Telcos are Not Suicidal

by on January 27, 2007 · 12 comments

I’ve finished reading Bill Herman’s paper. It’s got a lot of interesting material in it, so I’ll be discussing it in several posts over the next few days.

Having finished the paper, I remain convinced that Herman hasn’t given much thought to the details of how the discriminatory pricing regime he envisions would actually work. He seems to imagine that AT&T can simply send Google a bill for ten million dollars and Google will whip out its checkbook and pay it. This, it seems to me, is highly improbable. To see why, let’s look at the other end of the market—the millions of tiny websites like this one that are only frequented by a few hundred people every day.

The remarkably low cost of Internet production puts it within the reach of most Americans, skill permitting. A personal Web site costs less than four dollars per month to host; in many cases, Web hosting space is included with a broadband subscription. Millions of people, including people who otherwise produce and distribute no media, host Web sites for fun, for self-expression, or for some higher social purpose. Much of it is for vanity and amusement, but tens of thousands of gifted artists, seasoned experts, and enthusiastic hobbyists post irrefutably valuable content. Including both the fun and the serious reasons to love the Internet, these millions of hours of unpaid labor add incalculable value to our economy, not to mention our enjoyment of life and our democracy. In many endeavors, the dream of nearly frictionless transactions has been leapfrogged by the reality of nearly costless transactions and an entire subeconomy of “peer production and sharing.”

Adding even a small amount to the cost of these millions of nonmarket actors’ participation would cause more deadweight losses. Slowing or blocking their Web sites would likewise diminish their willingness to devote their time and energy to building the value of the network for no compensation, again piling up deadweight losses. Yoo supports the attempt by BSPs to capture all of the marginal value of increases in the worth of their networks, but a great deal of this value is produced for no direct economic gain. If BSPs attempt to capture the positive value of things that are not being bought or sold, they will kill the goose that laid the golden egg–or at least seriously reduce her golden egg production.

Now, Herman is absolutely right that these millions of sites are absolutely essential to the success and vibrancy of the Internet. Almost everyone who uses the web visits at least one of these sites, and many us visit dozens on a daily basis. I personally would cancel any Internet service that didn’t allow me to reach my favorite blogs, most of which are in the small-scale, non-commercial category Herman describes here.

Moreover, Herman is correct that most of these sites wouldn’t pay tolls to multiple third-party ISPs so their customers could reach them. I’m pretty sure we at TLF wouldn’t. We’re very happy we have a non-trivial number of readers around the country, but we do this blog as a hobby and none of us is rich. Moreover, a lot of our readers read the blog at work, and the market for higher-speed business connectivity tends to be rather more competitive than the home market, so we’d figure most of our readers could get us at work.

OK, so Herman and I agree that an ISP attempting to charge millions of website owners for the ISP’s customers to reach them would dramatically lower the benefit of Internet access for those users while generating very little revenue from ISPs (since most of them wouldn’t pay, and the ones who do pay aren’t going to pay very much). But it seems to me that for precisely those reasons, an ISP would be crazy to attempt such a stunt. Keep in mind that even a monopolist has some constraint on the price he can charge for a service, and that price is directly proportional to the perceived benefit of the service provided. Instituting policies that dramatically reduce the perceived value of a company’s services will reduce that company’s revenues, even if that company has a monopoly.

So I just don’t understand what Herman thinks is going to happen. It seems totally nuts to imagine AT&T sending Adam Thierer a bill saying “send us $10/month or we’ll cut off access to your site for our customers.” Adam wouldn’t pay it, but he would write a long, angry post denouncing AT&T, and then I imagine AT&T would get some angry phone calls from any TLF readers who happen to be AT&T customers.

  • http://tieguy.org/blog/ Luis Villa

    What is scary is not what they’d do to entrenched sources of information like blogs but what happens to the Next Youtube or Next BitTorrent or next Jabber- something that most people don’t yet know that they love (and hence won’t protest when it goes AWOL), but which catches the attention of the telcos because they are bandwidth hungry or run on non-standard ports and protocols. They’d happily put an innovation tax on new services that way, and the telcos are about the last people I want deciding what new services get implemented.

  • http://tieguy.org/blog/ Luis Villa

    What is scary is not what they’d do to entrenched sources of information like blogs but what happens to the Next Youtube or Next BitTorrent or next Jabber- something that most people don’t yet know that they love (and hence won’t protest when it goes AWOL), but which catches the attention of the telcos because they are bandwidth hungry or run on non-standard ports and protocols. They’d happily put an innovation tax on new services that way, and the telcos are about the last people I want deciding what new services get implemented.

  • http://www.techliberation.com/ Tim Lee

    But Luis, how would they do that? There are literally thousands of obscure applications using a lot of bandwidth and/or running on weird ports. Would they block everything except for an approved list of standard applications? Wouldn’t that create needless collateral damage on a lot of video games and obscure business applications? What happens when they “accidentally” block an application that’s more popular than they expect, or that happens to be used by someone influential? And how would the billing work? Presumably, you’d want to figure out who wrote the application in question and send the bill to him, but it won’t necessarily be easy to figure out who developed a given application. And even if you can locate the application developer and send him a bill, Herman’s point about non-commercial websites applies just as well to software products: a lot of them are going to be non-commercial sites that simply won’t pay.

    Moreover, if AT&T started blocking everything it didn’t recognize, they’d immediately get a serious evasion problem. Application developers would just start disguising their non-approved applications as the traffic of an approved application. You already see this with port-blocking schemes–a lot of applications that are frequently port-blocked have an option to run all their traffic on port 80. AT&T’s innovation tax would certainly annoying for users, but I don’t think it would actually make AT&T any money, because the small amount of revenues they would be able to extract from up-and-coming commercial application developers would be outweighed by the drop in revenues from customers who are pissed off that their favorite application mysteriously stopped working on AT&T’s network, despite the fact that it works fine at work and at their neighbor’s house.

    I think the fundamental problem for a potential “innovation taxer” is that the online world has a massive long tail–both in terms of the various applications, and in terms of the content available within each application. You can’t chop off a long tail without dramatically reducing the value of the service (which you’re charging your customers for), but it would be prohibitively expensive to individually inspect, bill, and filter each of the millions of different activities people are using their Internet access for. I think this argument applies equally well to the long tail of web content and to the long tail of non-web applications.

  • http://www.techliberation.com/ Tim Lee

    But Luis, how would they do that? There are literally thousands of obscure applications using a lot of bandwidth and/or running on weird ports. Would they block everything except for an approved list of standard applications? Wouldn’t that create needless collateral damage on a lot of video games and obscure business applications? What happens when they “accidentally” block an application that’s more popular than they expect, or that happens to be used by someone influential? And how would the billing work? Presumably, you’d want to figure out who wrote the application in question and send the bill to him, but it won’t necessarily be easy to figure out who developed a given application. And even if you can locate the application developer and send him a bill, Herman’s point about non-commercial websites applies just as well to software products: a lot of them are going to be non-commercial sites that simply won’t pay.

    Moreover, if AT&T; started blocking everything it didn’t recognize, they’d immediately get a serious evasion problem. Application developers would just start disguising their non-approved applications as the traffic of an approved application. You already see this with port-blocking schemes–a lot of applications that are frequently port-blocked have an option to run all their traffic on port 80. AT&T;’s innovation tax would certainly annoying for users, but I don’t think it would actually make AT&T; any money, because the small amount of revenues they would be able to extract from up-and-coming commercial application developers would be outweighed by the drop in revenues from customers who are pissed off that their favorite application mysteriously stopped working on AT&T;’s network, despite the fact that it works fine at work and at their neighbor’s house.

    I think the fundamental problem for a potential “innovation taxer” is that the online world has a massive long tail–both in terms of the various applications, and in terms of the content available within each application. You can’t chop off a long tail without dramatically reducing the value of the service (which you’re charging your customers for), but it would be prohibitively expensive to individually inspect, bill, and filter each of the millions of different activities people are using their Internet access for. I think this argument applies equally well to the long tail of web content and to the long tail of non-web applications.

  • http://tieguy.org/blog/ Luis Villa

    Would they block everything except for an approved list of standard applications?

    Why not? That is what they do on the cell network, and what many businesses do on their internal networks.

    And how would the billing work?

    Same way first-class works- airlines don’t find customers and ask ‘would you like to be in first class?’ You go to them and say ‘I’d like to pay for first class tickets- how much are they?’

