April 23, 2024

New Net Neutrality Paper

I just released a new paper on net neutrality, called Nuts and Bolts of Network Neutrality. It’s based on several of my earlier blog posts, with some new material.


  1. realizePhiladelphia says

    Check out some more information about Net Neutrality at http://web.illish.us

  2. Direct Real Estate Investment

  3. Translation to spanish of Felten’s paper on “Nuts and Bolts of Network Neutrality” (“Dificultades e imprevistos de la regulación legal de la neutralidad en Internet”).

    En el blog http://gruizlegal.blogspot.com está colgada mi traducción al español de trabajo de Ed Felten como “Dificultades e imprevistos de la regulación legal de la neutralidad en Internet”

  4. Hi, it seems to me that your paper addresses network neutrality as an attribute that exists solely due to technical limitation and therefore is assumed to be expendable. I would argue that neutrality is the defining characteristic of the network system.

    I suspect that because neutrality was not a purposeful attribute but rather a side-effect of the way the system evolved it is not recognized for its benefits to democracy.

    “Mild” or “harsh” discrimination of packets is a matter of degree but either would be a fundamental change to the system — would it not? It seems to me if innovation gives precedence to the possible efficiencies over the importance of preserving the benefits of the content-neutral system you will lose the main benefits of the system — but perhaps this is where the debate lies. Do you believe the benefits of the current system whereby all packets are treated equally are less important than improving the efficiency of the system?

  5. MnZ: Given a choice between the two incentives (people have to pay extra for decent service, vs. competition between ISPs) I’d choose the latter. What we need is to make internet service provision more competitive. Wideband wireless is probably the way to make it happen, or else a nationalization of the infrastructure and its leasing back to connectivity providers, other service providers, and other private actors.

    I’d rather not the latter, though — it’s liable to instantly start being massively overhauled with some sort of surveillance architecture. Turning it into what’s been described recently on this blog as “dangerous infrastructure”.

  6. Nick: This sort of article on a hot network-engineering topic is a prime candidate for a slashdotting, so that’s what I’m guessing may have happened.

  7. Ed,

    I really enjoyed your article. I found it very thought provoking. However, I don’t know if I understand your QoS argument. Aren’t QoS guarantees already being sold to businesses? I realize that these are on private networks, not the public internet. However, it does indicate that there is a demand for QoS in some segments of potential users.

    Moreover, many consumers (including me) and small businesses are keeping POTS, because they do not want to risk VoIP without QoS guarantees. In a way, POTS is a QoS guarantee.

    The rest of your article seems to actually lend support to paying for QoS guarantees. You say on page 9 that an ISP might have reasonable and nondiscriminatory rules in place, but still have network jitter. However, what is their incentive to fix network jitter in such cases? Certainly, competition between ISPs would give them the incentive to fix jitter. However, if there is no competition between ISPs (as some have argued), then QoS guarantees may well be the only incentive that the ISP would have to fix the problem.

  8. Eric: Nope, the website was clearly down, as was cs.princeton.edu. I was coming from I2, and obviously tried to go to the website first. Seems to be up now. Probably was fixed by 11am the next morning, but it was down for quite some time.

  9. @Hal, let’s first answer the question: Can there be a highly competative ISP market?

    For hosting and hosted services there is a highly competative market; everyone with an internet connection and a few servers can start such a service. Providing internet connectivity is more of a problem, because that requires an infrastructure. Telco’s and cable companies can add internet services to their network with a pretty low investment, whereas wireing an entire town with glassfiber gets quite expensive. (Wireless is another option.) The current state is that there are a few wired broadband providers (cable&telco) and there’s satellite; Wimax will enter the mix in some areas.
    That is a competative market, but not highly; the handfull of players can take good looks at the other’s offerings and calculate a nice margin for itself. Things will become more competative when telco’s and cable co’s are forced to unbundle their subscriber lines.

    So the economic motivation is relevant and will remain relevant for the coming years. Ed’s technical analysis remains of great value, including his observations that technically sound measures can be used to inconvenience competitors.

  10. % ping itpolicy.princeton.edu
    PING itpolicy.cs.princeton.edu ( 56(84) bytes of data.

    this is what I have been getting and I am not able to visit to that side. What should I do?

  11. The elephant in the room here is the role of economic motivations. You mention it in passing a few times but never give it the central position it deserves.

    Wouldn’t all this analysis go out the window if there were a highly competitive market suppling internet connections, so that each person could choose from multiple providers? I think so. Any efforts by ISPs to limit their customers would fail just like if Coke decided to make their drink taste like piss to save money.

  12. >Nick, I get the same thing, but I can still visit the website; clearly >itpolicy.princeton.edu just doesn’t respond to pings.

    Or you’re an SBC customer, which just started filtering icmp traffice in our area a month ago…

  13. Ugh!! Unmarked link to a .pdf file!

  14. Anonymous says

    Nick, I get the same thing, but I can still visit the website; clearly itpolicy.princeton.edu just doesn’t respond to pings.

  15. % ping itpolicy.princeton.edu
    PING itpolicy.cs.princeton.edu ( 56(84) bytes of data.

    — itpolicy.cs.princeton.edu ping statistics —
    14 packets transmitted, 0 received, 100% packet loss, time 13033ms

  16. Prof. Felten,

    How much of the earlier end-to-end neutral network might be explained by limitations in routers and network hardware? It seems that machine capacity increases more rapidly than network capacity, leading to ever more available cycles per packet.