Response To: Opening the Internet - With an Axe

After reading this article arguing against regulation of the Internet vis-a-vis net-neutrality, I felt compelled to compose a reply to the author. After spending so much time on it, I decided to spend a little more and make it a web page, so as to be more readily accessible to others. Since writing this page, I've found additional information that may help readers, and so have added that.

I was browsing today and read your article with some interest, as it pertains to a field with which I am intimately familiar and involved - I have been using the Internet in one capacity or another since about 1991, have been employed at a top 50 web site, an e-commerce hosting company, a Usenet service provider, an Internet-based financial services company, and so on.

Regarding the "three perils" to the Internet, I have the following comments:

  1. You mention that Genachowski claims the choice of providers has narrowed substantially with the shift from dial-up to broadband. In a strict sense, this isn't true; when most people were using dial-up, there was only one connection to their home - the "local loop" from the telephone company's central office to their home. You had to pay for that connection in order to dial anyone. However, the phone company did not treat calls to anyone in the LATA differently. This enabled you to access quite a few local ISPs relatively cheaply, and by and large, they were functionally equivalent; latency and bandwidth were approximately the same. Now, in actuality, the phone companies screamed bloody murder about modems "tying up" the phone lines; a typical modem connection was much longer than a typical phone call. However, despite their protests and attempts to squelch this "unfairness", they did not (or were not able to) take action. As a result, the Internet boom of the mid-90s was free to proceed unhindered.

    By contrast, in current day, most home users have two lines to their home; co-ax (cable) and telco local loop (DSL). In some cases, there are other options; satellite, for example, which has high latency due to speed-of-light delays (and often requires a dial-up line for the uplink), and in some cases, wireless access via Wi-Max and other technologies. However, these choices are not available to many consumers, as they require line-of-sight to geosynchronous satellites 22,300 miles over the equator, or to a tower or relay used by the wireless ISP. Also, there are substantial security concerns, since the traffic is transmitted through free space, and interceptable to anyone with the proper equipment. A surprising amount of Internet traffic uses unencrypted protocols.

    Despite this, the Internet that you access through these providers is relatively "flat"; one may connect to any Internet site, no matter where it is located, with limitations imposed by fundamental limits such as speed-of-light, and (more importantly) peering arrangements, long-haul and backbone infrastructure, current loads, etc. The ISPs would like to change this. It's akin to FedEx or UPS saying that they are going to delay or silently fail to deliver packages to a particular business, industry, or other destination, at their discretion.

    In response, you state that even with one operator, the customer is still "sovereign"; that is, he has the right to choose not to use the Internet. You suggest that this means that a provider who doesn't provide access to all sites will be economically pressured to do so, but the mechanism for this seems unclear, especially if providing access to particular sites that compete with their own offerings would cannibalize their revenue streams. Faced with the choice of being able to access some Internet sites or none at all, I wager few would choose none; indeed, in heavily censored countries, people still pay for Internet access, rather than have none at all, and there's little financial pressure on those countries to stop censoring the Internet, as evidenced by the fact that they still do so.

    You state later in the paper that a citizen may choose to pay for restricted access to the Internet versus having none at all. That is likely very true, but it seems to dodge the question of whether a consumer would prefer to have his options restricted to those two, compared to a regime where unrestricted access to the Internet was an option.

    I've been thinking about this a bit over the last decade or so, and I don't think I have a perfect grasp of it yet, but essentially, when offering a large and complex package, the provider may tinker with the details, and as long as the package they offer is still better than the competition (in your example, nothing), they have no incentive to improve the service on that particular detail. This may explain why an especially affluent area - be it a city, state, or country - may be able to enact fairly draconian laws in the form of taxation, or other encroachments on personal liberty and property - and yet, the overall package (being able to participate in that economic market) remains sufficiently desirable that people willingly move or stay there.

    NOTE: What I mean by this is that when services only come bundled, the customer is often forced to accept less-than-optimal solutions as part of the bundle. In this case, "access to the Internet" is is the bundle, and just because you want that doesn't mean that you don't want to be able to access all the sites on the Internet. To suggest that because people buy the bundle means they like it better than open, unfettered access (when that is not even an option available to them) seems to be a common fallacy for free-market advocates. I'm sympathetic to free-market principles, but find these kinds of arguments mendacious.

  2. In your second example, you state that a provider would lose revenue by blocking access to particular services. But it's not clear exactly what mechanism this would take. Indeed, if there are limited options to a consumer - in the extreme case, the only option is not to have any Internet access at all - it's not clear that they would leave in droves, especially if any other options available to them have similar policies. For example, in my locality, I have the choice of two broadband operators (or none at all, I suppose). One of them, has collaborated with the NSA to illegally eavesdrop on US citizens. The other is known for using DPI (mentioned later) to interfere with bittorrent sessions by forging TCP RST packets from the remote endpoints. Prior to their discovery, very few consumers knew about this - one of the basic tenents of the efficiency of a free market is perfect information about a provider's offering. In this case, it is clear that consumers did not have anything resembling perfect information of the service being offered. Now that I have more information about the service, I'm still faced with the rather unappealing choice of the lesser of two evils.