    What happens when they “accidentally” block an application that’s more popular than they expect, or that happens to be used by someone influential?

    Nothing, just like when it happens on the cell network. Customers bitch and moan, and have no meaningful alternative service provider, and so nothing happens. (We should probably drop this one, though, since we already know we disagree on whether or not there is meaningful competition in the data space.)

    [I]f AT&T started blocking everything it didn’t recognize, they’d immediately get a serious evasion problem.

    There is a big DRM evasion problem too, but that hasn’t stopped people from trying to apply DRM, nor has it helped businesses to innovate in the space (re: your ‘no innovation in DVD storage’ post.) If AT&T blocked bittorrent, people would certainly try to get around the blocks, and some might even succeed, but you can bet bittorrent.com would never have gotten a dime of VC. That is damaging to innovation- tons of non-commercial innovation would go into working around the new restrictions, instead of actual constructive work, and no commercial innovation would occur.

    You can’t chop off a long tail without dramatically reducing the value of the service

    Even in a long tail situation, the vast bulk of the value is still in the head- voice in the cell network; email/nytimes.com/espn.com in the data network. As long as that head is untouched, long-tail services (by their nature splintered and with small constituencies) don’t have enough of a voice to actually protest and effect change. Again, bittorrent- if AT&T sniffs and blocks that when it has its first serious public demo (http://www.ntk.net/2002/07/26/ ), who will protest it? AT&T really has this level of granularity in their traffic monitoring; it is quite impressive. They’ve implemented it in the name of network security, but it could trivially be used to sniff and slow new protocols/services. The readership of ntk will never be an effective protest group.

    it would be prohibitively expensive to individually inspect, bill, and filter each of the millions of different activities people are using their Internet access for

    Again, the cell network already does a rough form of this (try putting Skype on your phone), and AT&T already inspects every single packet that crosses their IP network- you seem to underestimate how much of the infrastructure is already in place, and overestimate how fine-grained it would need to be to be effective from AT&T’s POV.

  • http://tieguy.org/blog/ Luis Villa

    Would they block everything except for an approved list of standard applications?

    Why not? That is what they do on the cell network, and what many businesses do on their internal networks.

    And how would the billing work?

    Same way first-class works- airlines don’t find customers and ask ‘would you like to be in first class?’ You go to them and say ‘I’d like to pay for first class tickets- how much are they?’

    What happens when they “accidentally” block an application that’s more popular than they expect, or that happens to be used by someone influential?

    Nothing, just like when it happens on the cell network. Customers bitch and moan, and have no meaningful alternative service provider, and so nothing happens. (We should probably drop this one, though, since we already know we disagree on whether or not there is meaningful competition in the data space.)

    [I]f AT&T; started blocking everything it didn’t recognize, they’d immediately get a serious evasion problem.

    There is a big DRM evasion problem too, but that hasn’t stopped people from trying to apply DRM, nor has it helped businesses to innovate in the space (re: your ‘no innovation in DVD storage’ post.) If AT&T; blocked bittorrent, people would certainly try to get around the blocks, and some might even succeed, but you can bet bittorrent.com would never have gotten a dime of VC. That is damaging to innovation- tons of non-commercial innovation would go into working around the new restrictions, instead of actual constructive work, and no commercial innovation would occur.

    You can’t chop off a long tail without dramatically reducing the value of the service

    Even in a long tail situation, the vast bulk of the value is still in the head- voice in the cell network; email/nytimes.com/espn.com in the data network. As long as that head is untouched, long-tail services (by their nature splintered and with small constituencies) don’t have enough of a voice to actually protest and effect change. Again, bittorrent- if AT&T; sniffs and blocks that when it has its first serious public demo (http://www.ntk.net/2002/07/26/ ), who will protest it? AT&T; really has this level of granularity in their traffic monitoring; it is quite impressive. They’ve implemented it in the name of network security, but it could trivially be used to sniff and slow new protocols/services. The readership of ntk will never be an effective protest group.

    it would be prohibitively expensive to individually inspect, bill, and filter each of the millions of different activities people are using their Internet access for

    Again, the cell network already does a rough form of this (try putting Skype on your phone), and AT&T; already inspects every single packet that crosses their IP network- you seem to underestimate how much of the infrastructure is already in place, and overestimate how fine-grained it would need to be to be effective from AT&T;’s POV.

