Monday, November 10, 2014

This Vox NetNeutrality article is wrong

There is no reasoned debate over NetNeutrality because the press is so biased. An example is this article by Timothy B. Lee at Vox "explaining" NetNeutrality. It doesn't explain, it advocates.

1. Fast Lanes

Fast-lanes have been an integral part of the Internet since the beginning. Whenever somebody was unhappy with their speeds, they paid money to fix the problem. Most importantly, Facebook pays for fast-lanes, contrary to the example provided.

One prominent example of fast-lanes is "channels" in the local ISP network to avoid congestion. This allows them to provide VoIP and streaming video over their own private TCP/IP network that won't be impacted by the congestion that everything else experiences. That's why during prime-time (7pm to 10pm), your NetFlix streams are low-def (to reduce bandwidth), while your cable TV video-on-demand are hi-def.

Historically, these channels were all "MPEG-TS", transport streams based on the MPEG video standard. Even your Internet packets would be contained inside the MPEG streams on channels.

Today, the situation is usually reversed. New fiber-optic services have TCP/IP network everywhere, putting MPEG streams on top of TCP/IP. They just separate the channels into their private TCP/IP network that doesn't suffer congestion (for voice and video-on-demand), and the public Internet access that does. Their services don't suffer congestion, other people's services do.


The more important fast-lanes are known as "content delivery networks" or "CDNs". These companies pay ISPs to co-locate servers on their network, putting servers in every major city. Companies like Facebook then pay the CDNs to host their data.

If you monitor your traffic, you'll see that the vast majority goes to CDNs located in your city. When you access different, often competing companies like Facebook and Apple, your traffic may in fact go to the same IP address of the CDN server.

Smaller companies that cannot afford CDNs most host their content in just a couple locations. Since these locations are thousands of miles from most of their customers, access is slower than CDN hosted content like Facebook. Pay-for-play has, with preferred and faster access, has been an integral part of the Internet since the very beginning.

This demonstrates that the Vox example of Facebook is a complete lie. Their worst-case scenario already exists, and has existed since before the dot-com era even started, and has enabled competition and innovation rather than hindering it.

2. Innovation

Vox claims: "Advocates say the neutrality of the internet is a big reason there has been so much online innovation over the last two decades".

No, it's opponents who claim the lack of government regulation is the reason there has been so much online innovation in the last decades.

NetNeutality means sweeping government regulation that forces companies to ask permission first before innovating. NetNeutrality means spending money lobbying for government for special rules, surviving or failing based on the success of paying off politicians rather than surviving or failing based on the own merits.

Take GoGo Inflight broadband Internet service on airplanes. They block NetFlix in favor of their own video streaming service. This exactly the sort of thing that NetNeutrality regulations are supposed to block. However, it's technically necessary. A single person streaming video form NetFlix would overload the connection for everyone else. To satisfy video customers, GoGo puts servers on the plane for its streaming service -- allowing streaming without using the Internet connection to the ground.

If NetNeutrality became law, such things would be banned. But of course, since that would kill Internet service on airplanes, the FCC would immediately create rules to allow this. But then everyone would start lobbying the FCC for their own exceptions. In the end, you'd have the same thing with every other highly regulated industry, where companies with the most lobbying dollars win.

Innovation happens because companies innovate first and ask for permission (or forgiveness) later. A few years ago, Comcast throttled BitTorrent traffic during prime time. NetNeutrality proponents think this is bad, and use it as an example of why we need regulation. But no matter how bad it is, it's a healthy sign of innovation. Not all innovations are good, sometimes companies will try things, realize they are bad, then stop doing them. Under NetNeutrality regulations, nothing bad will happen ever again, because government regulators won't allow it. But that also means good innovations won't happen either -- companies won't be able to freely try them out without regulators putting a stop to it.

Right now, you can start a company like Facebook without spending any money lobbying the government. In the NetNeutrality future, that will no longer be possible. A significant amount of investor money will go toward lobbying the government for favorable regulation, to ask permission.



3. What's Taking So Long

Vox imagines that NetNeutality is such a good idea that the only thing stopping it is technicalities.

The opposite is true. The thing stopping NetNeutrality is that it's a horrible idea that kills innovation. It's not a technical idea, but a political one. It's pure left-wing wing politics that demands the government run everything. The thing stopping it is right-wing politics that wants the free-market to run things.