    I think this has some grounding in the issue of property ownership; in effect, the physical connection to the house is owned by one company. And as you must know, land is unlike most other forms of property in that it is not portable; if I, by some bizarre situation, own a private, unfiltered internet connection to my residence, and I move, I do not automatically own that same connection to my new residence. I believe that there are several restrictions on how land may be used in most countries due to this reason; for example, one may not purchase all the land around an area and prohibit the residents there from leaving or entering. In this case, the property in question is the wiring to the customer's home, and the FCC's role could be construed in a similar light. Only in this case, it is not the resident who is leaving or entering, but his information/communication. If we put your counterarguments in similar terms - that a person surrounded by another's property is still sovereign and may choose not to pay for transit through it, I think it perhaps more clear that it is an undesirable state of affairs for that resident - transit could be denied entirely, making his home a jail.

  3. The third threat he mentions refers to traffic management techniques. The latest generation of these techniques is DPI.

    You state that a customer doesn't care about how the goals of service provision are achieved, just the end product. You make the good point that a consumer (usually) doesn't care how a car is manufactured, only the end product.

    And yet, DPI and related "traffic management" strategies deeply affect the end product and its value to the consumer, but they have far from perfect knowledge of the service being offered. Typically these traffic management strategies are deployed without any disclosure to the consumer.

    They may be used to:

    The problem with many of these is that the policy is not transparent to the consumer, subject to change without notice, and potentially very difficult to detect - notably, there is no direct indication that the provider is dropping packets or forging TCP RST segments. Even if the providers were required by law to describe their practices, as some do with privacy policies, the language is usually very opaque, complex, and generalized. Typically the implications of such policies are not readily understood by a consumer.

    Nor is it the case that DPI is even necessary for traffic management - RED works quite well.

    The fact that less expensive and less intrusive "traffic management" strategies which accomplish the overtly stated goals - like RED - are being ignored implies that the motivation is not what is being stated publicly. In fact, it appears that technologies like RED are too fair to meet the covert goals; namely, the ability to deny or degrade access to competitors and competing technologies, or ones that they find morally or ideologically offensive.

Since the advent of broadband, the ISPs are increasingly coming under the control of media conglomerates - in my former residence, the cable "plant" was owned by Time-Warner. See Fairness and Accuracy in Reporting, specifically The Global Media Giants. Concurrent with this, the Internet has become an alternative source of information. The media companies are not unaware of this; see the Youtube video capture entitled FOX Attacks Bloggers. In a particularly absurd note, the FOX talking head declares that "the problem with blogs is that the communication is one way". This was on broadcast television (the most one-way of communications media). By controlling the means of distribution, the mass media companies hope to regain some degree of control over information on the Internet.

Telcos have their own problems; they don't wish to carry any data at all (that would be work), but be able to charge consumers whatever rate they wish. The Internet has seriously undermined their attempts to control offerings over their networks, and they'd like to return to not having to compete with others.

Most broadband ISPs are, at this point, over-subscribed; that is, the amount of bandwidth available to the customers in aggregate far exceeds their transit capacity to other networks. And they'd like it to stay that way, because inter-network traffic means paying money to other companies, whereas intra-network does not. So by making local offerings viable by denying access to competitor's offerings, they reduce the amount of traffic that leaves their network.

In addition to this, both the telcos and media companies see open access to competitor's services (e.g. VoIP, skype, Hulu, Youtube, etc.) as a threat to their own service. By making their competitor's services less desirable, or even completely denying access to them, they are able to offer their own, potentially inferior service offering at a price that would not be viable under more open conditions. It's hard to see how consumers would benefit from local monopolies being able to reduce the value of competitor's offerings. In the case of high-bandwidth data like streaming video, reducing the bandwidth seriously degrades the value of that service.

One may argue that a consumer could establish an encrypted VPN connection to escape such a traffic management regime, and indeed that is what some companies have done by offering (e.g.) the VyprVPN service to customers.

However, there are problems:

In the case of interactive moderately bandwidth-intensive real-time communication such as VoIP, latency and jitter significantly impact the value of the offering to consumers. Technologies such as VPN necessarily introduce more latency and jitter, both due to the routing topology differences and the processing time used for encryption.

The net result of all this is the mess we see here; providers are massively over-subscribed, consumers have incredible amount of bandwidth to their home (though still not nearly as much as some European countries such as Sweden), and providers are under economic pressure to decrease the fraction of the available bandwidth that consumers use, by putting in bandwidth caps, discriminating against heavy Internet users, and selectively discriminating against protocols like bittorrent that they now see as "unfair".

I wrote all this, and now I read that the concept of "perfect information" is criticized by heterodox economics (to include the Austrian and post-Keynsian schools). Ah well, it was a fun exercise. I could go on and list the other ways this market strays from perfect competition, and will stray further with tiered services, but I think it should be pretty obvious, plus I'd be lecturing you on your field of expertise, rather than mine...

PS: There is a neat tool by Google for end users to detect "bandwidth shaping" and "traffic management" by their ISP.

Go to this level's index
Travis's Homepage [email protected]
Original date: