« Our MoDo Problem | Main | Ambinder on VPs »

July 28, 2008

Comments

Dog-and-pony show. Verizon and AT&T's open access requirements they agreed for their merger agreements are expiring, and they're going to hammer Comcast once they're not required to keep the peering agreements where they are.

They want Comcast punished because they can provide bandwidth w/o throttling over their fiber pipes. The ploy is to argue that "regulation is bad" and swing in the other direction (i.e. against the regulations you're talking about that preserve the peering arrangements, etc.) to allow VZ and ATT to use metered or tiered pricing.

The effect on net neutrality will be just as negative, though, for reasons I think we discussed in the last post on this topic. It's going to be a bumpy ride. :(

so cynical...why so serious? (joker reference)

look, maybe. but that could come back and bite them if Obama appoints a solid Dem who imposes some real net neutrality requirements.

but, assuming i understand you, isn't that sort of the whole logic of NN requirements? comcast has upload problems. NN in this case punishes them and gives comparative advantage to the telcos who have better bandwidth (in the new fiber offerings)

that's sort of the plan right? we want to create incentives so that, to compete, you have to improve your infrastructure. that doesn't mean we can ignore V and ATT. but the advantage they receive from having better infrastructure seems very much part of the plan

Adam, can you remind me again why charging consumers per byte without regard to the type of data being transmitted might compromise NN? You explained before but for the life of me I can't remember, and I really don't want to go digging through that monster thread...

The problem isn't necessarily that Verizon and AT&T have good infrastructure, it's that they're going to achieve a stealth monopoly (the bad kind, not the Sirius-XM kind). The Comcast thing is a power play by VZ and to a lesser extent ATT, as we've discussed before -- they've thrown the weight of their advocacy groups behind the effort to punish Comcast, because they're trying to eliminate a competitor. So that's problem 1.

Problem 2 is net neutrality proper. Net neutrality is threatened in two ways right now -- throttling and metering. (I admit that we may disagree on the latter.)

Both are probably going to require some packet inspection, and both will put a drag on peer-to-peer architectures and centralize control of the network, which is contrary to the end-to-end design that's made the Internet so useful. So there's a collateral threat to net-neutrality principles to the extent that net neutrality demands that intelligence reside in the edges of the network, which I think is the case. (Pretty sure Zittrain discusses this, FWIW. I know Lessig does.)

A shift away from pipe-based infrastructures imposes a structural tariff on peer-to-peer (which privileges big firms that can afford per-byte pricing), just like throttling. That's additionally beneficial to VZ and ATT because they're aiming for triple-play services -- and the voice and video elements of the triple-play are seriously threatened by P2P apps like Skype and BitTorrent.

I also think that there are oversight issues with metering and with VZ's domination of the major backbones, insofar as it will probably disrupt the peering arrangements that I see as integral to net neutrality. But that problem may be a little too abstruse to discuss here.

Hopefully that post answers the question, Turb. To be brief, I think that metering threatens NN by (1) possible DPI to monitor BW usage, esp. in the presence of compression; (2) centralizing of network decisions via the drag on P2P and other shared architectures; (3) the potential danger to peering arrangements when other ISPs are inevitably asked to pay for their transit.

As to the infrastructure question specifically, publius, I am doubtful that the monopolized infrastructure really works that well -- I think the 96 Act has objectively failed to encourage competition, and in fact we're pretty much headed down the Bell monopoly road again. No one is going to overbuild VZ's network -- it might be OK now, but the problem is that there's already hardly any competition in the markets, despite FTA96, and it's only going to get worse.

Moreover, the current regulation assumes a "siloed" world -- internet, cable, spectrum, voice, etc. The current legislation isn't designed to accommodate the single-network model, and it's going to fail. VZ and ATT are going to end up with control over the entire telecom infrastructure and none of the existing regulation is going to stop them from squeezing the rest of the market dry.

i'll say more tomorrow (rough baby night last night so i'm about to go to bed), but adam's interpretation is consistent with the Kevin Martin conspiracy theorists who think he's a Bell hack.

I actually the 700 MHz stuff disproves that. i buy the feld theory that martin actually believes in making markets work and is concerned by a lack of competition. i mean, he's a partisan republican, but could be worse.

also, moving from the tech to legal world, there's a huge issue about the authority to act on neutrality/openness. Even if Vz and T vanquish Big Cable (that's a bold claim -- not saying i disagree, but need to think about), the act against Comcast is an enormous step in establishing the legal authority to act.

it's not merely that they're acting, but that they're doing so with bipartisan support. that adds some political weight too.

and even if the FCC does nothing, and Congress has to step in, this step still really helps in a legislative fight. "You see, Senator, even Kevin Martin was concerned enough that he did X."

Thsi thread reminds me that i need to go back over the other one -- i tagged it but haven't read through it completely yet.

(I admit that we may disagree on the latter.)

Heh indeed.

Both are probably going to require some packet inspection

If the telco is only interested in counting bytes, no packet inspection is required. Byte counts are trivial and are already being done.

(Pretty sure Zittrain discusses this, FWIW. I know Lessig does.)

Can you give any more specific info as to where? I'm not challenging you, I just want to go read some without having to read through hundreds of pages. If you can't think of any more specific references, no worries.

A shift away from pipe-based infrastructures imposes a structural tariff on peer-to-peer (which privileges big firms that can afford per-byte pricing), just like throttling.

This part confuses me; if you're a company purchasing connectivity from a datacenter, then you're paying for per byte. If you're really really tiny, you might get away with flat rate up to a cap, but that's just a convenience since all transmission beyond the cap is metered. What am I missing? Or are you saying that when Verizon/ATT have killed off Comcast, they're going to raise backbone prices so high that only large businesses will be able to afford it? If so, can you explain in more detail because that really doesn't make sense to me (how many dollars you bring in per byte transferred has little to do with how big your business is).

That's additionally beneficial to VZ and ATT because they're aiming for triple-play services -- and the voice and video elements of the triple-play are seriously threatened by P2P apps like Skype and BitTorrent.

The concern that telecos might use technological means to crimp VOIP competitors seems extremely serious to me (given how much of this we've already observed). What confuses me is that I think metered pricing won't do squat to hurt VOIP; the average person blows through a lot more bandwidth watching crappy youtube videos than they do chatting with Skype.

Also, I can't help but think that the correct manner to deal with this problem is through serious enforcement of anti-trust law rather than arcane frobbing of NN in a way that doesn't make sense to lots of NN proponents (i.e., the vast army of people typified by yours truly). But I'm no purist and if the only good way to get openness is by using NN to strangle metered pricing, I'll be happy to fight for that.

I'm not sure I see how BT threatens the video service being offered by telecos. BT does not deliver on demand: it takes at least an hour to download a movie on most connections while the telecos can stream immediately. BT can make anything available while the telecos seem limited to much much smaller catalogs for which they've managed to clear the rights. From the telecos' perspective, I don't think BT is their problem. Rather, their problem is getting rightsholders to acquiesce to downloading video. BT video can't (generally) provide the liability guarantee that the telecos can provide so these aren't really competing services in my mind.

If you meant something like Netflix's or iTunes' or Blockbuster's streaming video service, I think your point would be much stronger here. But those services (right now) are download only and don't require significant upload capacity, which might frobulate your argument.

I also think that there are oversight issues with metering and with VZ's domination of the major backbones, insofar as it will probably disrupt the peering arrangements that I see as integral to net neutrality. But that problem may be a little too abstruse to discuss here.

I'd love to see you flesh this out; I suspect it will require a pretty thorough explanation of backbone bandwidth economics today.

i'll say more tomorrow (rough baby night last night so i'm about to go to bed), but adam's interpretation is consistent with the Kevin Martin conspiracy theorists who think he's a Bell hack.

I actually the 700 MHz stuff disproves that. i buy the feld theory that martin actually believes in making markets work and is concerned by a lack of competition. i mean, he's a partisan republican, but could be worse.

Well, based on secondhand accounts (never met Martin but know quite a few people who have), he kind of is a Bell hack, but not in the sense you mean, I think. He's a hack in that he doesn't really understand how competition works in an interlinked world -- he sees everything through the lens of the phone network.

Because of that, the way he enshrines "competition" isn't always a positive thing. I absolutely agree that his perspective is good in the context of, e.g., the 700 Mhz auction. It hasn't been all that great in the voice world -- FTA96 really hasn't been all that great at producing competitive markets. But lots of people like FTA96 and it's not surprising that Martin's a fan. (And it probably is better than Bell at any rate, though personally I don't see that as a great standard for comparison.)

But I definitely don't like how it's shaping up in the Internet world, because he's really being taken for a ride; VZ and ATT are playing him against their competitors while they leverage their dominance in one market (voice) into the others (video and internet).

It's not a conspiracy, he just doesn't really get it and that's creating problems because VZ and ATT most definitely do get it and they're running circles around him.

Ooh, I really want to respond to Turb's comment, but I need to sleep too. Curse you guys.

I did think of a historical example/argument that I'll paste (I typed this up but it wouldn't post, so I'm not staying up to write this... just wanted to throw it out there):

Metered pricing is essentially how the phone system has always worked -- pay-per-minute -- and we can all see pretty clearly how that worked out. It gave us Bell, and then after FTA96 precisely no competitive local exchange carriers succeeded. We're now basically back to AT&T.

The internet, on the other hand, used pipeline-based pricing, which is essentially what created the successful infrastructure that publius discusses in this post. That historical accident (metered pricing wasn't really feasible, information services were insulated from the incumbents, etc.) gave us various distributed technologies (e.g. P2P) that wouldn't have been feasible in a metered system.

Pipelines also provided the structural basis for the high-level peering arrangements that have proved remarkably resilient in maintaining a neutral Internet, despite competing Tier 1 ISPs -- something that really wasn't feasible in other areas of telecom where the infrastructure encouraged monopolies and price-fixing. If the Internet had been run like broadcast, cable, or voice, we never would have had a neutral net in the first place.

Unfortunately, I think that's where we're headed right now. And more specifics on that tomorrow.

Ooh, I really want to respond to Turb's comment, but I need to sleep too.

Excuses excuses. You tease.

Metered pricing is essentially how the phone system has always worked -- pay-per-minute -- and we can all see pretty clearly how that worked out. It gave us Bell, and then after FTA96 precisely no competitive local exchange carriers succeeded. We're now basically back to AT&T.

This sounds...not believable. If you want to argue that metered pricing was the thing that made Bell suck, then you're going to have to explain a mechanism by which it did so, because just asserting it isn't really doing much for me. I mean, the DC metro charges you based on the distance you travel while the Boston subway is flat fee, but I don't think you could just assert that that is why the MBTA sucks compared to the WMTA.

Metered pricing didn't give us Bell; Bell came about for a whole host of complex causes.

Pipelines also provided the structural basis for the high-level peering arrangements that have proved remarkably resilient in maintaining a neutral Internet, despite competing Tier 1 ISPs -- something that really wasn't feasible in other areas of telecom where the infrastructure encouraged monopolies and price-fixing.

I've read this again and again and I still don't know what it means. Pricing per physical connection is real when it comes to an individual customer but I don't see what this has to do with peering. When two tier 1 providers (call them A and B) hook up at MAE East, the individual customer pipelines are far far away and pretty close to irrelevant. What matters there is how many bytes go from A to B and how many bytes go from B to A every hour. You can be damn sure that both A and B keep very very close tabs on this data and if the amount of traffic going in one direction starts to dramatically exceed the amount going in the other direction, action is taken. There are no tier 1 providers stupid enough to allow infinite bandwidth to one of their peers in exchange for nothing.

"Actually, the Internet is successful because the government regulated the bejesus out of it."

This link seems to lend no substantive support to the claim whatever. It's just a handful of paragraphs of assertion.

It's not that I disagree with you, but could you perhaps find a link that actually substantiates your claim beyond simply asserting it? With some detail? There's essentially no relevant information there whatever, other than a completely unsubstantiated repetition of your claim; it seems rather a random, as well as utterly uninformative, choice. What are the relevant regulations? When were they passed? What did they do? In other words, what are you talking about?

It's useful for links to provide such information; simple repetition is not so useful.

Here might be one starting place. Here is a piece by Lessig. Perhaps this paper might help. This is also substantive.

Given your expertise, perhaps you might educate us about the relevant details, since you believe they are relevant?

"Actually, the Internet is successful because the government regulated the bejesus out of it."

I'm not totally sure what you are trying to say with the link.

The main conceptual areas of regulation was intitally treating it all under common-carrier type rules (i.e. no or little discrimination on the basis of content) and some of the domain name issues. On the other hand, actual delivery systems have come by all sorts of means and in those areas where government intervention was to be had, it typically was to enforce monopolies which retarded the growth of high speed internet penetration rather than enhanced it (the early cable modem years).

So I'm not sure what area you think represents "regulated the bejesus out of it" for the purposes of this discussion.

Turb, sorry for the delay. I've been thinking about your comments today (which raise all the right questions, I think) and was hoping to type up a proper response when I finish work.

As to Sebastian and Gary, I think publius is correct in that the common-carrier regulations are arcane and draconian in the way they insulate voice from "information" services.

publius' link points to how the different Titles in the Federal Telecommunications Act enforce divisions that aren't really based on any substantive technical distinctions. The internet was "unregulated" because of the efforts of legacy providers to keep the new kids off their turf.

Can you give any more specific info as to where? I'm not challenging you, I just want to go read some without having to read through hundreds of pages.

Lessig, Lawrence. The Future of Ideas, pp. 26 (history of AT&T), 34 (end-to-end). (download here) (bottom of page)

Wikipedia's network neutrality entry that Gary linked is also surprisingly good. For the sake of reference, I agree with Google's definition of net neutrality here, which defines it in terms of discrimination "against competing applications or content."

This part confuses me; if you're a company purchasing connectivity from a datacenter, then you're paying for per byte.

If you're paying for managed hosting, maybe. Personally, I've never dealt with anything but pipes when dealing with colocation.

can you explain in more detail because that really doesn't make sense to me (how many dollars you bring in per byte transferred has little to do with how big your business is).

I disagree -- pipes are burstable and it becomes the provider's responsibility to mediate between the number of clients they have versus their capacity (e.g. you can have multiple clients that don't use all their BW allocation by muxing the pipes, assuming you have enough backbone capacity). That flexibility is important for different applications, I think -- more on that in the discussion of the parallels between the phone and data infrastructures.

What confuses me is that I think metered pricing won't do squat to hurt VOIP; the average person blows through a lot more bandwidth watching crappy youtube videos than they do chatting with Skype.

OK, this I strongly disagree with. Skype relies on supernodes and a distributed architecture that moves a couple kilobytes a second through the network, much like BitTorrent. That works in a world where users pay for connections, but not where they pay for bytes. Skype's decentralized design couldn't survive without that.

Decentralized networks simply won't survive in a world of per-byte pricing. Skype and BitTorrent (to take two important examples) rely on being free, and the background traffic on both networks is integral to their design. If that traffic's no longer free (the cost of constant background traffic adds up under per-byte pricing), you're looking at far fewer users, which means far fewer nodes, which means significantly degraded service. Since Skype competes directly with telcos, and BitTorrent with content distribution networks (and there are lots of legitimate uses of BT -- even Blizzard uses BitTorrent to distribute Warcraft patches and such), that raises a serious competition issue.

Also, I can't help but think that the correct manner to deal with this problem is through serious enforcement of anti-trust law rather than arcane frobbing of NN in a way that doesn't make sense to lots of NN proponents (i.e., the vast army of people typified by yours truly). But I'm no purist and if the only good way to get openness is by using NN to strangle metered pricing, I'll be happy to fight for that.

I don't necessarily disagree, but if antitrust laws worked, Skype wouldn't have to write their client like a virus (fascinating PDF here) and pull all kinds of legal tricks to avoid being smashed by large telcos.

The telecom laws are the problem here -- a notable example is the spurious distinction between "telecommunications" and "information" services highlighted in publius' initial link. The fact is that ATT and VZ won't let Skype or any other VoIP services terminate calls on the public network, and no one's been able to get around that. None of the competitive local exchange carriers (CLECs) that FTA96 envisioned have succeeded, because the legal game is rigged.

If you want to argue that metered pricing was the thing that made Bell suck, then you're going to have to explain a mechanism by which it did so, because just asserting it isn't really doing much for me.

There's a decent discussion in that Lessig book that I suggest you read, but the short version is that Bell used metered pricing to cross-subsidize various services and crowd out competition -- e.g. by pricing local service low and long-distance high because no one else wanted to build a long-distance network when Bell already had one. This is essentially the position VZ will be in because of the lead they have with FiOS.

Metered pricing didn't give us Bell; Bell came about for a whole host of complex causes.

Bell came about because they were allowed to leverage their protected markets against unprotected markets, and differential pricing was one of the main tools they used to accomplish that. I suppose there's a way to do metered pricing fairly, but I can rattle off a number of things VZ's done that suggest to me they won't (aside from the structural problems I discussed above with P2P), and no matter what there's no way to do it more fairly than just letting people pay for their pipes.

Since VZ and ATT's fiber networks have the bandwidth to accommodate all current applications and then some, I'm curious why they keep pushing per-byte pricing. The "efficiency" argument doesn't wash, since they're already building the networks. If the demand is there, it shouldn't make any difference whether you use connection- or byte-based pricing -- it's still the same backbone connection. The only rational reason I can think of is to depress low-priority traffic like P2P networks by structural means -- since it's obvious that consumers have issues with traffic-shaping, they're just trying a different route.

And regardless, I've never gone wrong betting on the venality of big telcos, so I'll err on the side of caution on this one in any event. :) Call me paranoid, but my experience has been that no matter what tricky scheme I come up with, the thing that actually ends up happening is even worse. ATT and VZ in particular are scary good at what they do.

adam - the skype per-byte stuff seems like a critically important point. and it's not one I've seen in academic literature, though maybe I missed it. i haven't thought of it like that, but that's a key key point. any extended discussions/links/articles on that specific point?

also,
1 - antitrust won't work. i actually just wrote a paper with a section on this. among other things, it's like implicitly precluded by extensive telecom regulation (i.e., antitrust courts won't mess around in fields that are heavily regulated).

2 - Adam, i'm not sure your "creation of Bell" point is quite right. but maybe I'm misunderstanding you. from what I understand, it was more that they switched from a policy of non-interconnection to a "buyoff competitors" strategy. essentially, it was an early version of the anbar awakening.

there's an old Mueller book on this ("Universal Service"), though it's out of print. but the history is impressive (one of his points is that cross-subsidizing long distance didn't have all that much to do with Bell's success)

Adam, thanks for explaining. You've given me a lot to think about. Unfortunately, I need to get to bed pretty soon, so the reading will have to wait till tomorrow.

publius, the paper sounds very interesting...is there any way we can get enough info on it to track it down?

If you're paying for managed hosting, maybe. Personally, I've never dealt with anything but pipes when dealing with colocation.

Wha? Are you saying that your colo contracts allow you unlimited bandwidth? The only ones I've seen have a cap for total monthly bandwidth and usually a max burst speed. I can imagine a colo facility selling rack space without mentioning bandwidth limits while imposing a hard cap at very small number of bytes per second, but that's very different. If you have unlimited bandwidth at a colo, please hook me up with the provider.

can you explain in more detail because that really doesn't make sense to me (how many dollars you bring in per byte transferred has little to do with how big your business is).

I disagree -- pipes are burstable and it becomes the provider's responsibility to mediate between the number of clients they have versus their capacity (e.g. you can have multiple clients that don't use all their BW allocation by muxing the pipes, assuming you have enough backbone capacity). That flexibility is important for different applications, I think -- more on that in the discussion of the parallels between the phone and data infrastructures.

Now I'm confused. Here's my argument. Amazon moves very few bits for each dollar in profit they earn because they're just schlepping around web pages. In contrast, Netflix's streaming video service moves many many bits for each dollar in profit. Some random home-based ebay business is more like Amazon whereas a solo pro photographer or videographer is closer to Netflix. In all these cases, the variable that matters is how much profit they earn relative to how many bits they exchange; the size of the business in and of itself doesn't explain much about how the business will react to changes in bandwidth pricing.

I don't get why you're talking about muxing and flexibility; all that stuff is basic packet switching 101.

OK, this I strongly disagree with. Skype relies on supernodes and a distributed architecture that moves a couple kilobytes a second through the network, much like BitTorrent. That works in a world where users pay for connections, but not where they pay for bytes. Skype's decentralized design couldn't survive without that.

Are you talking about Skype's bandwidth needs for individual users or Skype's corporate needs for its central servers? For individual users, a couple of kilobytes per second is nothing. People blow way more than that uploading photos to flickr.

I think we're getting stuck on how the price changes in a world where end users are billed per-byte. I'm assuming that the price won't differ significantly beyond the current wholesale cost of bandwidth, say $0.50 per gigabyte. Double that if you like. I suppose VZ/ATT could raise prices dramatically right now, but if they want to do that, can't they do that for their wholesale customers right now anyway? If their ability to price the backbone is the real source of their power, then what's stopping them from utilizing it right now? Raising prices on backbone providers seems a lot less likely to trigger political problems than raising prices for your customers directly. Rising prices on the backbone will be absorbed by customers indirectly.

Decentralized networks simply won't survive in a world of per-byte pricing. Skype and BitTorrent (to take two important examples) rely on being free, and the background traffic on both networks is integral to their design. If that traffic's no longer free (the cost of constant background traffic adds up under per-byte pricing), you're looking at far fewer users, which means far fewer nodes, which means significantly degraded service. Since Skype competes directly with telcos, and BitTorrent with content distribution networks (and there are lots of legitimate uses of BT -- even Blizzard uses BitTorrent to distribute Warcraft patches and such), that raises a serious competition issue.

It all depends on what the per-GB price is. At $0.50/GB, downloading a single episode of TV costs me $0.10. Boo hoo. There's a limited amount of TV I can watch. At this rate, I blow a couple of dollars a month in the worst case. Even if we're talking about shipping Linux distributions around, I end up paying $2.00 maybe every time there's a new Ubuntu release in the worst case. I think most people can handle the $4.00 per year that would cost. In contrast, purchasing the same show from iTunes will cost me $0.63.

BT is cool and all, but it is not something I'd kill over. My informal sampling indicates that 95% of BT use is for illegal transfers. BT has two advantages: lower bandwidth costs for seeders and greatly reduced liability footprint. But given that even hard core BT users would only burn $10 a month in bandwidth costs if they had to download directly from a central server, it seems the only significant benefit BT has is reduced liability. I'm not sure I'm willing to stake major regulatory decisions on the need to make particular technologies primarily used for piracy economically viable. I mean, I think IP laws in the US are well beyond braindead and I think the RIAA/MPAA are insane criminals, but BT isn't the hill I'd want to die on for NN if you get my drift.

You still haven't really addressed the questions I asked earlier about whether BT can in any sense be said to compete with telco's video offerings. There's the time delay feature, the liability angle, and have you seen how limited VZ's Starz Play's catalog is?

"Bell used metered pricing to cross-subsidize various services and crowd out competition -- e.g. by pricing local service low and long-distance high because no one else wanted to build a long-distance network when Bell already had one."

That is wrong. The idea of using long distance service to subsidize local service originated during WWII. The concern was that if the FCC lowered long distance prices, people would want to make more long distance calls, and the network wouldn't be able to handle the additional traffic because we weren't investing in domestic infrastructure during WWII. So the FCC lowered local rates instead. After the war, the subsidies were increased in the name of promoting universal service.

Rather than crowding out competition, the subsidization played a major role in opening up the long distance market to competition. MCI built its own long distance network, and started offering long distance service for considerably less than AT&T. The way it worked was that an MCI customer would make a local call to an MCI office, then dial the number that the customer actually wanted to reach. The call would would be carried over MCI wires to another MCI office, which would connect to the target telephone via a second local call. This was significantly less expensive than using AT&T because the MCI customer wasn't paying any money to subsidize local service.

The FCC probably had the power to stop MCI from doing this, but it was blindsided. The FCC simply didn't understand what MCI was doing until after MCI had done it. Once the MCI network was in place, there wasn't much that the FCC could do about it. Political considerations meant that the subsidies for local service had to continue, and the FCC had no legal authority to force MCI to help pay for those subsidies.

The eventual result was that AT&T was broken up to allow long distance companies to compete on a level playing field. All long distance companies were given equal connections to the local network (no more having to dial extra digits if you were an MCI customer), and all long distance companies had to pay "access fees" in order to subsidize local service. The breakup was arranged as a way to settle an anti-trust suit against AT&T, but I doubt it would have happened without the subsidization of local service, which allowed MCI (and later Sprint) to effectively compete in the long distance market even without the breakup.

Read about the OSI networking model.

Each layer of the model is supposed to be completely agnostic about the layers above it. IP doesn't care if it is routing a TCP packet or a UDP packet.

Comcast violated that model with deep packet inspection, where they took the packet, inspected it at the application layer to see if it was a torrent looking packet, and would refuse to route that packet, violating the terms of the OSI model, in effect controlling what applications you can run on your computer by interrupting traffic from applications it doesn't like.

publius asks for cites, and I'll try to oblige, but I think this was just one of those things that seems so obvious to me that I never bothered to see if there were studies. However, if you're interested, here is an interview with a big AT&T muckety-muck explaining how they plan to use usage-based pricing to reduce P2P traffic that clogs up the network.

He basically lays it all on the table, before he screws up and says that there's no real stress on their networks anyway since most BT activity is at off-peak hours -- oops! so he changes the subject quick since he just admitted they wanted to structurally throttle P2P, and he doesn't want Skype to hit him with an antitrust suit.

publius, how would preclusion work if D is allegedly using two unrelated regulations (e.g. changing the pricing for internet access for P who are using that internet access for VoIP that competes with D's phone lines) -- i.e., the regulations are being leveraged against each other.

w/r/t publius and Kenneth, I was imprecise when I said that metered pricing made Bell or whatever it was. What I meant was that the pricing scheme (whether it was tiered or not -- it was always usage-based) was always one of the big hammers used on the competition before cross-subsidizing was formalized after the way. But price tricks could also include driving out competitors with loss leaders, market segmentation, extracting monopoly rents, etc.

Yes, interconnection is the big trick, of course, but usage-based pricing is what gives it bite. In voice, there's a huge issue with who pays for where the call terminates or originates or who carrier what and do you cut in the upstream and blah blah...

But in the internet world it's all peering agreements and best-efforts routing and the system works astoundingly well (a few threats to depeer Cogent notwithstanding) because people aren't providing "bytes," they're providing "connectivity." I am arguing that pricing on the basis of content has no inherent advantages over connection-based pricing. The issues outlined elsewhere aside,

I think that what this comes down to is a strange new obsession with content rather than transit. A neutral internet provides equal transit and ignores content completely. It treats all apps and data streams alike.

If we're worried about capacity, the solution is to build more infrastructure, not try to game what isn't working by differentiating on price of content (Verizon / AT&T) or types of content (Comcast) but on simple old upstream and downstream.

Finally, Kenneth, you're talking about the cross-subsidization that occurs under the aegis of "universal service" policies. I'm talking about competitive cross-subsidization, which Bell used from the start, often simply to push out competition via loss leaders. It wasn't until later that they began cross-subsidizing urban v. rural areas at the behest of government officials, which is what you're discussing.

Your discussion below demonstrates why you're getting this wrong.

The way it worked was that an MCI customer would make a local call to an MCI office, then dial the number that the customer actually wanted to reach. The call would would be carried over MCI wires to another MCI office, which would connect to the target telephone via a second local call. This was significantly less expensive than using AT&T because the MCI customer wasn't paying any money to subsidize local service.

Wrong and wrong. Is this some sort of mistake? I certainly hope so. MCI didn't carry anything "over MCI wires" -- as indicated by the name of the company, Microwave Communications Inc. -- they used long-distance microwaves to circumvent the high-cost long-distance lines that ATT was leveraging (i.e. using to cross-subsidize) against local phone companies.

Note that this predatory cross-subsidization had nothing to do with subsidizing local service. In fact, you're getting "local" and "rural" confused -- the latter was what was targeted by universal service programs, not local phone service. The way Bell made money was by extracting monopoly rents from every other carrier using their long distance infrastructure.

The comments to this entry are closed.

Blog powered by Typepad