Network Neutrality Confuses Me

Network neutrality – “treating all bits alike” – has been top-of-mind for Congress and for many geeks / lawyers / telco companies. I’m hoping some of you who are smarter / better-informed than I can help me answer this question: at a technical level, how does one define or implement network neutrality?

I approach this with both my former systems engineer and current lawyer hats on. Assume Congress decides network neutrality is a wonderful idea, and legislates to that effect. How would such legal code translate into computer code – in this case, network management policies? What would ISPs or other telecommunications companies be enjoined from doing?

I’ve considered a few possibilities, none satisfactory:

  • No discrimination by sender: ISPs could not handle packets from different senders differently. However, ISPs already do this to filter spam (preventing connections or mail from known bad actors) or to mitigate Denial of Service attacks. Both policies seem unobjectionable.
  • No discrimination by recipient: ISPs could not handle packets to different recipients differently. However, would this change spam filtering (for example, flagging or quarantining messages addressed to many users – often a sign of spam), or filtering based on parental controls?
  • No discrimination by application: ISPs could not handle packets for different applications in a different fashion. Yes, please tell providers that they must allow users to port scan the network.
  • No differential charges: ISPs could not charge 1) users or 2) senders differently based on application, content, or destination. But ISPs already offer tiered service plans (based on connection speed, storage space, services offered — try running a DNS server on your network, or upstream bandwidth) to users. Plus, some anti-spam strategies require charging senders differently, and it’s not clear how to craft a policy that lets ISPs block smurfing but prevents differential pricing. (Remember, users are senders too…)

In short, imagine you’re the IT person charged with implementing a law that mandates network neutrality. What switches do you flip? Remember that you have to keep your users and management happy by blocking spam, dealing with virus / worm traffic (including botnets, and perhaps kicking infected machines off the network), and perhaps preventing vulnerability identification with nmap or the like, at the same time you keep Congress happy by avoiding discriminatory treatment. What do you do?

I hope someone will help me with this. Comments appreciated.

8 Responses to “Network Neutrality Confuses Me”

  1. I have admittedly not been following the network neutrality issue closely, but from what I gather, all of the problems you are discussing are not really at the core net neutrality.

    As I understand it, there are many, many mid-level connections between service providers, and in general there is little or no filtering between those connections based on application. It is all about bandwidth, not content.

    Thus, as far as I can tell, spam blocking, Denial of Service (which is pings), etc. and other “end user” issues aren’t really relevant. What the ISP’s do with that will be governed by their agreements with their users.

    Instead, the implementation would be simple – each “connection” must allow all packets through to the end destination, at which point the end destination will do whatever it does.

    For example, when SBC hooks into some cable company’s network, there is no port blocking, etc. at that point. Either it lets the packets through or it doesn’t. I suppose it could throttle the bandwidth a bit – that would likely be outlawed as well.

    Thus, when we talk about content filtering via NON-net neutrality, it means that packets are blocked based on sender, not based on the actual content of the packet.

    I hope that helps clarify things a bit.

  2. Thanks Michael! This is interesting. If I’m understanding correctly, the problem is essentially one of interconnection: of behavior where two (or more) connectivity providers transfer data. Is there anything in the standard peering agreements that prevents “discriminatory” behavior? If I’m AT&T and I connect with SBC, but find that SBC is currently flooding my network with SYN TCP packets, and I thereby shut off SBC, does that violate anything?

    What interests me as well is that it seems from your analysis that net neutrality may be, at the moment, a construct of private agreements (peering, etc.) between network providers. That’s fascinating – to a degree, it means that legislated neutrality would be restoring the status quo.

    Thanks for writing!

  3. I don’t have the peering agreements, but I think that they cover what can and can’t be done, and perhaps they will change. Before the internet grew, getting your data onto another network was NOT a given – I recall when the MAE’s were first started, your packets were routed all over the country based on interconnects, and if any one was down, you were dead in the water.

    I view net neutrality as a simple problem of economics and transaction costs. Somebody has to pay for the bandwidth. Right now, SBC charges a cheap rate for 1.5MB, on the theory that every user will not fill his or her 1.5MB pipe. Thus a 10MB pipe at the Central Office can serve 20 or more people. BUT, with video and voice, the likelihood of each user filling that 1.5MB goes up, and so the provider has to install 20MB at the CO or even 30MB (making the numbers up of course).

    Who pays for this? In my view, it is the users one way or another – either they pay more for their 1.5MB pipe (because they now really are using the full 1.5MB more often) or they pay the video and voice providers, who wind up paying SBC. To me, none of this is shocking. It seems rational that it should be the service providers who pay, and then charge their customers. That way, those who really use the most bandwidth pay for it. The alternative is to charge users for what they actually use (which is what wireless companies often do).

    The scary part for people (including me) is that charging for the type of usage means all the great free stuff out there (like YouTube) goes away because people won’t pay for it. That argues for charging the end users for the bandwidth directly. Of course, if they are paying more, then there will be less users willing to pay for fluff like YouTube, so it goes either way.

    I’m not addressing blocking of “first amendment” type discrimination by service providers, but I have a belief that this is a minor part of what is going on in actuality, but it is a good rallying cry for net neutrality proponents. I think SBC, etc. do not care what the content is, so long as someone is paying for the bandwidth.

    That’s my 2 cents. I guess I should write an article on this some time because I think my views are contrary to most.

    MR

  4. Michael, I’m in agreement with you on the central point: as slack bandwidth decreases, prices will likely increase. I’d argue that there are a couple of questions here, related but not identical. First, the telcos seem to be claiming that the real key for preventing net neutrality is to allow them to recover the capital investment of building out the last mile of fiber or other physical connectivity. This is in tension with your point: if available bandwidth is dropping due to increased demand, then building out fiber, for example, is a much surer bet financially.

    Second, people may (psychologically) prefer less visible payments. We “pay” for free over-the-air television via ads. We pay for newspapers via classified advertising. Arguably, movie ticket costs are held down by the advent of pre-show “entertainment” (read: ads). FICA taxes are technically split between employers and employees; though most economists agree that employees wind up bearing the full cost of these taxes, this distribution somehow strikes us as more appealing. Hence, in terms of looking at the various ways that consumers might eventually pay for more bandwidth, some may be preferable to others for reasons that have little to do with their ultimate financial impact.

    Finally, it’s an interesting question as to how much of the new demand / bandwidth use will simply soak up unlit fiber / slack capacity, and how much will necessitate new, and more expensive, infrastructure investment. In essence, this is a question about marginal costs versus average costs, but that may be too boring even for a geeky law blog.

    And I’d love to read your article if you write it!

    /D

  5. You’re right to be confused — beyond the usual “they’ll block Google unless it pays” or “it’ll be the death of YouTube” scare stories, the net neutrality debate in DC has been notably lacking in technical details.

    But let me suggest a few ideas that I’ve picked up from some of the more level-headed advocates. First, the fear here is that major broadband network providers (i.e., The Duopoly in Your Town) will begin to arrange their business models around treating bits differently based on who is sending them. So many are asking not that all discrimination between bits be banned, but rather that any special treatment be made available to all comers on equal terms.

    Of course, this just begs more questions. For examples, if a network provider’s infrastructure configuration happens to favor Vonage packets (perhaps because Vonage designed its protocol with that particular configuration in mind), do other VOIP providers get to force the provider to reengineer/reconfigure the network?

    A second point to keep in mind: I don’t think anyone believes that network neutrality would be an absolute rule (any more than the First Amendment is an absolute rule). Presumably regulators would be asked to create a regime of exceptions that approves “non-neutral” treatment of certain kinds of traffic (like spam, port scanning, etc). Of course, that turf then will become its own battlefield, as regulators struggle to figure out whether an exception is legit or just a pretext for discrimination (or both), and then others try to sneak through the exceptions defined in the regulations. Whether you think this will be an improvement over the status quo depends on your view of regulatory decision-making.

  6. Thanks, Fred – really helpful. A few quick thoughts:

    1. Equal treatment is a strong starting point, but of course it’s vulnerable to gaming: make sure that the “fast lane” for bits is sufficiently expensive, for example, that only Google / Disney can afford it. This effectively lets the telcos engage in price discrimination (rather bluntly), which seems to be what they seek anyway.

    2. I’d be interested in whether a regulatory – the FCC? – would seek to codify existing practices (e.g., spam filtering) as exceptions to net neutrality. Not only is the devil in the details, it’s in the defaults.

    3. Your first point — about the Duopoly Problem — really seems key. We’d be less interested in network neutrality if there were more competition in Internet access. Does either side contend that their position will, in fact, widen competition or diversify access paths?

    thanks for reading!

  7. [...] By the way, correct me if I’m wrong because I haven’t double-checked, but I assume the same goes for network neutrality, trademark bills, and patent reform. [...]

  8. [...] Yoo has set himself apart as one of the most thoughtful and consistent opponents of legislated limits on service providers’ discrimination among content. He suggests that we should instead look for “network diversity” and worries that forced equality among dramatically different applications with different needs (technical and otherwise) will stifle innovation. Wu has been a supporter of mandatory network neutrality, although a careful one who warns that it will be very important to get the precise definition of neutrality requirements exactly right. (My co-blogger Derek pointed out months ago that there is a similar challenge in defining the precise approach on a technical level.) [...]