Saturday, 21 February 2015

Network Neutrality. Conceptually Wrong. Time For it to Die.

"Network neutrality is the idea that all web content, regardless of its form or who provides it, should be treated equally by internet providers."

Network Neutrality, like so many things, is a high-minded concept that aspires to a wonderful ideal, but an impractical reality.

But it's biggest challenge is that its advocates often make incoherent contradictory arguments like: "If they're saying I can watch all the TV I want through their internet network, but I can't go browsing Facebook on this, as much as I want, is that legal through net neutrality?"

If you are an advocate for an open, transparent and accessible Internet, Facebook is the last platform you use as an example. It is entirely conceivable that Mark Zuckerberg goes to sleep each night masturbating to his ideal version of the Internet where every single piece of content is provided through a (or address. Facebook's business model is to drive as much Internet traffic as possible through it's closed ecosystem. And this is just one example of how consumers choosing to lock-in with commercial organizations online flies in the face of the stated goals of an open, neutral and equal Internet.

It is also striking that most of the outrage (covered by the press I will grant) over the so-called violations of Network Neutrality start from a cost or billing perspective. Take this new Telus charge for exceeding arbitrary bandwidth caps. Nothing about Network Neutrality states that Internet Service Providers cannot charge for the service they provide. And quite frankly, in a capitalist economy, what the market will bear applies as much to Internet access as it does to luxury watches or $900 smart phones. "But the Telus Optik TV works through the same fibre optic network as the internet!" Oy vey. If Telus wants to provide Telus content over Telus owned infrastructure to Telus customers at a price of its own choosing, it has that right. Network Neutrality is a concept, not a law. By the logic presented, if Telus provided Optik TV over a private IP network, or through traditional cable TV infrastructure, then "Network Neutrality" wouldn't apply because it isn't on "The Internet." Which is bullshit logic. A company should not have to build parallel infrastructure to provide an IP network based service because it violates strongly held ideals of misguided Internet denizens. If Telus owns IP infrastructure, it has a right to sell service over it how it sees fit, regardless if that infrastructure is connected to the Internet or not. Telus' customers do not own their Internet circuits. Telus owns them. (I can already hear the But! But! screams, and I know what you're going to say, I'll get to that in a bit.)

And quite frankly, the "internet traffic should be treated equally" concept is a technical none-starter. As the Internet has evolved, the services provided over it have challenged the concept of "equal." The problem with any network is that it is a complete technical impossibility to provide all of the bandwidth needed to everyone all of the time. Every Internet circuit is over-subscribed.

Consider the following simple three node network:

In this example, N1 is a server that provides content, and N2 and N3 are clients that interact with that server. Let's assume that N1, N2 and N3 each have a network link of 1 Mbps. That means that if N3 is using all of it's 1 Mbps connection interacting with N1, then N1 has no available bandwidth to provide to N2. N2 will get the equivalent of "Page not Found error." This is an equal network design, and it relies on a principle of FIFO (First In, First Out. "Or he who asks first, gets answered first.") On this network, N1 is "over-subscribed." It is hoping that neither N2 or N3 will take full advantage of their network speed that often, when they do, the connection to N1 is congested.

The simple solution is to give N1 a 2 Mbps circuit so that it can equally provide content to N2 and N3. Of course, if N4 is added to the network, N1 will be oversubscribed again. But at least the calculation here is simple. The network bandwidth N1 needs is expressed in the formula N1=N2+N3+....N99. There is only one node that needs to have enough capacity to interact with every other node.

But what if everyone is a server? Like participants on a BitTorrent network, sharing data with each other.

Here we have a problem. Because any one node can be providing data to any other node, it becomes impossible to avoid congestion. The formula for this is N=N+N(+N+...N). Solve for N. The math doesn't lie. It's impossible.

BitTorrent use is a great example. Because if you use BitTorrent, and you're a "Network Neutrality" proponent, you are likely a hypocrite. Why? Well many BitTorrent users deny or limit the amount of bandwidth they are willing to share with others. This is to avoid having their circuits congested by other Torrenters downloading content from them. That is not being neutral, you are degrading quality of service for others to limit the impact on you. Granted it is you, not the ISP making that decision, but from the perspective of the other Torrenters, what difference does that make?

Enter streaming services, particularly live-streaming services like IP Phones, Video Chats, and live broadcasts. These services cannot withstand any congestion at all. Network bandwidth has to be "guaranteed" in order for these services to work. Neutrality in this circumstance essentially means that more congestion tolerant services get downgraded network access because they can handle it better. Is that a neutral choice though? Those tolerant services will experience slower access. They still work, just not as fast.

This is where the reality of networking limitations run into the idealism of network neutrality. Limitations means we have to make choices. And depending on who is affected by those choices, not everyone will see it through the same neutral lens.

But let's look at it through a commercial lens. If the option was there to pay a little bit more to guarantee NetFlix, would you? If no, but someone else is willing to, can they? If not, why not? Why do you get to choose whether someone else can purchase better service? In a commercial marketplace, if someone is willing to provide a legal service, I'm allowed to choose if I want to pay for it. We have already accepted this model for television services. That's how Rogers Cable got started in Canada. Everyone had equal free access to broadcast VHF/UHF signals, but paying for cable provided more reliable service and more channels. Why can't that ideal extend to IP enabled services over the Internet?

Because the Internet is a public good. Right? It is an essential service required to live in a modern society.


That "essential service" isn't as essential as people say it is, or at least the way they mean it. Essential doesn't mean being able to download gigs and gigs of copyright content for free. Essential doesn't mean being able to play massive online video games without lag or jitter. Essential doesn't mean interruption free NetFlix binge watching.

Essential, at most, means access to e-mail and to Google and Wikipedia (yes I am being a little bit sarcastic). Those services do not require anything more than a 4 or 5 Mb internet circuit. People are not being entirely honest when they shout about "public good" and "neutral networks." What they really want is cheap fast access that allows them to get as much content online for as little money as possible. Preferably free, without ads or commercials. We have greedy consumers yelling about greedy Internet companies. Excuse me if I can't take that argument very seriously.

There are compelling arguments to be made for more transparency, and better competition (I've had some thoughts on that.) But as long as Internet access is a commercial enterprise, the rule "What the market will bear" will always apply. As consumers it's perfectly fine to bitch about costs, but if you're going to turn it into a moral argument, you better make sure your own shit don't stink.

And "Network Neutrality" is a dead idea. Choices will always have to made, and others will not like those choices. It's a reality. Deal with it.


Anonymous said...

Actually, your concept of Net Neutrality is what is wrong.

When I signed on to Bell Simpatico in 1997, I was buying access to the internet. Bell didn't have IP TV or anything other offerings - I was paying for a pipe to the internet. They were happy to provide it - I was happy to pay.

Internet providers rely on public right-of-ways - their cables run through municipally maintained land, and are accessed through tax-payers' backyards.

Network Neutrality is the necessary reality. If an ISP wants to charge me extra for special services, fine - if I sign up for them. If I sign up to use 10 gigs a month, after which I will pay an arbitrary increase, fine, but they should charge buy the bit, regardless of where those bits come from.

Network Neutrality means providing unrestricted access to the internet. If a company wants to provide restricted access to the internet, that might be OK, as long as they aren't relying on tax-funded land or services to provide it.

Ken Breadner said...

Thanks for this...I think most people still believe the Internet is some kind of magical data fairy that shouldn't have any limits on it at all because FREEDOM.
TANSTAAFL: There Ain't No Such Thing As A Free Lunch.
I keep hearing people say the Internet should be a public utility, and maybe they're even right. I fail to see how that would change anything as far as the logistics you have outlined go.

Catelli said...

Making part (or all) of Internet Communications a none-profit public utility has some benefits. To be honest, I happen to agree this is the route we need to go.

The problem with capitalist enterprise is two things 1) It is risk averse and 2) The goal of a business is to sell as little as possible for as much $$$ as possible. "What the market will bear"

Both of those points mean that our Telcos have little incentive to dramatically improve our communications infrastructure. As long as people are willing to pay for crappy DSL and Cable, why upgrade the circuits to Fiber?

A public utility (as done in other countries) would be charged with providing X bandwidth to everyone by Y date. It would also remove the FUD factor and more transparently show why and when congestion is happening. Private enterprise has no obligation to be completely honest about the performance of their network.

We'd still have part of the "Neutrality" debate (because that's the nature of networks) but it should become less frequent and we'd have better data from our public provider showing us why certain problems exist.

Ken Breadner said...

Fair enough. I think what gets people's goat is "willing to pay".
I come at things from a grocery perspective. Toilet paper. It's like what would happen if every retailer decided that toilet paper should be $3 a roll, then justified it by saying, well, look, people are willing to pay it. Maybe we should make it $4, I bet people would be willing to pay that, too.
The only difference is that in retail, somebody would go, wait a second, if I drop my price to $2 a roll I will corner the market on butt wipe. That never happens with ISPs and I think we all know why.

Catelli said...

Yep. When people say Canadian ISPs are protectionists benefiting from a closed market, I agree with them 100%.

ADHR said...

Net neutrality seems like on of those silly debates, like public vs. private healthcare. It's a stupid debate that ends up pivoting around a slogan, rather than an appreciation of the core issues.

In healthcare, the issue is that healthcare resources are scarce (probably necessarily scarce), and you need some sort of distribution principle. That can be ability to pay -- although a lot of people object to that -- or it can be on a triage basis -- most severe need is dealt with first -- or other possibilities. But "I'm for public healthcare" or "I'm for private healthcare" are both stupid, incoherent positions.

It's the same with net neutrality. Of course you have to manage the network -- bandwidth is limited, and you need to allocate it on the basis of some principle or another. Of course the networks are privately-owned, so (within some limits) the owners can do what they like with them.

The question is what limits, if any, should be placed on what they can do with their property; and thus what is the principle we use to manage the network.

What the net neutrality proponents get right is that the owners of the network are abusing their ownership position. They are also content creators, and they are using their distribution networks to push the content that they own, and degrade or eliminate competing content.

It's the good old tendency of capitalism towards monopoly (or near-monopoly). The content creators (primarily big media companies) have bought or built a distribution network, and bought or built a library of content, and are now trying to limit access to that network, in order to privilege their content libraries.

(Related problem: as the first comment noted, when we initially bought internet access back in the day, we were sold that it was a dumb pipe. To now change the rules and turn it into a walled garden is a pretty blatant bait and switch. I think that's a minor issue, but it's not a non-issue.)

So, the point of who owns the networks if a bit of a red herring. There's all kinds of privately-owned resources which are under government regulation, given their size and significance. Banks, for example; and, more to the point, railroads. Railroad companies can't just put anything on a rail car and send it along the lines. They also, interestingly, don't limit access to their lines -- if Ford wants to ship its cars on CP Rail's lines, CP Rail isn't going to tell Mercedes to go to hell. CP will cheerfully carry any automobile whose manufacturer is willing to pay for the use of the line.

But, CP Rail isn't a monopoly. They have a big national competitor for provision of a distribution network; and, they don't make the things which travel on that network. So, they have a big incentive to ensure that as many manufacturers can access their lines as possible, and no incentive to privilege their own products (because they don't have any products!).

Similarly, the idea that people should be able to pay more to access more is also a red herring. That only works if what you want to buy is actually for sale. In a truly free market, everything is for sale at a price. But a market dominated by a handful of competitors is, by definition, not free (not even reasonably or relatively free) -- the dominant competitors can crowd new entrants out, which defeats one of the hallmarks of a free market. So in the market as it is, some things just aren't for sale.

Netflix is an interesting case -- as it's not owned by a media company, it may end up getting forced off the network (either in fact, or in effect, such as by having its service degraded to the point that customers switch to alternatives owned by the media companies). And what about the next Netflix? If I start up a new internet service, then I have to negotiate my way around whatever exclusive deals exist already; otherwise, my customers won't get the option to purchase a sufficiently high level of access to use my service effectively.

ADHR said...

One way governments could deal with this (near-) monopoly is to create a public utility to compete. If memory serves, though, some ISPs in the US have actually sued to try to stop municipalities from doing so. So this might be very difficult from a legal perspective. It's also possible that content creators (i.e., the same big media companies who don't like these public ISPs) will refuse to provide their content on the public networks, which winds back to the problem of their monopoly.

I think the real solution is to acknowledge that big media companies are just too damn big. Their distribution networks are too big, and they control too much content -- they're just too damn powerful.

So, the companies need to be broken up. This can be done by breaking up their size -- so, national ISPs have to be broken up into regional ISPs. Or, they could be forced to split their businesses -- i.e., you can either be a content creator, or a distributor, but not both. (At least, past a certain size. You'd have to finesse the regulation a bit to make sure that you weren't cracking down on some poor guy self-publishing his own comic book or something.)

Of course, they don't want to lose their monopoly, so they will fight tooth and nail to keep it.

Catelli said...


Well put, as always. There is a huge issue with competition, and dominance in Canada's ISP space. But my issue is framing that debate under the Net Neutrality red herring, which is why I composed this post in the manner I did.

Fixing this issue will require some kind of regulation, and I happen to think that breaking up the huge media monoliths and preventing content creators from owning content distribution is a good idea. I would even go one step further and separate access to the "internet core" from the "last mile" and not allow one company to dominate both. But focusing regulation on ensuring "Neutrality" is a horrid idea. It doesn't actually address the root issues, and when it comes down to it, people can't agree on what Neutrality truly means.

Neutrality principles should arise as a property of the network. Properly limiting who can own what (or even make part or whole a public utility) would go a long way to ensuring that.

But right now, what Telus is doing is entirely legal, and the Neutrality argument as expressed, is the wrong way to illustrate what is wrong with Telus both distributing and creating content.

Keith said...

Though I'm someone who leans pro-net neutrality, I appreciated the way you framed this.

Principles like net-neutrality shouldn't be ends in themselves, but ways to get at the real goals: promoting innovation and open communication. An implementation should be judged on that basis.

An analogy can be drawn to road transportation. Are roads an end in themselves? No, the point is to promote commerce and trade. As in your network example, congestion is a problem that can't be regulated away. But suppose more people telecommute than drive. Is it bad that fewer people are making use of roads? No, the needs of people finding ways to do business are more important than a quota to have more lanes and streets.

You also bring up a great point about unbundling the last mile. Have you read on how this is done in other countries?

Catelli said...


I actually lean Net Neutrality myself. Network Management is a PITA from personal experience. ;)

But as ADHR points out, a label as a rally point for a popular movement leads to silly arguments. Debate about what we want/need from our communication infrastructure needs to be more focused.

As far as unbundling the last mile, the (little) research I've done suggests this was baked into the telco model adopted from the beginning. I.e., it was never bundled to begin with. My guess or strong suspicion is that North American ISP issues are somewhat unique as we have a lot more legacy design/sunk-costs/habits to overcome. Which, from my point of view, means we need to have more urgent focused debates about fixing it than we have had up until now. The Network Neutrality shout-fest is getting in the way, and is working against us, not for us.