  • http://tieguy.org/blog/ Luis Villa

    (I should note that I agree with Prof. Felten that crafting non-sucky net neutrality legislation would be difficult-to-impossible, but pretending that there is no potential problem, and hence removing the threat of legislation- the biggest factor in keeping the telcos honest- altogether, is insane.)

  • http://tieguy.org/blog/ Luis Villa

    (I should note that I agree with Prof. Felten that crafting non-sucky net neutrality legislation would be difficult-to-impossible, but pretending that there is no potential problem, and hence removing the threat of legislation- the biggest factor in keeping the telcos honest- altogether, is insane.)

  • http://www.techliberation.com/ Tim Lee

    Luis,

    I don’t think we disagree all that much about this. If your point is that filtering would be a disaster for the telcos, but that some of them might be incompetent enough to try it anyway, I can’t really disagree with that. The question then becomes how much of a disaster would it be, and how long would it take for them to realize that they’d shot themselves in the foot. A couple of further thoughts:

    With respect to the cell network analogy, I think you underestimate the forces of inertia involved. People don’t expect to run Skype on their cell phones, and so they’re annoyed, but not especially angry, when they’re not able to do so. In contrast, there are several million people who use Skype on their computers. They will be a lot more angry if it suddenly stops working than they would be if it had never worked in the first place.

    Second, people react very strongly when they feel a company is treating them unfairly. If you cut off an application that accounts for, say, 10 percent of a user’s traffic, he’s likely to feel that the value of his service has been reduced by a lot more than 10 percent. This is especially true since a lot of niche applications have very strong followings who care a great deal about being able to use those applications. SSH is a good example of this–block that and you’ll have every single one of your Unix admin customers rioting in the streets. Likewise, a significant number of people play a variety of networked games. Cut off WoW and you’ll have a million angry customers (and voters).

    And lord help you if you cut off a category of service whose users are photogenic. Imagine if you block some kind of children’s distance learning program or any kind of medical software. Suddenly AT&T is standing in the way of Johnny learning to read and Grandma treating her cancer. Ted Koppel would have a field day.

    Same way first-class works- airlines don’t find customers and ask ‘would you like to be in first class?’ You go to them and say ‘I’d like to pay for first class tickets- how much are they?’

    But the difference, again, is that our hypothetical telco isn’t building a new network from scratch. It already has millions of existing customers who rely on hundreds of different applications. The question is how do you induce Google to come to the telco and say “we’d like to pay more for the privilege of our website reaching your customers.” My guess is that you can’t.

    As long as that head is untouched, long-tail services (by their nature splintered and with small constituencies) don’t have enough of a voice to actually protest and effect change. Again, bittorrent- if AT&T sniffs and blocks that when it has its first serious public demo (http://www.ntk.net/2002/07/26/ ), who will protest it? AT&T really has this level of granularity in their traffic monitoring; it is quite impressive. They’ve implemented it in the name of network security, but it could trivially be used to sniff and slow new protocols/services. The readership of ntk will never be an effective protest group.

    I see two problems with this. First, if most of the value is in the head, then you’re not going to generate much revenue if you grandfather the head into your blocking scheme free of charge.

    Secondly, the reason AT&T’s traffic monitoring tools are able to distinguish the various protocols so cleanly is that the people who designed them had no particular incentive to obfuscate them. If AT&T started to block traffic it didn’t like, you’d quickly see libraries to disguise arbitrary data transactions as web traffic, or World of Warcraft traffic, or Skype traffic, or whatever. And you’d obviously have to block encrypted protocols, since those could be hiding disfavored data transmissions.

    Should we totally remove the threat of regulation? Well, I don’t think there’s much chance of that happening in any event. There will always be a non-trivial number of people who favor NN regulations, and perhaps that’s a good thing. So while I’m more worried about bad regulation passing now than about necessary regulation failing to pass in the future, I do think there’s something to be said for Felten’s “wait and see” position.

  • http://www.techliberation.com/ Tim Lee

    Luis,

    I don’t think we disagree all that much about this. If your point is that filtering would be a disaster for the telcos, but that some of them might be incompetent enough to try it anyway, I can’t really disagree with that. The question then becomes how much of a disaster would it be, and how long would it take for them to realize that they’d shot themselves in the foot. A couple of further thoughts:

    With respect to the cell network analogy, I think you underestimate the forces of inertia involved. People don’t expect to run Skype on their cell phones, and so they’re annoyed, but not especially angry, when they’re not able to do so. In contrast, there are several million people who use Skype on their computers. They will be a lot more angry if it suddenly stops working than they would be if it had never worked in the first place.

    Second, people react very strongly when they feel a company is treating them unfairly. If you cut off an application that accounts for, say, 10 percent of a user’s traffic, he’s likely to feel that the value of his service has been reduced by a lot more than 10 percent. This is especially true since a lot of niche applications have very strong followings who care a great deal about being able to use those applications. SSH is a good example of this–block that and you’ll have every single one of your Unix admin customers rioting in the streets. Likewise, a significant number of people play a variety of networked games. Cut off WoW and you’ll have a million angry customers (and voters).

    And lord help you if you cut off a category of service whose users are photogenic. Imagine if you block some kind of children’s distance learning program or any kind of medical software. Suddenly AT&T; is standing in the way of Johnny learning to read and Grandma treating her cancer. Ted Koppel would have a field day.

    Same way first-class works- airlines don’t find customers and ask ‘would you like to be in first class?’ You go to them and say ‘I’d like to pay for first class tickets- how much are they?’

    But the difference, again, is that our hypothetical telco isn’t building a new network from scratch. It already has millions of existing customers who rely on hundreds of different applications. The question is how do you induce Google to come to the telco and say “we’d like to pay more for the privilege of our website reaching your customers.” My guess is that you can’t.

    As long as that head is untouched, long-tail services (by their nature splintered and with small constituencies) don’t have enough of a voice to actually protest and effect change. Again, bittorrent- if AT&T; sniffs and blocks that when it has its first serious public demo (http://www.ntk.net/2002/07/26/ ), who will protest it? AT&T; really has this level of granularity in their traffic monitoring; it is quite impressive. They’ve implemented it in the name of network security, but it could trivially be used to sniff and slow new protocols/services. The readership of ntk will never be an effective protest group.

    I see two problems with this. First, if most of the value is in the head, then you’re not going to generate much revenue if you grandfather the head into your blocking scheme free of charge.

    Secondly, the reason AT&T;’s traffic monitoring tools are able to distinguish the various protocols so cleanly is that the people who designed them had no particular incentive to obfuscate them. If AT&T; started to block traffic it didn’t like, you’d quickly see libraries to disguise arbitrary data transactions as web traffic, or World of Warcraft traffic, or Skype traffic, or whatever. And you’d obviously have to block encrypted protocols, since those could be hiding disfavored data transmissions.

    Should we totally remove the threat of regulation? Well, I don’t think there’s much chance of that happening in any event. There will always be a non-trivial number of people who favor NN regulations, and perhaps that’s a good thing. So while I’m more worried about bad regulation passing now than about necessary regulation failing to pass in the future, I do think there’s something to be said for Felten’s “wait and see” position.

  • http://bennett.com/blog Richard Bennett

    Luis is playing the what-if game. What if AT&T decided not to exchange packets with networks outside the US because it costs too much? Hey, that would be bad, so let’s pass a law to make them exchange packets world-wide with not restriction, especially to Nigeria where the generous dictators all seem to live.

    Do we need to do that?

    No, because the telcos aren’t suicidal. It would have been perfectly legal for cablecos to do all of Luis’ what-if scenarios at any time in the past, but they never have.

    Why didn’t they do that? Because cablecos aren’t suicidal.

  • http://bennett.com/blog Richard Bennett

    Luis is playing the what-if game. What if AT&T; decided not to exchange packets with networks outside the US because it costs too much? Hey, that would be bad, so let’s pass a law to make them exchange packets world-wide with not restriction, especially to Nigeria where the generous dictators all seem to live.

    Do we need to do that?

    No, because the telcos aren’t suicidal. It would have been perfectly legal for cablecos to do all of Luis’ what-if scenarios at any time in the past, but they never have.

    Why didn’t they do that? Because cablecos aren’t suicidal.

Previous post:

Next post: