Comcast and Netflix announced a deal yesterday that the companies say is all about providing “Comcast’s U.S. broadband customers with a high-quality Netflix video experience for years to come.” And to a large extent, the deal is going to do just that, eliminating a major bottleneck that has caused buffering, low video-quality, slow start times and other misery for Comcast subscribers trying to watch House of Cards on Comcast recently. But because the deal apparently involves money changing hands — from Netflix to Comcast, though neither party is confirming that, it’ll be clear in a moment why that’s precisely what’s happening — it has network-neutrality advocates in a frenzy about the impending death of the open internet. The good news is (1) this has nothing to do with net neutrality (2) the deal being struck looks an awful like scores of deals done before it. The bad news is as Comcast prepares to swallow Time Warner Cable, the Cassandras of the world aren’t entirely off base in fearing a world with internet haves and have-nots.
Don't Miss: The best Deal on a new Samsung Galaxy S8
So what the heck is happening anyway?
Because Netflix is so popular, with 33 million customers in the U.S. alone, it’s a prodigious generator of internet traffic, accounting for nearly 30% of all the bits moving around, according to Sandvine. To get all that traffic to folks like you and I, the company relies on a something called content-delivery networks (CDNs) that cache its content broadly around the country. Some of it is stored with third parties, companies like Level 3 and Limelight. Much of it is stored by Netflix itself, but not on infrastructure it completely controls. Netflix has its own CDN that exists on capacity the company leases from infrastructure providers like Cogent Communications. It also has equipment inside the networks of some internet providers, including Google Fiber, RCN and Cablevision. Netflix calls this Open Connect but what’s important about is that by getting the content burrowed deep inside the networks, it gets clear of as many traffic jams as possible on the way to you.
Comcast, until yesterday, wasn’t interested in Open Connect — or really any solution that was going to help make Netflix work better. And part of the reason was about money, but partly it was about principle. The reason goes to a somewhat arcane principle on the internet known as peering. Normally, when a really big internet company like Comcast needs to move traffic between itself and a big “backbone” provider like Level 3, it makes a peering deal to do that. As the word implies, the two companies are more or less equal and so peering has historically been done without dollars coming with the traffic. Each sends traffic to the other and all is good.
Gorillas in the mesh
But wherever Netflix went, the concept of peering would go out the window. And conflicts arose. Most recently, Cogent and Verizon have been in a dispute over peering, which is supposedly leading to trouble for Netflix customers. as ArsTechnica reports. Cogent, you see, is carrying a lot of Netflix traffic and the story is that Verizon would be happy to upgrade its connections to Cogent, but only if Cogent pays for it — not just once, but on a continuing basis. Exactly how much of this is true is open to debate; Cogent is often at the center of peering disputes. But the key here is that peering is becoming more complicated as relationships change over time. Verizon is, as the Washington Post article linked in the first paragraph notes, both a backbone provider and a consumer ISP. It carries lots of traffic. It’s becoming more and more unequal to more and more former “peers.”
The FCC, for its part, knows about this. But when it originally imposed the recently struck-down net-neutrality rules, it actually chose not to try to regulate peering. The neutrality rules were concerned, instead, with how ISPs treat traffic that reaches you. The idea was that all data should be treated equally and that an ISP couldn’t, for example, block Pandora from reaching your home without an extra fee while letting Spotify through for free (perhaps because Spotify paid Comcast to make that happen).
Monkeys in the wrench
What’s happened is that traffic-carrying charges have been a potential de facto pay-to-play cost to get services to you. Except the key word there is potential. Today, Netflix is paying Comcast something that is likely in the millions of dollars. And it’s not paying a bribe, but rather the very money it would be paying someone else in the CDN business. This is critical to understand amid all the hand-wringing. Consider this misapprehension from The New York Times:
Tim Wu, a Columbia Law School professor and advocate for net neutrality, said the interconnection agreement between Comcast and Netflix was one of the first such arrangements where a broadband provider like Comcast has extracted payment to send specific content through the “on ramp” to its network. “This is the water in the basement for the Internet industry,” Mr. Wu said, the first in what could be a flood of such arrangements. “I think it is going to be bad for consumers,” he added, because such costs are often passed through to the customer. One fear is that if such deals become common, only the wealthiest content companies will be able to afford to pay for them, which could stifle the next Netflix from ever getting off the ground.
Wait, what? Netflix is paying this because it’s so wildly successful. It’s not even possible a startup would be in this position. It would, instead, pay a third-party CDN to carry its tiny traffic until it theoretically grew to a size where it mattered. Then it would likely follow in the footsteps of Netflix, Google and soon Apple by building its own CDN and make its own traffic deals according to the asymmetry of how much it was sending versus what it was receiving. And that’s key here. Netflix is almost entirely a sender of bits. You request a program with very little data, perhaps a few hundred bytes, but in return you can receive gigabytes — millions of times as much or more. For years, Netflix has been spending real money to distribute its content to CDNs to make sure those gigabytes get where they need to be. And those CDNs have borne the cost of the traffic.
There will be numerous calls now for government regulation of the backbone side of the internet, much as network neutrality tried to regulate the “last mile.” Even with the FCC’s authority kneecapped, Comcast is still under neutrality rules as a condition of its merger with NBC. It’s likely those will continue into any approvals of the deal proposed with Time Warner. What’s less likely is that stopping the merger — which about 2/3 of the comments on most articles about this deal seem to want — will do anything to make the peering and traffic payments problems disappear. And solutions like declaring cable companies “common carriers,” might create broad regulatory authority but seem unlikely to ultimately be favorable for innovation. The pre-breakup AT&T wasn’t exactly adept at pushing the envelope at technological change even though it had some of the most brilliant minds in the country developing technology at Bell Labs.
The Big C
The best way to get pro-consumer change would be to have robust competition for consumer broadband. In a lot of areas, that’s already hard but trends are marginally favorable. Where Verizon offers Fios, it’s the better ISP than the cable incumbent. AT&T either needs to get its U-verse build completed and upgraded to compete or sell off the network to someone who will; the asset is in tens of millions of homes. Fiber is slowly becoming more available from upstarts, too. While wireless advocates hold out hope, the idea of “last mile” access providing high-definition video wirelessly to a whole neighborhood will require something miraculous like Steve Perlman’s pCell.
In theory, the government could build a wholesale fiber network to every home and business in the U.S. and lease it out to all comers. That would allow for multiple competitors while virtually no regulation of any kind would be required. It could even create an “open peering system” on the backbone of such a network. Of course, the politics of a $100 billion infrastructure program might be more impossible than believing that Comcast-Time Warner could ultimately be good for customers (through increased bargaining power over channel owners). Google recently announced an expansion of its fiber network, but at the rate its going, the country will be covered by sometime in the next century. Of course, if somehow other deep-pocketed tech companies with interests in greater network connectivity and perhaps openness– think Apple, Microsoft, Cisco — got involved, that might be advanced to 10 years from now.
For now, the deal between Comcast and Netflix is actually a qualified win for consumers. But neither party did this out of the goodness of their hearts. Netflix needs good video quality to keep subscribers happy. This deal will help make that happen for Netflix’s customers on Comcast — a base that could someday be a majority of the post-merger 30 million. For Comcast, this will help sell more broadband and establish a precedent that there’s a tipping point beyond which you can’t drive on its roads without paying a toll. We’ll see how frightening that is someday. But not for a while.