Should we worry that Netflix is buying transit rights from Comcast?

I say no (for background read here, and Dan Rayburn has very useful coverage, dispelling a variety of myths).  To be sure, one may believe there are monopoly problems at the retail level in the cable sector.  We could alleviate those with local loop unbundling and/or deregulation, as I discussed yesterday.  But within that setting, Netflix paying Comcast won’t make that monopoly worse.

In support of this conclusion, I would cite two literatures.  The first is Ronald Coase’s analysis of payola.  If a gatekeeper can extract payments from input suppliers, the end result of that process need not be bad for consumer welfare and very often is positively good for consumers.  In a nutshell, the gatekeeper won’t want to exclude the programs which consumers really want.  Those programs contribute to the profits of the gatekeeper.

The second literature is that on double marginalization.  This literature considers settings where you have a retailer with market power and an input supplier with market power, and the input supplier needs the retailer for access to consumers.  In those cases either integration or Coasean bargaining is usually in the interests of consumers, as it minimizes the double mark-up and thus lowers the costs of the market power.  To put this more concretely, the two parties will deal so that the marginal cost of the input to the monopolist is lowered, the monopolist expands output (and profit), and those gains are shared between the two institutions with market power.

When you put those two theories together, Netflix buying transit rights from Comcast is likely fine.  Don’t translate your opposition to cable monopoly into opposition to this agreement.  Seton Motley asks a good question:

Should we worry Amazon (Prime) buys transit rights from UPS & USPS?

So, if someone criticizes this new deal, but cannot put the argument in Coasean language, they probably have not thought it through carefully enough.

Moving beyond that, can we think of reasons why the basic Coasean results may not hold?

1. Expected joint profit doesn’t map perfectly well into consumer surplus.

2. The status quo ex ante was based on some amount of queuing of Netflix access, rather than a dollar-based marginal cost for selling the input, as in basic Coasean models.

3. None of these transactions are purely Coasean when regulatory threats beckon and thus a wider range of outcomes is possible.

4. Comcast has read Doug Bernheim and can now construct a scheme to preempt Netflix from this market altogether,

I’ve pondered those long and hard, but the standard Coasean results still seem reasonably likely.  And if they are not, it would be for reasons so convoluted it is unlikely to represent anyone’s actual worry.  From that list only #4 seems to have any bite, but I don’t see that #4 applies empirically.  Netflix has risen greatly in value over the last year, this new development is hardly a surprise, and the fees to Comcast, while secret, have been described as “de minimis.”  There is quite a good chance that Netflix benefits from this deal and this is more of a “we are here for good” statement than Netflix falling off a cliff.

You also could try this argument:

5. By agreeing to pay a price for transit rights, Netflix imposes a negative pecuniary externality on smaller streaming services and content providers, which in the longer run will mean a negative non-pecuniary externality for variety-seeking consumers.

Maybe, maybe so.  I do take that argument seriously.  But it’s also unconfirmed, Coase on payola implies it won’t be so bad, and furthermore the Netflix transaction, in stand-alone terms, still would seem to be welfare-improving.  Besides, what happens if Netflix expands by 5x or 10x, should the company never have to pay anything?  Is it so terrible if tomorrow’s necessary equilibrium shows up today?

Addendum: Timothy Lee offers a different perspective.  Perhaps I am failing to understand his argument, but I don’t see why having a cluster of mid-level intermediaries should make the market as a whole more competitive.

Or you may be tempted to write a screed about the dangers of moving away from net neutrality, and associate this development with that movement.  I say you would do better to stick to the specific economics of this particular issue and explain, in terms related to the Coasean model, exactly what will go wrong.

Further addendum: Joshua Gans offers comment.


I say no we should not be worried. The ball is in your court, opponents. TC makes a strong case and like a true master can play both sides of the board.

I agree with the Tyler's conclusion (that this deal is nothing to worry about) but the technology of implementation also makes a difference here. Instead of the naive model of data interconnection Tyler is considering, where fatter pipes carry more Netflix data from a central data center to Comcast's customers, what will actually be installed is an infrastructure of data caching servers throughout Comcast's network that will serve up content locally, avoiding the need for more expensive long-haul data transfer of the same traffic over and over again each time a customer wants to watch something. This traffic reduction will result in substantial cost savings overall. So it's a Coasean situation but the externalities in this case are positive and not negative.

If the Comcast pipes were as wide as advertised, as big as the monthly bill said they were, would any more infrastructure have been needed?

The link that TC posted above from Dan Rayburn has an excellent explanation of why this is not the case. (By the way, I retract my comments about 'naive model', I should have read through all the links before commenting).

I'm an old programmer, and think I read Rayburn with fair understanding. I think what he said is that yes, many companies provide services to deliver ... well he calls it "throughput" doesn't he? He even goes off with "SPEED AND THROUGHPUT ARE NOT THE SAME THING".

Now, what does your cable internet contract say? Mine says I get 20 (or is it 30?) Mbps. Of course, using the speed test apps I've never seen about 16.

So ... Rayburn is in sense saying "don't worry, you never get what you think you are getting anyway" and "this deal is only get give you a little more of what you thought you'd already paid for."

(I think practice is pretty close to theory, that backbones can be treated as infinite speed. Both providers and recipients buy access to that backbone. If providers don't want to buy one crazy big connection at home base, they can set up local servers around the world and redirect to them. So when I try for I don't get New Jersey or wherever, I get an Akamai server or whatever here in California. And sure, Netflix wants local servers. But that sure doesn't justify "speed is not throughput" sophistry.)

Going back even further, our local providers sold us "10 Mbps" not based on a true capacity that they could deliver 10 Mbps to all of us at the same time. They used a statistical model to provide their own trunks and branches with enough bandwidth that we each could think we had 10 Mbps when we used it. Now, I can feel for the providers a bit, when popular technologies come along. It happened when t he internet went from text to graphics, again from graphics to video, and now from video to HD. At each junction the statistical model changed. Our actual usage of our "10 Mbps" account approached 10 Mbps more often.

This is not "Netflix's fault." This is HD streaming, regardless of source.

Now, if my contract was "real" and I really got 10 Mbps whenever I wanted it, I'd be fine.

"Now, if my contract was “real” and I really got 10 Mbps whenever I wanted it, I’d be fine. "

Pay for a dedicated business class line. That's the point. A consumer line is similar to the older phone network shared phone lines, for exactly the same reason.

And does your contract really say it's a 10Mbps line? Because when I signed up my contract said "Download speeds of up to XXMbps and upload speeds of up to YYMbps".

So JWatts, you are seriously fine with a "pig in a poke" contract? If you pony up for mid-tier speeds, and they just tell you "hey, Download speeds of up to XXMbps and upload speeds of up to YYMbps” then you are good?

Remember, this is not market competition we are talking about. I'd be fine too if I could measure my connection, find it lacking and then shop elsewhere.

But what you are defending is "pay us what we ask, and we will actually be mum on what 'throughput' you get."

Commenter "fwiw" below has a 20 Mbps line that won't reliably deliver video. Is your answer to him and his neighbors that they should all buy dedicated lines?

"So JWatts, you are seriously fine with a “pig in a poke” contract? If you pony up for mid-tier speeds, and they just tell you “hey, Download speeds of up to XXMbps and upload speeds of up to YYMbps” then you are good?"

How is a contract that works as advertised a "pig in a poke” contract?

Easy answer, JWatts. Rayburn is right that some final, bottom line, functional speed is what matters. We don't care about mystical speed. We want to understand real speed.

If your contract is silent on real speed, you are buying a pig in a poke.

You might end up like poor "fwiw" with a "20 Mbps" contract that cannot even provide 3 Mbps.

Do you think that comcast doesn't admit that it is shared connectivity? They've never denied that to me. They also sell dedicated bandwidth. Its just a lot more expensive.

