase in Oregon and tell me how a couple of guys in a garage can compete with it. Hint: buy 900,000 computer and multi-gigabits of bandwidth for starters; get a better algorithm if you have money left over.
If Google has a video file download service and a voice service, they can choose to interleave voice and data packets in such a way that their customers effectively get QoS, and NN laws don’t prohibit that. So we get a scenario where the Telco gets to pay for the last mile, but Google gets to manage it, and that’s not fair under any interpretation of personal liberty.
Richard has pointed out in numerous posts that the barrier for entry for the “Little guy in his garage who wants to compete w/ google” is the fact that google has massive server farms located at numerous POPs that are close to the users.
The costs and prices assocaited with QoS tiering are peanuts compared with the capex necessary to replicate massive server farms, something the little guy will be forced to do without the ability to buy Enhanced QoS services. In this scenario, tiered pricing benefits everyone but the entrenched “content” providers.
If, according to your theory, Comcast does something dumb, I’ve still got options:
Satellite
ADSL2+ Video Delivery
FIOS
etc.
Hardly a monopoly scenario.
And if they all the last mile providers decided to do the same thing, you’ve got a whole bunch of current laws to hold the network operators accountable, from FCC regulations to antitrust laws.
All we are saying is that perhaps we should see if these theoretical abuses actually occur. If the current mechanisms don’t work to resolve it, then we can address it with legislation.