The refusal of Vox to recognize that this is a left-wing vs. right-wing debate demonstrates their overwhelming political bias on this issue.

4. FCC Bypassing Congress

The Internet is new and different. If regulating it like a utility is a good idea, then it's Congress who should pass a law to do this.

What Obama wants to do is bypass congress and seize control of the Internet himself.

5. Opponent's arguments

Vox gets this partly right, but fundamentally wrong.

The fundamental argument by opponents is that nothing bad is happening now. None of the evil scenarios of what might happen are actually happening now.

Sure, sometimes companies do bad things, but the market immediately corrects. That's the consequence of permission-free innovation: innovate first, and ask for permission (or forgiveness) later. That sometimes companies have to ask for forgiveness is a good sign.

Let's wait until Comcast actually permanently blocks content, or charges NetFlix more than other CDNs, or any of the other hypothetical evils, then let's start talking about the government taking control.

6. Red Tape

Strangling with red-tape isn't a binary proposition.

What red-tape means is that network access becomes politicized, as only those with the right political connections get to act. What red-tape means is that only huge corporations can afford the cost. If you like a world dominated by big, connected corporations, then you want NetNeutrality regulations.

While it won't strangle innovation, it'll drastically slow it down.

7. YouTube

Vox claims that startups like YouTube would have difficulty getting off the ground with NetNeutrality regulation. The opposite is true: companies like YouTube would no longer be able to get off the ground without lobbying the government for permission.

8. Level Playing Field

Vox description of the NetFlix-Comcast situation is completely biased on wrong, taking NetFlix's and leftist description at face value. It's not true.

Descriptions of the NetFlix-Comcast issue completely ignore the technical details, but the technical details matter. For one thing, it doesn't stream "across the Internet". The long-distance links between cities cannot support that level of traffic. Instead, NetFlix puts servers in every major city to stream from. These servers are often co-located in the same building as Comcast's major peering points.

In other words, what we are often talking about is how to get video streaming from NetFlix servers from one end of a building to another.

During prime time (7pm to 10pm), NetFlix's bandwidth requirements are many times greater than all non-video traffic put together. That essentially means that companies like Comcast have to specially engineer their networks just to handle NetFlix. So far, NetFlix has been exploiting loopholes in "peering agreements" designed for non-video traffic in order to get a free ride.

Re-architecting the Internet to make NetFlix work requires a lot of money. Right now, those costs are born by all Comcast subscribers -- even those who don't watch NetFlix. The 90% of customers with low-bandwidth needs are subsidizing those 10% who watch NetFlix at prime time. We like to think of Comcast as having monopolistic power, but it doesn't. The truth is that Comcast has very little power in pricing. It can't meter traffic, charging those who abuse the network during prime time to account for their costs. Thus, instead of charging NetFlix abusers directly, it just passes its costs to NetFlix.

Converting the Internet into a public-utility wouldn't change this. It simply means that instead of fighting in the market place, the Comcast-NetFlix battle would be decided by regulators. And, the result of the decision would be whichever company did the best job lobbying the FCC and paying off politicians -- which would probably be Comcast.















8 comments:

Roland Dobbins said...

There is no such thing as a 'fast lane' on the Internet (or on private networks, for that matter); you can't make traffic go faster.

What you can do is to use QoS to rate-limit traffic based upon various layer-4 characteristics.

So, you can police down some traffic, whilst not policing down other traffic. I guess you could call it a 'fast lane via negation', but what you're really doing is simply giving some traffic priority over other types of traffic within the context of finite network bandwidth available via a given network link/path of links.

This illustrates one of the many ways that the brouhaha over 'net neutrality' is partially the result of an incorrect understanding of how the Internet works. Couple that with non-technical political motivations on all sides of the debate (such as it is), and you end up with people who don't know what they're talking about making policy recommendations which will have unexpected (to them) negative consequences.

Evan said...

Net Neutrality or not, the more important point being missed is the overall state of Internet quality in the U.S. I live in the UK and I get 120mb down via Virgin Cable. That's 15 megabytes a second. Wifi is typically the bottleneck in parts of my house. I get at least 90% of that as usable bandwidth during peak hours. All of my friends in the U.S. don't see close to those levels of service. Why is that? Is it ISP monopoly in most of the U.S.? Is it the "socialist" UK government forcing companies to improve their networks? Why do the "inventors" of the Internet lag so far behind the rest of the civilized world in terms of Internet speed and cost to consumers? I have no answers on Net Neutrality, but the idea that there is "nothing wrong" with the way things are in the U.S. today seems inaccurate from this side of the world. Good luck to those of you suffering with the Comcasts and Time Warners of the world. I've been there and it was miserable.

Katie said...

Evan -- I'm going to guess it has to do with 3 things:

1. distances -- I mean, there are physical distance limits to wires/etc.. I'm not sure why people think wires are magic and have no energy loss. The US is huge. That has to have some effect? Maybe I'm just obsessed with wires these days.

2. No competition. In the US you frequently have the choice of maybe 1-3 content providers, so they don't need to do much to make themselves slightly better.

Because:
3. Local loop -- the UK forced the local loop to be shared, if I recall? I think it was for a while in the US (Mid 90s to mid 00s?), but it no longer is. After that happened, our choices went from dozens to that 1-3 we have now. No competition, because only one company owns the last mile.

Stop me if I'm wrong, but when the idea of "net neutrality" was coined, it was about the local loop, not about QoS.

Scott said...

Honestly, I think you both make good points. Inman's main point (and similarly, John Oliver's from the much shared "Last Week Tonight" video) is that consumers want their ISPs to provide a dumb pipe of internet bandwidth, with little to no restrictions on where they go or what they do with it. The Netflix graph is certainly not a neutral source, and therefore should be taken with a decent dose of salt. However, there are several reports of Netflix performance improving when using some kind of VPN tunnel service out there. Likewise for Youtube and other streaming services at various points in the past. All those stories point to common theme. ISP's degrading services they don't like, in favor of ones they do. In a truly competitive market, consumers would just migrate away from the ISPs that did this to ones that didn't. Unfortunately, in most parts of the US, this isn't an option. When consumers don't have a market option, they turn to government to step in.

Your technical points are correct, and I agree with most of your interpretations. Your point about GoGo inflight internet is an especially salient one. That's one of the major problems with any legislation around this problem is all the edge cases. In-flight internet is a good case. Hotel internet has also popped up recently as preventing certain traffic (VPN's mainly). Another one I haven't seen discussed is public Wifi a la coffee shops and such. Who decides what traffic travels over "their" network? What constitutes an "ISP" for Net Neutrality's sake?

Unknown said...

Your example of the GoGo inflight broadband doesn't hold water. GoGo isn't giving *preference* to their own video streams over the air-to-ground link. Rather, they're not *using* the air-to-ground link because they keep their servers on the *plane*. They don't need to *block* Netflix; they only need to give each user such a low bandwidth (or charge so much per MB) that streaming video from *anybody* over the link is just not feasible.

I also see problems with your assertion that "the market immediately corrects" when companies do bad things. As exhibit A, look at the banks leading up to 2008. That's an example where: 1) the correction wasn't immediate, and 2) the correction was pretty painful. I think your faith in the free-market as a self-driving car which automatically picks the best route is unfounded. However, in this case, the assertion is even more outlandish. So, you want me to believe that, when the ISP's start monkeying with Netflix's throughput during contract negotiations, that entrepreneurs will spot this market opportunity and a new crop of nation-wide ISPs with billions of dollars of switching infrastructure will just "sprout up" to meet the demand? That isn't going to happen. The current crop of ISPs are the result of decades of mergers and acquisitions of smaller providers. The number of providers has gotten fewer, not larger. Nobody can just jump in and roll out a competing product. Even *Google*, with mountains of cash, is rolling out internet provision (via their fiber) at the rate of a few cities per year. The free market... doesn't. work. here.

Regarding your point about "Re-architecting the Internet to make NetFlix work requires a lot of money"... Comcast shouldn't be bearing those costs to make *NetFlix* work. That's NetFlix's problem (which they can address with CDNs and whatnot). However, if the advent of NetFlix causes the *aggregate* of internet traffic traversing Comcast's networks to become too large, then that's Comcast's problem; Comcast has been selling a certain throughput of broadband service to its users, and now increasing numbers of those users are trying to use what they've paid for.

The crux of net-neutrality, as I understand it, is that the ISP's are able to control the data flow of a lot of users, and they're able to use that as a bargaining chip. If they have, say, 10 million subscribers, they can go to NetFlix and say "Gosh, it'd be a *shame* if 10 million of your customers suddenly started having streaming issues... and only with *your* servers". In itself, that's bad enough. However, to make things worse, the customers can't easily punish Comcast for these actions against the users' interest (ie, the market can't correct) because the ISPs, effectively, have monopolies. A Comcast customer usually cannot switch to Time-Warner or Charter. In other words, the ISPs are able to control how the resources of millions of users are directed, and they're able to use that control to enrich themselves to the detriment of the users and the services they're trying to use.

In this way, this scenario bears some resemblance to corruption in public-works projects. In those cases, a public official is controlling how the citizens spend their tax dollars to, say, build a bridge. The public official is able to solicit bribes or favors from the contractors in order to win the contract, but this is working against the best interests of the citizens and the contractor.

Wheaties said...

Where are you getting your asinine statement about Obama. I'm no fan of the guy but I read thoroughly through his press release and no where in there can any sane person come to this conclusion.


Overall very good article that touches on some very needed points. Most of the readers here are missing the points.

1. Fast lanes happen, that is not what net neutrality is about. Its about preventing slow lanes.

2. The US have a very large internet backbone and infrastructure. Its very difficult to get local municipalities to get permits to work and upgrade the lines in the ground. They are upgrading them, it just takes time. an Example is right here. I'm slated to move up to 150Mbps download early next year.


I'm not talking about you robert but im frankly very sick of "people in technology" getting net neutrality completely wrong.

Sincerely, A ISP Network engineer

Unknown said...

A few comments for Wheaties:
1 - I'm not sure anybody can be "getting net neutrality wrong" when everybody seems to have a different definition of it. To some, it means that Comcast can't niggle around with Netflix's throughput during contract negotiations. To others, it means that ISP's can't let websites which are NOT their customers pay for faster throughput. For a simplistic example, suppose that all of Amazon's servers are connected to the internet through AT&T. Now, suppose that Amazon is able to pay Charter, Time-Warner, Comcast, Google-Fiber, whoever-else extra money to give preferential treatment to traffic to/from Amazon. This is something (first, because it is cost-prohibitive and, 2nd, because they don't have the resources to negotiate/maintain contracts with dozens of ISPs) which a mom-n-pop web store just won't be able to do. So, Amazon, simply because of their size, enjoys a competitive advantage which locks out the smaller guys. It's akin to BestBuy being able to add a special lane on all of the freeways reserved just for people going to BestBuy. If that's viewed as anti-competitive in the corporeal world, why isn't it considered the same in the virtual world?

Lastly, your comment about "Fast lanes happen... it's about preventing slow lanes". They go hand-in-hand, since fast and slow are subjective. Compared to 10 years ago, everything, today, is fast. If everybody else has a "fast lane" and I don't, that's going to feel (to me and to anybody who connects to me) as being *slow*. In other words, allowing "fast lanes", necessarily, creates "slow(er) lanes".

Unknown said...

A few comments for Wheaties:
1 - I'm not sure anybody can be "getting net neutrality wrong" when everybody seems to have a different definition of it. To some, it means that Comcast can't niggle around with Netflix's throughput during contract negotiations. To others, it means that ISP's can't let websites which are NOT their customers pay for faster throughput. For a simplistic example, suppose that all of Amazon's servers are connected to the internet through AT&T. Now, suppose that Amazon is able to pay Charter, Time-Warner, Comcast, Google-Fiber, whoever-else extra money to give preferential treatment to traffic to/from Amazon. This is something (first, because it is cost-prohibitive and, 2nd, because they don't have the resources to negotiate/maintain contracts with dozens of ISPs) which a mom-n-pop web store just won't be able to do. So, Amazon, simply because of their size, enjoys a competitive advantage which locks out the smaller guys. It's akin to BestBuy being able to add a special lane on all of the freeways reserved just for people going to BestBuy. If that's viewed as anti-competitive in the corporeal world, why isn't it considered the same in the virtual world?

Lastly, your comment about "Fast lanes happen... it's about preventing slow lanes". They go hand-in-hand, since fast and slow are subjective. Compared to 10 years ago, everything, today, is fast. If everybody else has a "fast lane" and I don't, that's going to feel (to me and to anybody who connects to me) as being *slow*. In other words, allowing "fast lanes", necessarily, creates "slow(er) lanes".