Well, as a monopoly provider they are able to make a pretty nebulous claim about what we've bought. In fact, they can try to put the shoe on the other foot and say that people who used advertised bandwidth too often are "hogs." Is a plain old Netflix subscriber a "hog" these days?

One advantage to not having "net neutrality" is that Netflix has more bargaining power relative to Comcast than Comcast's retail customers have, so they could, for example, insist that the fees they pay Comcast be partially used for capacity expansion.

A) Comcast is investing in capacity expansion all the time
B) money is fungible

Guys (everyone on this thread), this Netflix deal is inre to PEERING congestion, not local shared medium (DOCSIS CM to CMTS) plant congestion. Netflix is NOT getting a specific service flow (a way to differentiate and, thereby, prioritize traffic on the DOCSIS plant). This helps Netflix avoid PEERING congestion (Google does this RIGHT NOW) at Transit peering points that Comcast currently pays Cogent and other "rent seekers". Comcast (like Verizon, AT&T, etc.) have EYE BALLS. They have customers who "suck content" off the of the Internet. This deal BYPASSES the Cogents (and bullshit company btw who can't afford to upgrade their peering points to Comcast, et al) and this deal improves BOTH Netflix's customer's experience AND Comcast's by NOT bottlenecking these other peering points with their traffic (between 20-60% of all Internet traffic at various times of the day).

False. As noted below, the Comcast deal places Netflix servers WITHIN the Comcast local network.

they are not going to place them below the CMTS.

Sure Emil, above the CMTS would seem like the place where they could both provide preferred access and monitor (or shape) that usage.

So why are you contesting JR for saying exactly this?

On payola, the music industry in China was described to me like this:

Label must buy the radio time, to get their artists played.

The labels finds brand sponsors to pick up the tab.

The artists tours and gets a piece of ticket t-shirt sales.

The label and venue get the rest.

The brand is highly visible at venues.

Theres no shortage of musicians who want the deal.

On the other hand with so many countries, such as South Korea, making strides in offering quality and relatively cheap cable and broadband serivces I would be a little cautious of making sure high priced media doesn't make the US a less attractive place to emigrate to or do business in.

50% of South Korea lives in the Seoul metro area (20% within Seoul itself) so it's very easy to wire those people up.

In 2010, a total of 80.7 percent of Americans lived in urban areas, up from 79 percent in 2000.

When all your country's people are in *one* metro area, it's a lot easier to lay all the wires that connects them all to each other.

And South Korea's metro areas are a lot denser than most of America's. The Seoul metro area has a density of 16,700 people per square kilometer. There are only three cities in the US with a population over 1 million and a density over 10,000 people per square kilometer, and just 16 metro areas in the US with a density over 10,000 people per square kilometer.

New York City is of course denser, with 56,000 for the metro area and 69,000 for Manhattan itself. How is their broadband?

I'm afraid that the quality of Manhattan broadband can't be guessed from their city density. It is at the whim of their monopoly provider, right?

"How is their broadband?" wasn't a rhetorical question. How is it, really? My theory is that it should be great, but if it's not my theory has a serious problem.

Are you sure I am better at the google than you are?

I did find this:

"The Center for The Urban Future has issued a report – New Tech City – that details NYC’s rapid growth to tech industry pre-eminence in the last few years. However, the report states that that development is hampered by the poor state of Internet infrastructure in the city, citing the lack of availability, bandwidth, and competition."

I live in Manhattan, and bandwidth is pretty terrible. I pay for the 20Mbps, and often (>8x/week) can't stream video, and less often (>2x/week) resort to my phone to use things as light as Google.

If TWC and every single one of its terrible employees went bankrupt tomorrow, I would literally throw a party.

That seems a good example, fwiw. Netflix recommends "3.0 Megabits per second - Recommended for DVD quality and 5.0 Megabits per second - Recommended for HD quality" Here your "20 Mbps" line apparently isn't giving you 3 Mbps, real world.

Urban in the US has very little to do with urban in many parts of the world. To get 80%, you have to call the suburbs of Dallas urban. How far out of Madrid do you have to go to find that level of population density?

Single family homes destroy the economics of both public transportation and fiber.

Is that true? Is Google avoiding single family homes in their roll-out?

The google deployment is just brand advertising (not that this is bad for the consumers) so they don't care too much about the economics

FWIW, Akamai rates connection speeds. By country, the US is in 8th place by average speeds, although there are other measures. South Korea usually sits on top of every list.

Several states would be in the top 5 if they were treated as their own country. The District of Columbia would be second only to South Korea. (hope the link to the raw PDF works)

"How far out of Madrid do you have to go to find that level of population density?"

It's 3,300 miles to suburban Portland, so 3,300 miles. American suburbs are pretty rare in the world as they require extreme Soviet-style anti-market central planning to achieve. Markets have never chosen that kind of lot size and parking mandates.

"Single family homes destroy the economics of both public transportation and fiber."

That will astonish the majority of citizens of Seoul that live in single family homes and have the world's fastest internet. Also the 85% of Tokyoites who live in single family homes and have the world's highest rate of transit use.

"American suburbs are pretty rare in the world as they require extreme Soviet-style anti-market central planning to achieve. "

Surely you mean they resist extreme Soviet-style anti-market central planning to achieve.
The market is quite clear in Americans choosing suburbs.

A lot more immigration is going to be needed to help wire up the US. The labor just isn't there right now, we've seen tech companies complaining already about massive shortages which I'm sure exist in many other sectors as well.

Why do you say that 'the labor just isn't there right now"? Tech companies already complain about "massive shortages" but I'd find these complaints more credible if the unemployment rate were lower, or if I saw them recruiting more determinedly -- for instance, by funding college for students who commit to working at the firm -- or by providing on-the-job training. Instead, I've heard repeatedly that they're very particular and turn down older workers with all the requisite programming skills.

"Why do you say that ‘the labor just isn’t there right now”?"

I'm guessing sarcasm.

I don't think the net neutrality argument can be so easily ignored. This so far is non-exclusionary non-neutrality. That is, Netflix gets a guarantee and nobody else is hurt. I'd argue it might even be Pareto optimal.

But I'm extremely (irrationally?) tempted by the "slippery slope" argument here. It's pretty obvious that the tech world is one of copycat technologies an business models, even among the largest and most successful companies. I suspect that both the online content and ISP industries are eager to see how this deal turns out. If the economics work in favor of both firms (collateral consumer surplus aside), others are sure to follow. The nightmare scenario is twofold: non-corporate content providers get edged out, and 2) some kind of exclusionary cost disparity (like menu pricing) is passed onto end consumers. Those are scary thoughts.

I retract everything I just wrote after reading Dan Rayburn's piece.

@ Greg Werbin - I also thought Dan's piece was good and persuasive, but I am skeptical of his tone that as an insider his opinion is that this deal is just like any other deal done in the past. I don't buy that. I think there's something unique about this deal, but I also agree with TC's overall argument that the market will sort it out and society should not worry too much since there's no transaction costs that indicate a market failure.

If anyone should be peering directly with the ISPs instead of going through other backbones, you would expect it to be the company responsible for 50% of traffic.

There are other CDNs, but I can be pretty confident that their prices are natural. They enjoy no monopoly and must compete with each other. It may be that Comcast is just offering CDN on better terms, but that is pretty opaque to us. What we do know is that with their monopoly on last miles they are in a position to extract monopoly rent on last miles CDN architecture.

Basically if the libertarians would come out for free markets, or a limited term of cable company monopoly (something like 10 years from date of first service), then we wouldn't be having these hand-waving discussions about markets which are "free enough."

TCP/IP networking is vital national infrastructure, and will be for decades at a minimum (possibly centuries). We should ensure that it enjoys a competitive market, right down the the last mile.

(Rather than libertarians arguing in inverse logic, for state sponsored monopoly power in a critical section of our economy. I mean, if you want socialism, why not go for it then, with a reunited AT&T, closely monitored and regulated?)

Well, Coase said that payola is Pareto-optimal, so you're wrong. Case closed. /s

Double marginalization hurts consumers through higher prices. Tyler seems to be implying that this contract (partial integration) will reduce the double marginalization, leading to lower prices. Consumers happy.

I think some commentators (e.g. Timothy Lee, linked in post), suggest that the deal is turning the marking from a rather competitive organization to a more monopolistic organization. Big Netflix plus Big Verizon. Thus, the less competitive market leads to higher vertical externality.

So the empirical question should be: is this deal making the market less competitive. The way to answer is to look at the shape of the demand function (good luck) and the markups of each player (good luck).

You seem to be shortchanging the uninformed screed. Without that, net traffic would be reduced substantially.

Especially video screeds. But most of those are on YouTube, not Netflix (Michael Moore excepted), and therefore will be cut out of this arrangement.

Youtube (google) has spent lots of money on creating their own CDN several years ago

Timothy Lee offers a different perspective. Perhaps I am failing to understand his argument

Timothy Lee doesn't have arguments. He writes things designed to get the tech-world good and angry, regardless of how valid or invalid, and he thrives off of the resulting rage-views.

Netflix could get almost the exact same benefit by just setting up a bunch of peering sites all over the US, or by contracting with Akamai, but at its scale its better off just being its own Akamai.

high "social dominance orientation," much?

um, yeah, methinks so.....

I'm such a rube. I thought Amazon's power over shipping was a problem.

It's not just about Netflix and Comcast. It's also about the "next" Netflix. But assuming that new content providers grow organically, and Comcast doesn't throttle/isn't allowed to throttle this hypothetical smaller company, then a bad precedent isn't being set.

Where is innovation in that theory? (For me, it looks pretty dead.)

A startup has no market power, and by the time it aquires this power, it's not needed anymore, as biger companies cuold quite well supply the same thing. Now, what does the theory say it will happen between a player with lots of market power, and one without it?

Similarly, what will happen to the millions of ninche players (some not even companies) that make the net usefull today?

@Marcos--the millions must pay up or shut up. Here in the Philippines you must pay $100 a month to get 3 Mbps--too much money for too little bandwidth. Ironic since it's also the Call Center for the World. I think it has something to do with the fact it's an archipelago and with typhoons possible every month, the wireless transmission towers (I don't think they have undersea cables yet) are expensive. But if people pay enough--and if there's not a 0.1% elite of crony capitalists who stifle competition (a big 'if' here)--then the free market will eventually sort things out, like Coase and Wilkins Micawber predict.

There aren't 2 monopolies involved, only 1. Cable companies are both service providers and competitors to Netflix and derive most of their income from selling content, not transport. The expected result is that companies like Verizon and Comcast will try to raise Netflix prices to the point that maximizes long term profit.

It's also likely that cable companies have overpaid for content with their bundled model versus Netflix a la carte model if there were a competitive content market versus a monopolistic one. Netflix and Hulu plus broadcast TV thru an antenna are vastly cheaper with no loss of quality. They can't afford to let this continue while still increasing cable TV prices at twice CPI, and they can't realistically increase their internet connection pricing enough to compensate.

This rate of increase is probably understated, based on how the BLS calculates prices. A change from 100 channels at $20 to 200 channels at $30 would be a 25% decrease. This, of course, assumes that the added channels have the same utility as the old ones.

Geez, dude. That Netflix has to pay Comcast for something has already sold (bandwidth to its subscribers) is a naked demonstration of monopoly power, abuse.

Rayburn makes the claim that bandwidth is not throughput. Taken to the extreme, that would mean I could buy 1Gbps bandwidth and receive 1/1000th, just 1Mbps. Would I have recourse? If my contract with Comcast is just naming some theoretical bandwidth, perhaps not. But then we'd all need better contracts, IMO.

You're confusing last mile internet access services with backbone network services. Netflix is buying the latter likely in order to mitigate bottleneck issues at Netflix's current network services provider's interconnection point(s) with Comcast. The retail end user customer is purchasing last mile service in order to access the Internet. They're two different kinds of services.

I understand, roughly, the range of content distribution systems that can be done given the Internet architecture. It sounds to me that Comcast is actually offering a "sit close to your end-users" deal. Akamai can't offer that, because they can't sit closer than the edges of the Comcast network. Akamai certainly cannot offer a "deliver deep into the Comcast network." That's the network topology.


Netflix is NOT getting deeper than Akami. They are connecting directly to Comcast as Akami ALREADY does. This deal is good for Comcast because it gets a BW monster (Netflix) off of peering points that can now be less congested.

A link would be nice. I did read the short statement of agreement between Comcast and Netflix which was mum on the subject. I'd be looking for an engineering statement from Comcast or Netflix, rather than someone's claim that it doesn't have to be deeper than Akami & etc.

Whoops! "Comcast, Time Warner Cable, Verizon, and AT&T have all refused to hook their data centers up to Netflix's servers without payment from the video streaming service. Their logic: when they move data for content networks like Akamai, they get paid for delivering that content to end users. Netflix is doing the same thing, so they should pay up too."

So what this source says is it's even WORSE than I thought. Comcast, Time Warner Cable, Verizon, and AT&T have used their monopoly positions to extract rents from Akamai too.

Oh, and exactly as I said, "deeper." "Netflix wanted to hook up one of their Open Connect servers inside a Comcast data center and theoretically save money for both of them."

That's normal practice since decades

Finally, someone (C) gets it.

>Should we worry Amazon (Prime) buys transit rights from UPS & USPS?

That's a terrible analogy. Netflix already paid for the transit rights (and the consumer has, too), they're being extorted for additional payment.

If it turns out that this deal saves Netflix money because they can cut out a middle-man, will you change your tune?

The only issue is the local loop. There is no monopoly issue with peering. It might be a minor concern that the owner of the local loop is faster than competitors like cogent.

Tyler linked to Dan Rayburn's piece, and probably read it, but didn't understand it.

Everything Tyler wrote addresses the scenario that he (and everyone else) imagines just happened, when in fact nothing at all like it happened.

Yes. The only company really worried here is Akamai.

Possibly. Though as I say that is opaque to us. It may be that Comcast is offering a product that Akamai could offer. Or it could be that Comcast is offering a service (with last miles interconnect) that Akamai is blocked from providing. Akamai can only get "so close" to the end user.

+1. Sometimes the technical details really matter, and this is one of those times.

(OK, I replied to Charlie and it went under John, so that is suboptimal. I agree with Charlie and disagree with John.)

In what sense? That "Akamai can only get 'so close' to the end user." is just topological fact.

No, it's not. Akamai doesn't *want* to get closer to the end user, because that's a different market with different regulatory burdens. If they wanted to, and had the money, they could set up competition for internet connectivity in some town or another, then expand from there. But it's a very different business, both technically and politically, and they'd probably get a better return investing the money into Comcast or TWC shares.

That's a pretty big change of subject, Anthony. I'll I'm talking about is a deal like "position caches at Comcast branch and trunk connections." There is no regulation blocking that.

Don't keep us in suspense. Tell us what "really happened."

Dan Rayburn's piece explains it pretty well.

A short summary is: Netflix used to reach Comcast (and thus Comcast's customers) by going through third parties, which is the same way they reached everyone else. Now they will stop using those third parties to reach Comcast, and will connect to Comcast directly.

By analogy:

Imagine Amazon stopped using the USPS, FedEx, and UPS to deliver packages to, say, DisneyWorld, and instead just brought them to the Disney corporate mailroom themselves.

You ask "Should we worry Amazon (Prime) buys transit rights from UPS & USPS?"

Of course not. thats because in order to deliver packages to you or Me Amazon has a choice of supplier - USPS, DHL, fedex and UPS. NFLX doesnt have that luxury - if youre a CMCSA subscribers NFLX has no choice but to do bizness with CMCSA. And chances are if you are unhappy that CMCSA isnt treating NFLX fairly in many places your choices are minimal if you want to switch broadband providers.

There are *some* places where Comcast is the only provider, and that is a problem that should be addressed. (Artifically introducing a competitor? Demand they lease the last-mile to whomever asks? Strong regulation as a monopoly-like utility? There are many ideas here.)

But it's not the typical story. In the vast majority of Comcast's markets, it has competition.

As I've said, I have Time Warner, and my regulated option (AT&T) is only theoretical. They have not wired my end of the city for faster than 768K. So if I want multi-Mbps it is TW or no one.

That's an argument for deregulating AT&T, not for regulating Time Warner

Again, that's play deregulation. The sort of play deregulation that gave us "one cable company, one phone company, and a couple satellite companies" as a phony competitive market.

Seriously consider the possibility of technological plateau, and that wired (or optical) internet will be as good as it ever gets for last-mile networking. (There are reasons it could be, with much less packet collision on the wire than in the aether.)

So, the physical connection monopoly might have high-value 100 years from now. Should that value really just accrue to whoever buys monopoly rights today?

Sorry to be a frequent poster here today, but it really shocks me that there are "libertarians for monopoly rents."

Not me. I believe in free market competiton and distrust concentrated power, whether government or corporate. As the corporation geta larger its ability to affect government policy in its favor increases. This is dangerous to both present and future market competiton and present and future non-corrupt government.

In case you hadn't noticed, that is what the government is for. It is supposed to serve as my large institution serving as the check on the large and powerful corporations. You know something like what happened at the beginning of the 20th century.

In case you hadn’t noticed, that is what the government is for. It is supposed to serve as my large institution serving as the check on the large and powerful corporations

Yeah, I think that was in The Federalist Papers #85.

Yeah I agree that is a useful end government can be used towards, which is why I am (most of the time) not an anarchist.

I don’t see why having a cluster of mid-level intermediaries should make the market as a whole more competitive

You don't see why having multiple independent providers leads to a more competitive market than a single government enforced monopoly provider?

Who are you, and what have you done with the real Tyler Cowen?

"The cluster of mid-level intermediaries" is not about cable companies having monopolies granted by local jurisdictions. It's about the middlemen of backbone providers.

OT: Speed-reading. I score consistently 600 to 800 wpm speed reading, with 2 out of 3 to 3/3 question correct on the Staples speed reading test and 90% on the ReadingSoft test. Note college professors score 700-800+ consistently.

Take the test and see how you do: and

Is a high rate on that test a good thing or a bad thing?

I score 1,050 feet per minute on the art gallery visit door to door. Much better than those lollygaggers.

That must be why so many young Filipinas want you to father their children.

Double marginalization argument doesn't work: the carrier, like Comcast, is integrated backward into production.

The reason one should be concerned is because combat competes with Netflix through on-demand. The ability for Comcast to raise the price for Netflix and use the markup to subsidize it's own offering is alarming. The more competition friendly, & thus consumer friendly, approach would be to charge cable internet subscribers based on their bandwidth use so that the cost of consumption of a commoditized good is at an equal market rate. Of course, Comcast it's unlikely to do this as it directly raises it's asp rather than it's competitors'.

Honestly, I think this whole debate is much ado about a temporary situation. There are at least 4 different potential transmission mechanisms into the majority of American homes. Cable, phone line, cell transmission and satellite. All four medium's cost per kilobyte delivered has been dropping for decades. (Granted, not as fast for the phone line, but that's why they are quickly becoming obsolete.)

If cable company's were to become a decisive bottleneck, cell networks and satellite providers would quickly become dominant.

But if coax and then fiber-optic are the clear winners, you have only 1-2 players, by mandate. And those 1-2 player can easily drift toward higher prices together. It doesn't even have to be collusion. It's just optimum strategy for both.

With all respect, I'm not sure you understand what's happening here Tyler. Netflix is not paying some kind of "extortion" fee so Comcast would pass their traffic.

Companies like Netflix (everyone actually) buys cable capacity from someone. Every website pays network fees (usually it's included in a single fee). In this case Netflix buys capacity straight from Comcast, instead of paying someone else to deliver it to Comcast. Netflix probably saves money here. It's business as usual for companies building CDNs.

Forgot to attach link:

This link was retweeted by Adrian Cockcroft, ex Netflix Cloud Architect:

12 words.

12 words into the article that Tyler wrote that you say demonstrates that he doesn't know what's going on, and should read this blog link to find out why he's wrong/

12 words into the article, Tyler put that link,.

Yeah, comments like this make me wonder how many commenters never read any further than the headline.

UPS and USPS are both not monopolists and not downstream competitors with Amazon.

For some reason, no one is talking about the horizontal aspects of this deal, in which one provider of streaming content is getting paid by another for preferential treatment.

Please explain how are they getting preferential treatment.

Imagine the government complaining that Amazon was placing a big burden on the roads, as well as Fedex placing a big burden on the roads, so the police and traffic control and snow plows will not provide reliable service to Amazon, not do anything to get Fedex trucks moved along. USPS trucks, being government trucks, will be escorted by the police and a snow plow dispatched to clear the road in front if needed.

Unless Amazon and Fedex paid extra to the government to get the same service level as USPS trucks.

Comcast is demanding extra payments from Netflix so it will get the same quality of service Comcast will give its own on demand programming.

"Comcast is demanding extra payments from Netflix..."

No, try re-reading the original post.

Tyler, sorry to keep on repeating myself but as I've already commented on yesterdays post: you can't unbundle cable. It's just not technically feasible.

In copper (and point-to-point (but not GPON) fibre) networks you have a large cable coming into the main distribution frame in the local exchange containing multiple pairs of copper (fibre) strands. Each one of those connects to a specific end-user. When you unbundle you attach those specific copper (fibre) strands to the equipment of an alternative operator serving that specific end-user i stead of to the equipment of the incumbent. A coax cable used by cable companies is instead one big shared cable serving all end-users. There are no individual copper pairs for anyone end-user. The broadband stream in there is managed by a CMTS system which needs to control all of the cable modems connected to it to work

Ergo you can't physically unbundle cable. I'm sorry but it's just not an option.

Apologies but Joshua Ganz (addendum 2) doesn't understand the basics of how the Internet works. It is, and always has been, a two-sided market in which:
1) a consumer has to pay for a certain download speed (which can, at very different prices for the same bandwidth, be best effort or guaranteed)
2) a producer has to pay for upload bandwidth

So, it is a good thing for you to pay Netflix $8 per month to pay Comcast to force you to pay Comcast $40 per month to get the programming you bought from Netflix that Netflix already paid Comcast to carry using the money you paid Netflix?

My guess is Netflix will hike the $8 per month citing the need to pay Comcast to carry your viewing, and Comcast will hike the $40 per month they charge you, blaming the costs that Netflix is forcing on Netflix.

How many times does Comcast get to take money from your wallet?

Imagine if Comcast ran the highways. They would be charging you to have the road go to your house, charging Amazon for the road to it, then charging Amazon for actually shipping stuff on the roads, and charging you for having the things you order from Amazon delivered on the road to your house. But then Comcast would argue that Fedex is forcing Comcast to pay for the roads that Amazon ships your order to you with, so that Fedex would need to pay Comcast for the use of the roads that Amazon has paid to use and that you paid to use to deliver things to you.

mulp, take a moment and actually read the original post and links.

Netflix was paying 3rd party providers to deliver it's content to Comcast. Now, it's delivering the content directly to Comcast and paying Comcast instead of paying the 3rd party providers. There's no indication that Netflix is paying Comcast more than it was paying the 3rd party providers.

Liu was one of the 31 migrant workers elected to the National People's Congress last year. She is a foot masseuse in southeast China's Xiamen City. A native of south China's Anhui Province, Liu was a school dropout at the age of 14 and worked to support the schooling of her siblings.

Comments for this post are closed