The Rise of the Self-Contradictory Network

Re-reading my Berkman Center slanderer David Isenberg’s seminal paper The Rise of the Stupid Network, I was struck by the contradictory nature of the two paragraphs at the heart of the polemic. First, he says his “Stupid Network” is aware of the types of messages presented to it, and handles each with appropriate service: [In] … Continue reading “The Rise of the Self-Contradictory Network”

Re-reading my Berkman Center slanderer David Isenberg’s seminal paper The Rise of the Stupid Network, I was struck by the contradictory nature of the two paragraphs at the heart of the polemic.

First, he says his “Stupid Network” is aware of the types of messages presented to it, and handles each with appropriate service:

[In] the Stupid Network, because the data is the boss, it can tell the network, in real time, what kind of service it needs. And the Stupid Network would have a small repertoire of idiot-savant behaviors to treat different data types appropriately. If the data identified itself as financial data, the Stupid Network would deliver it accurately, no matter how many milliseconds of delay the error checking would take. If the data were two-way voice or video, the Stupid Network would provide low delay, even at the price of an occasional flipped bit. If the data were entertainment audio or video, the Stupid Network would provide wider bandwidth, but would not necessarily give low delay or absolute accuracy. And if there were a need for unique transmission characteristics, the data would tell the Stupid Network in more detail how to treat it, and the Stupid Network would do what it was told.

Yet this is the very behavior that Net Neutrality laws would ban carriers from embedding in their fee agreements, and moreover it contradicts the very next paragraph:

You would not have to ask your Stupid Network provider for any special network modifications – its only function would be to, “Deliver the Bits, Stupid.”

Network neutrality advocates say true neutrality is simply delivering the bits, first-come, first-served. But delivering the bits in ways that are sensitive to application needs is blasphemy, monopolistic, price-gouging and extortion. See Susan Crawford for an example of the “bits is bits” point of view:

There are lots of people out there saying “we need to treat all VoIP alike, all video alike, and all blogs alike.” For them, that’s network neutrality. That’s not what I hope we’ll end up meaning by net neutrality. That would require a heavy-handed regulator enforcing a provider’s determination of what packets are “like” other packets. I am not in favor of that approach. I have a different vision. I hope, someday, we’ll treat broadband access like the utility it is. That would mean separating transport from other activities, and separating access from backbone and backhaul transport. That doesn’t require a great deal of discretion to repose in any particular actor.

Yesterday’s debate at PDF seemed to be focused on the fuzzier definition of network neutrality (“treat all VoIP alike”). That definition plays directly into the arguments of the telcos. It would give the FCC an enormous amount of discretion and power.

(UPDATE) In a subsequent re-write of this article, Isenberg came over to the Crawford side and abandoned the special treatment idea:

Intelligent Network advocates point out that networks need to treat different data types differently. Right now, they’re absolutely correct. There is a network for telephony, another network for TV, and proprietary leased-line networks for financial transactions – and none of these are ideal for public Internet traffic. You need to have low delay for voice telephony, the ability to handle megabit data streams with ease for TV, and low error rates and strong security for financial transactions.

Quality of Service (QOS) is an intermediate step in the journey from separate networks to a single, simple Stupid Network…

But suppose technology improves so much that the worst QOS is perfectly fine for all kinds of traffic, without a repertoire of different data handling techniques. Suppose, for example, that everyday normal latency becomes low enough to support voice telephony, while at the same time allowing enough capacity for video, plus data integrity strong enough for financial transactions. This would be a true Stupid Network – one treatment for all kinds of traffic.

Why should anybody build a network to transport raw bits without packet inspection? None that Isenberg can see:

One thing about the Stupid Network is clear – the physical elements that comprise the network would be neither expensive nor scarce. There would be little profit margin in shipping dumb bits. There would be lots of high value Business Ideas supported by the Stupid Network, above and beyond transport.

As I read that, he’s justifying the Telco program to pay for the network by selling services. That’s a “high value Business Idea” instead of a low-profit transport business.

And indeed, Isenberg has come to recognize that nobody will build a high-speed, stupid network simply to carry bits, as there would be no money in it:

The best network is the hardest one to make money running.

So this realization ultimately leads to the real end-goal of network neutrality: broadband Internet access networks should not be built by private companies, they should be built by government and maintained as public utilities. The goal of network neutrality legislation, then, should be to discourage private investment in broadband networks, the quicker to energize local governments to jump into the networking business.

The end of Evslin’s talk was all about doing that in Santa Barbara, CA, where Doc Searls is on the case.

36 thoughts on “The Rise of the Self-Contradictory Network”

  1. > Network neutrality advocates say true neutrality is simply delivering
    > the bits, first-come, first-served. But delivering the bits in ways that
    > are sensitive to application needs is blasphemy, monopolistic, price-
    > gouging and extortion.

    It is _not_ blasphemy, monopolistic and price-gouging extortion if, once QoS is instituted, service remains, as you point out earlier in your blogs, at a comparable level as it is now, and higher priority services get their packets delivered at a different schedule than lower ones.

    What particulary annoys me, having read about 5 or so of your journal entries that someone linked to (from a non-net neutrality site) is your obnoxious, conceited style of delivering what everyone already knows.

  2. Actually it is. Re-read the post, there’s an example of one highly-respected neut, ICANN board member Susan Crawford, calling for a ban on application-sensitive forwarding.

    I don’t have to make this stuff up, these people are that crazy.

  3. You are dead right in seeing the inherent contradiction in the neutr-elites thinkng. It is not driven by principle, because a principle would guide them to consistent thinking and logic. The neutr-elites are simply driven by opposing the telco view. That’s why their positions so often get so twisted and convoluted.

    In short, the neutr-elites don’t know what they truly stand for or what they want, but they are darn sure that if the telcos want it, it must be bad and they should oppose it. That’s the best explanation I have for their double standards and sloppy thinking.

  4. Richard, I am not certain I see the contradictions. David Isenberg and Susan Crawford appear to be supporting the two competing conceptions of net neutrality.

    Crawford appears to be in favor of “pure” net neutrality, that is, a world where routers treat all data packets on a first-come, first served basis. However, the downside to mandating this protocol (IP)–to which Isenberg presumably gave significant weight when writing the piece–is that the protocol does not handle time-sensitive applications well.

    To accommodate time-sensitive applications, Isenberg seemed to support the “fuzzier” or softer conception of net neutrality, where QoS is permitted but not QoS fees. In his vision, the routers would identify, classify and route data packets–presumably using a standard classification system and routing algorythm–according to needs of the data. Realizing this version of net neutrality–and not the “pure” version–is what was attempted by the S/D and Markey amendments.

    Crawford sees a weakness in this version: agreeing on (or mandating) a standard classification system and routing algorythm. She sees this as something that would require FCC oversight and presumably fears that agency capture could lead to ways to give preference to the telcos’ and cable companies’ Internet content and applications.

    The only source of confusion is that Isenberg’s position seems to have evolved over time. It was 1997 when he wote “The Rise of the Stupid Network,” and over time he seems to prefer “pure” net neutrality and possibly an “open access” regime that forbids network owners from integrating into Internet content services and applications. But he can speak for himself on where he stands now…

    My point is that it seems inaccurate to call the concept of net neutrality “self-contradictory” when conflicting statements by advocates result from two competing conceptions.

  5. Dignan makes five points, none of which seem compelling:

    (1) He says, “Congress will screw it [net neutrality regulation] up.” Maybe, maybe not. If they do, they might be able to fix it. That’s the nature of legislation. Congress rewrites the tax code every few years, but only a few say we should abandon taxes. After all, they provide a benefit, even if the details are sometimes suboptimal.

    (2) He says, “Fast lanes exist today.” My understanding is that is true for some VPNs but not the public Internet. His example–Akamai–is not a fast lane but a collection of server farms that reduce the hops required for data packets to travel. Akamai has no impact on routing protocol and–contrary to Dignan’s suggestion–would not be outlawed by the proposed net neutrality regulation.

    (3) He says, “All traffic isn’t created equal.” Conceded, but NN regulation allows (in theory at least) QoS so important or time-sensitive data packets get priority.

    (4) He says, “Telecom giants are already doomed.” He cites future broadband platforms, but they are not here and the telecoms are likely to control some of them. He then says that without NN, Google could charge the telecoms just like ESPN charges cable. However, he doesn’t explain how that market dynamic has anything to do NN. If demand for Google on the Internet was akin to demand for ESPN on cable, why couldn’t Google charge AT&T with NN legislation?

    (5) He says, “Laws exist to thwart net neutrality concerns today.” Arguably, the FCC could prevent anticompetitive blocking under its August 2005 Policy Statement, but the strength of that power and the FCC’s jurisdiction under Title I is a debated issue among scholars. As for antitrust, Trinko and the “agreement” requirement for collusion make the antitrust laws of doubtful effectiveness.

    He ends with the “wait-and-see” argument. But why wait for the problem to get bad? Moreover, the problem–lack of innovation (and competition to some extent) at the edges as the ISPs squeeze out rival content providers–will be difficult to quantify. We won’t be able to identify what never was. The issue probaby has to be resolved through a best guess method a priori–one way or the other.

  6. Remember, Chad, that QoS that doesn’t cost anything is QoS that doesn’t exist.

    If there’s no additional fee, nothing stops people from marking their every tiniest packet as highest-priority, and there’s an incentive to do so… and when everything is market highest-priority, the effect is identical to nothing being so-marked.

    (If same-day airmail and bulk third class cost the same, would anyone ever use anything but same-day (assuming no restrictions on what could be airmailed, etc)?

    If so, why would they?)

  7. Sivigald, that’s what I meant by “in theory.” The language explicitly permits prioritization, but the economics might not make it viable as a practical matter.

    However, there might be ways to prevent everyone from painting their data packets with the highest priority. For example, couldn’t the ISPs petition the FCC for a rule that standardizes the data type marker on the packet’s header and makes mis-marking the headers illegal? The ISPs could police their networks by occasionally sampling data packets, and violators would be fined.

    I am not suggesting that this is a good solution, but rather that a solution might be available. Indeed, even if you can reduce abuse, you still might have too much demand for QoS for the network to support if QoS is free. (Well, eventually anyway, as demand for time-sensitive applications increases.) However, increased bandwidth, distributed server providers (Akamai), and other ideas not yet conceived could help allay the demand for QoS.

    I am still not convinced that point 3 of his argument is “compelling.”

  8. Chad,

    I have yet to see evidence of a problem so why fix something that is rumored to perhaps become broken?

    It would be more prudent to strengthen the existing FCC controls – I like some of the efforts that took place in France to deregulate Internet access by requiring and enforcing a fair open access policy. I don’t know the details of any QoS restrictions but I doubt that backbone providers are prevented from offering better performance levels to premium subscribers. (We use ATT MPLS connections that indeed offer such premium services.)

  9. ChadB,

    If we agree that QoS might be OK, but you are concerned about abuses and overcharging, why not introduce regulated QoS fees?

    For example, QoS rates would be regulated in areas where there is “no” broadband competition, while they would be unregulated in areas with a lot of broadband competition. The regulators could use the competitive fees to help determine the appropriate regulated fees.

    Why is this not a superior alternative to the Snowe and Markey amendments, which would regulate all QoS fees to zero regardless of the costs or competition?

  10. Stuart,

    I believe that you have missed the primary concern that net neutrality regulation is designed to address. The concrn is that QoS fees would hurt competition in the Internet content markets where QoS is needed and the ISPs compete (for now, voice and broadcast video). QoS fees would hurt competition in two ways. First, QoS fees would prevent some competitors from entering the market by increasing the costs of doing so. Second, it would give the ISPs an advantage because they can use the QoS fees to subsidize their own internet offerings and offer lower prices, thereby squeezing any major competitors from the market. The fear is that we move from a duopoly in broadband access to effectively a duopoly in video and voice services.

    Three points:

    (1) The problem is very likely to happen, so why wait? Even Richard (who strongly opposes NN regulation) recognizes that broadcast video and voice services can be provided more efficiently via the Internet because that would allow consumers to use all the available bandwidth instead of having some reserved for broadcast video, some for other services, and some for the Internet (which is what cable basically does now). A little prevention is often less painful. Further, it will be difficult to measure the harm to innovation. If NN regulation proponents are right, consumers simply won’t know what they are missing because the innovation didn’t happen. The “If it’s not broken …” theory is a bad one to follow if you cannot see that something is broken. Consider the poor child who plays with a headless doll but doesn’t know it is missing a head.

    (2) We once had an “open access” policy under the Computer Inquiries regime, which required the telcos to provide access to independent ISPs. The ISPs could buy broadband access wholesale or interconnect with their own equipment and then resell the broadband access to consumers. Rates were on a tariff basis (i.e. the FCC has to ok them). The FCC abandoned that regime last August 2005, but, like you, some are in favor of going back to it.

    (3) You are right that the proposed regulations do not prohibit the ISPs from charging more for those who want to buy more bandwidth (faster speed). However, the proposed regulations prohibit the ISPs from charging for QoS, which is special treatment for data packets provided by routers within the ISP’s network. Importantly, QoS is different from bandwidth, and the distinction needs to remain clear. So you need to identify what “premium services” you are talking about.

    MnZ, you raise an interesting idea about trying to target the regulation to geographic markets that lack broadband competition. It raises some regulation issues, such as how to define a competitive market. Recall that some NN opponents have used the zip code analysis done by the FCC to suggest that most markets are competitive when the analysis was not designed to measure competition. Even if the FCC could do this and was not subject to regulatory capture, the costs of such a regulatory regime might be very expensive. A simpler rule–even if not ideal–might promote social welfare more because of its simplicity.

    But maybe it could work if you gave the FCC clear directions and put the burden on the ISPs to prove that a market was competitive before permitting them to charge QoS fees in that market. I would like to hear more thoughts about this idea.

  11. I suspect I don’t really grok the detail to this dispute, but in my simple minded estimation there seems to be a really simple solution.

    Abort the (from my reading not very technically literate) net neutrality bill, and replace it with one where imposing varying QoS restrictions (limitations of bandwidth dependant on the apparent use of the packets) is allowed, but with the proviso that it must never ever depend on source or destination addresses.

    This would allow different tiers of bandwidth purchase for different purposes, but would prevent the megaliths from forcing their content and their content alone upon the people who can only afford the lowest level of service. The above is the demon that the net neutrality folk are (IMHO quite rightly) afraid of. There is enough domination, especially in the US, of those who command multim(b)illion pound budgets over those who don’t, without allowing them another mechanism to buy their way to power. Imagine most search engines being too slow to use (on your limited connection) and only google showing any sign of bandwidth. This already happens thanks to the purchase of raw bandwidth, but to add to that by allowing google to purchase priority for its traffic over end-user pipelines would magnify the (already exponential) tendency by another order of magnitude.

    At the end of the day though, and again only from my simplistic POV, the suggested fix above is only a stopgap and the real solution is very very obvious. People pay for bandwidth. If there is not enough bandwidth then they should pay for more. If the folk that want to use P2P transfer want to continue doing so then the rates they pay should be sufficient to ensure adequate network growth. With enough core network, and with adequate throttling on a per user basis, I can’t see that there’s any remaining problem. Qos is irrelevant if you pay for exactly the bandwidth you wish to have available, provided the core (transnational and peer grouping) networks can handle the total.

    The difference between the curent situation and that ideal must unfortunately be covered by us (the users) though perhaps with some governmental help given the importance of the internet to national commerce.

    I’m sure I’ve missed a point or three and I hope this contribution remains long enough for someone more knowledgable to put me right.

  12. ChadB,

    You are landing on the core of the problem in that the ‘debate’ centers around very ambiguous terminology which means whatever someone thinks it means. To be specific and say that QoS is acceptable provided you don’t charge more for it help to focus the discussion. (Or to enforce regulated QoS rates.) If indeed QoS is the primary concern then it should be dealt with plainly.

    I prefer free market economics to deal with the issue of the crisis which I am, for some reason, unable to see. (The premium service I referred to is extra cost QoS (3 additional price tiers) which allow us to provide International carriage of our corporate VoIP and video.) [BTW a recent TIP study showed that most network managers are against additional fees for this kind of QoS but I attribute this to their vested interest in trying to reduce/control their expenses. However they will pay gladly if their VoIP systems are unreliable under the conditions Sigivald calls out.]

    Dave J – “forcing their content and their content alone upon the people who can only afford the lowest level of service” this condition already exists today. I can’t ‘afford’ the highest levels of Cable TV content. I am stuck with the lowest tier of basic cable (plus broadband) I am stuck with C-SPAN and local networks but miss out on HBO and MTV [and to my actual disappointment SciFi channel]. Call me cheap or poor but I don’t insist that the government ‘solve’ this. I could move to the Dish – but the Laws of Inertia prevent it.

  13. ChadB,

    You’re assuming that QoS could be used as an anti-competitive weapon. That assumption would be fundamentally wrong. Your hypothetical fear is totally illogical and misinformed.

    Your hypothetical scenario would be:
    ISP offers VoIP service and then charges customers for prioritization to get to Vonage or any other VoIP company. ISP then offers free QoS to its own VoIP service. This is where you scream antitrust.

    Here is why your hypothetical scenario is fundamentally silly:
    If we reversed the QoS offering and say that the ISP had to offer free QoS to Vonage but not to itself, what do you think would happen here? Latency to use your own ISP for VoIP services would still be FAR superior even without QoS than latency to Vonage even with QoS. How can that be? Your ISP is at worst 10 milliseconds away and Vonage is probably 30 milliseconds away. The reliability and responsiveness of something that you’re directly attached to will ALWAYS be superior. Your hypothetical scenario simply cannot exist because there is no need for the ISP to cheat in this manner because they already have the supreme advantage to begin with without playing QoS favorites.

    ChadB says:
    “However, increased bandwidth, distributed server providers (Akamai), and other ideas not yet conceived could help allay the demand for QoS.”

    This is where you show a complete lack of understanding about QoS and Bandwidth. Bandwidth is NOT a substitute for QoS just like QoS is NOT a substitute for more Bandwidth. Bandwidth and QoS are not mutually exclusive. All the bandwidth in the world cannot guarantee that an application like BitTorrent won’t simply commandeer 99.99% of the pipe can starve latency-sensitive applications. This is why people turn off BitTorrent in their own homes when they want to use Vonage and that applies to the 1.5 mbps DSL customers to the 25 mbps VDSL customers. Either that or they self-throttle their BitTorrent client or use a router configured for QoS. The problem is that you can only throttle your own BitTorrent client; you can’t throttle the thousand BitTorrent clients your neighbors are using. The only thing that will properly do this is QoS.

    You admit that Isenberg’s stupid network can only exist in theory yet you fail to offer any explanation on how to enforce QoS without fees.

  14. George,

    You seem to be arguing that the ISPs’ advantage in providing service with fewer hops–“the supreme advantage”– means that ISPs won’t use QoS to give them an additional advantage. Most companies exploit all the competitive advantages they have, but if you say that’s silly here because the ISPs are somehow different, I would be interested in hearing why. BTW what happens when a rival content provider (say, Google) sets up servers across the country to reduce the hop advantage? Will the ISPs be willing to employ the QoS advantage against their rivals then?

    I am not really one to scream. I am certainly not “screaming antitrust” here. Indeed, I am not sure antitrust can solve this problem, which is why I am interested in the pros and cons of net neutrality legislation. If you have an antitrust theory that could solve the problem, I would love to hear it.

    You’re right that I don’t have as good an understanding about bandwidth and QoS as you or many others out there. However, I agree with you that bandwidth is not a good substitute for QoS exactly because of things like BitTorrent. Nevertheless, demand for QoS exists only because bandwidth is necessarily limited, which creates the possibility (indeed, reality) of congestion. Congestion causes things like jitter, latency and loss, which can hurt the delivery of time-sensitive applications. Giving time-sensitive applications priority through QoS can help reduce these problems. As a general principle, more bandwidth means less congestion, less jitter, latency and loss and, therefore, less demand for QoS. As you point out, you or your neighbors can use BitTorrent to commandeer all of the available bandwidth, which is why I said bandwidth can “help allay the demand for QoS,” not eliminate it.

    As for your last comment, George, I’m not sure what you mean.

    I think you are asking, if we mandate that ISPs provide QoS, isn’t that a violation of the Takings Clause? If so, and we cannot mandate QoS, why would the ISPs offer QoS without QoS fees on their own? I’m not up on the Taking Clause, but if you’re right, you have asked a good question. However, the ISPs might have incentives to offer QoS on their own, and that seems to be what the NN proponents in Congress are banking on. First, QoS would generally increase the value of time-sensitive applications and, therefore, the value of Internet access. If the value for Internet access increases, ISPs can charge more for access. Second, the ISPs have stated their intention to offer time-sensitive services and content (VoIP and broadcast video). If the ISPs offer QoS, the demand for these offerings increases, as does the amount that the ISPs can charge for them. (This assumes, of course, that they cannot reserve bandwidth for their own content and applications (which is super-QoS), which they do now or have announced they will do.)

    You might also be asking how to make network owners agree on a standard algorythm for routers and how to make users paint their data packet headers with the right type code. Standardization practices in other industries suggest that it could work here for router algorythms, especially with guidance from the FCC. As for user compliance in painting data packet headers, I offer one possibility in my post above on August 10th, 2006 at 1:47 pm.

    Stuart,

    I agree that “net neutrality” is not the best descriptive term. However, the ambiguity you perceive is not so much an ambiguity as the effect of differing levels of abstraction. There is the concept of “net neutrality,” which is an abstract idea about extending market power in the Internet content markets via the market power in broadband access. Then there are conceptions of how to prevent that extension of market power, which is where NN proponents diverge. It’s just people with different solutions to the same problem, not ambiguity about the problem itself.

  15. Response to ChadB

    “You seem to be arguing that the ISPs’ advantage in providing service with fewer hops–”the supreme advantage”– means that ISPs won’t use QoS to give them an additional advantage.”

    I’m arguing that this is MOOT because of the “supreme advantage”.

    “BTW what happens when a rival content provider (say, Google) sets up servers across the country to reduce the hop advantage?”

    Google doesn’t need QoS for any of their businesses, not even video. Video (unless you’re talking about video conferencing) has at least 1000 ms buffers which make QoS moot since that buffer is the ultimate jitter-insurance. Video and file transfers require solely on your ability to cache, and carriers are NOT a good place to buy caching from because you’d have to deal with every access layer carrier in the country. Akamai is in a much better position to offer caching services. Now if Vonage moved in to your carrier’s data center, the carrier still has the advantage since they can host their own things for free. But you can take this line of thought to insanity and you won’t ever stomp out all the “inequalities”, but that’s called LIFE. Surely you would not suggest that we mandate free rack space for competing companies on the carrier’s facilities would you?

    “You’re right that I don’t have as good an understanding about bandwidth and QoS as you or many others out there. However, I agree with you that bandwidth is not a good substitute for QoS exactly because of things like BitTorrent. Nevertheless, demand for QoS exists only because bandwidth is necessarily limited, which creates the possibility (indeed, reality) of congestion. Congestion causes things like jitter, latency and loss, which can hurt the delivery of time-sensitive applications.”

    You admit you don’t understand and then you concede that bandwidth is not a substitute for QoS yet you completely contradict yourself by saying that if only we had more bandwidth then we wouldn’t need QoS. This is my problem here, you don’t even make any sense and among the pro-NN camp you probably make one of best arguments. Most in the NN camp just flat out lie about NN so I can certainly respect the fact that you’re at least honest.

    “I think you are asking, if we mandate that ISPs provide QoS, isn’t that a violation of the Takings Clause? If so, and we cannot mandate QoS, why would the ISPs offer QoS without QoS fees on their own?”

    You’re getting confused on the whole subject Chad. I said no such thing. All I said is that it’s insane to have congress get in to the business of network engineering especially when they flat out ban for-fee QoS. What I support is the true spirit of Net Neutrality, which is ban on blocking. I also support on the use of QoS as a weapon against your opponents. What I mean by that is so long as a carrier doesn’t deliberately sabotage someone like Vonage by deliberately flipping a few packets to introduce jitter (out of order packets) or deliberately drop Vonage to “background” priority from regular “best effort” service, then there is nothing wrong with QoS. Most of the QoS to be gained for most home networks is done by implementing QoS on their own routers anyways. Most people wouldn’t mind 80 ms round trip delays to Vonage’s data center and an occasional jitter or dropped packet. It is understood that cheap 99.9% reliable non-QoS VoIP is good enough for the money for most people. But if you want super reliable 99.99% QoS level business-class VoIP service, you pay more for it. Even Tim Berners-Lee admits that you must be able to charge more for better quality Voice and less for medium quality Voice. The problem there is that Tim seems to be afraid of addressing his own hypocrisy in supporting the NN camp which pushes for Markey and Snowe-Dorgan.

    Again I must stress that your fears are strictly hypothetical and irrational that a Carrier might exploit QoS in their own favor when they already have the ultimate advantage to begin with. The fact of the matter is Carriers have not done this and Vonage is still one of the biggest VoIP providers in the country. Vonage’s enemy is NOT the carriers competing with superior latency; Vonage’s enemy is EBay/Skype which recently decided to dump its product for free outbound calling just long enough to kill the Vonage IPO. But you’re so busy worrying about a hypothetical that you’re totally blind to the real injustice in Skype dumping their Skype-Out product for the sole purpose of killing a competitor’s IPO so that they will wither on the vine. Vonage by the way is already feeling the heat because their IPO sunk and they’re being sued by their own investors (which is ridiculous that you can sue for a stock dropping). If they should be suing someone, it should be Skype’s anti-competitive tactics. Vonage is also running out of money quickly and they’ve slashed their advertising spending. AOL just recently came out and started offering a free phone number along with free inbound calling and a flat $9.99 fee to call 30 countries! http://blogs.zdnet.com/Ou/?p=291 Now if only you took a breath and looked at reality versus your hypothetical fears, you’ll see just how silly you are.

  16. George, thanks for the informative response.

    Your mootness argument has some merit. You cannot stomp out all the inequities that give the ISPs an advantage over rival content providers (unless, perhaps, you prohibit them from vertically integrating). ISPs have an advantage because they do not pay access fees, have fewer hops to content customers, do not have to pay for facilities access, and do not have to pay for QoS. Eliminating the QoS advantage won’t eliminate the other ones. But on the margin, it might allow one more competitor to exist or allow for some marginal increase in innovation. The proposed NN legislation is a patch to preserve competition until more platforms become viable, not a permanent fix.

    Perhaps I need help on bandwidth and QoS. It was my understanding that data packets get delayed or dropped under IP, which makes time-sensitive applications like VoIP and video-conferencing less effective. One way to treat this problem is to allow these time-sensitive packets to have priority in the router queues, which is generally QoS. (This isn’t a complete or great description of QoS, but I think it’s good enough for our purposes.) A second way is to reduce the frequency of packet delay and loss, which, to some extent, can be done through increased bandwidth. For example, it is my understanding that networks experience periodic spikes in the number of data packets traveling at a moment, and it is during these spikes that time-sensitive applications have problems. Why wouldn’t more bandwidth reduce (but perhaps not eliminate) the severity of the spikes? A reduction in severity, it seems to me, would reduce (but not eliminate) the packet delay and loss and, therefore, the consumer DEMAND for QoS because time-sensitive applications would work a little better and would fail less often. For an increase in bandwidth to have this effect, the increase might need to be substantial and cost-prohibitive. (That’s what an AT&T guy seemed to argue here.)

    I can see how QoS probably is a more effective and less costly way of handling congestion. But you are telling me that I’m getting the concepts wrong. Can you point out specifically where I go astray? Or point to reliable literature that would explain my error.

    As for Skype and Vonage, it sounds like their competing. I am not sure that they are in a separate market from the ISPs, which means that if one fails (for whatever reason), there will still be competition in voice services. You seem to have described predatory pricing, but that is effective only if a company can recoup costs after driving competitors out, which means charging supra-competitive prices later. With the ISPs offering voice services and relatively low barriers to entry for VoIP companies, I am not sure how Skype could charge supra-competitive prices after driving Vonage out and still keep market share.

  17. BTW I am not in the pro-NN camp. I am still trying to learn about the issue. I sometimes challenge NN opponents here, but I have also challenged NN-proponents at savetheinternet.com. It seems like the best way to sift through the hyperbole.

  18. So this realization ultimately leads to the real end-goal of network neutrality: broadband Internet access networks should not be built by private companies, they should be built by government and maintained as public utilities. The goal of network neutrality legislation, then, should be to discourage private investment in broadband networks, the quicker to energize local governments to jump into the networking business.

    I suspect that is the modus operandi of most regulation. Rather than pass regulations that work efficiently with minimal side effects, regulations with wide-reaching unintended* consquences are passed. The unintended consequences demand even more regulation. Eventually, the market becomes so gummed up that the government has to step in and run the entire operation.

    *Sometimes I wonder how unintended the consequences actually are.

  19. “BTW I am not in the pro-NN camp. I am still trying to learn about the issue.”

    That’s because you’re intellectually honest. Most extremists on NN flat out don’t know what they’re talking about and they lie about the issue.

    “Perhaps I need help on bandwidth and QoS.”

    I was chatting with Dan Kaminski at DEFCON who is genius level security researcher that did the research on Domaincasting and wrote a new Net Neutrality tester. He only tests for bandwidth throttling but not latency injection but thought it was a very interesting challenge when I raised the issue with him. He even fell for this “bandwidth” fixes all latency issues and was espousing the virtues of the stupid network and stated that there is no need for QoS if only we had more bandwidth. Then I presented him a simple scenario and he backed off his stance that bandwidth solves all problems.

    Let’s say you’ve got a 1 mbps connection to the Internet and you’re making an 80 kbps VoIP call. On a stupid network, turning on BitTorrent or any other burstable application will simply consume as much bandwidth as it needs and your VoIP call will lag like crazy and experience dropped packets like crazy. Let’s say the bandwidth hungry application goes on for 100 minutes, then that means you can’t make good quality phone calls for 100 minutes if any at all. This is the jamming effect of applications that will take as much of the pipe as they can get. Now let’s say we increase bandwidth by 10 times (which isn’t cheap or trivial). Do you really think you’ve solved the jamming problem? If the pipe is 10 times faster, all you’ve done is make the jamming period 10 times shorter. This means instead of getting jammed for 100 minutes, you’re now jammed for 10 minutes. If the BitTorrent application continues to run, then all bets are off and the jamming goes on indefinitely. When I raised this scenario, Kaminski conceded the point that maybe QoS is a legitimate technology.

    On the Skype versus Vonage issue, we should note that Skype will simply raise the prices after Vonage is dead or near death. They’ve already stated they don’t intend to have free skypeout forever. That is the classic definition of dumping. A larger company with better financial resources can temporarily dump their product at below market value or in this case free. Once their competitor dies, they’re free to charge the public even more than before.

  20. George,

    Thanks for the explanation. From what you are saying, increased bandwidth can reduce the duration of the jamming problem but does not eliminate it. I think that’s consistent with my point about demand for QoS. The demand for QoS should be greater if the jamming problem lasts longer, so the demand for QoS should decrease as bandwidth increases. Increased bandwidth does not mean that you won’t have interruptions. It just means that people would value QoS less. My point is about economic demand: A person would value more fixing a 100-minute interruption in service than a 10-minute interruption in service. (Graphically, an increase in bandwidth would shift the demand curve for QoS down and to the left. What that means is that an increase in bandwidth would reduce the social welfare loss if QoS were not offered at all–which was my original point but could have been clearer.)

    As for “dumping” (a term which is usually used in the international trade context) or predatory pricing, many economists are skeptical that it happens very often, which is why the antitrust standards are pretty steep to prove a predatory pricing claim. Promotional discounts (ones of limited duration like Skype’s) to attract market share are generally considered fair play as long as they are not being used to achieve market power. Arguing that Skype has market power (or will have it if Vonage is marginalized) seems like a difficult task in light of the alternative services being offered by the telcos and cable companies. If Skype raises its prices above competitive levels (afer Vonage is marginalized), customers are likely to switch to other providers like the telcos and cable companies. That means Skype probably would not have market power, even if Vonage went under. Meanwhile, consumers benefit from Skype’s temporary cheap rates. For a good general discussion, check out Areeda & Hovenkamp’s Antitrust Law, which is widely considered the leading treatise on antitrust law.

    BTW investors probably realized (or should have) that Skype’s offer was temporary and discounted its impact on Vonage’s profitability. If Vonage is having trouble going public, the problem is probably something more permanent or fundamental (i.e. competition from ISPs, regulation regarding 911 services, corporate management, business model, etc.).

  21. “From what you are saying, increased bandwidth can reduce the duration of the jamming problem but does not eliminate it. I think that’s consistent with my point about demand for QoS. The demand for QoS should be greater if the jamming problem lasts longer, so the demand for QoS should decrease as bandwidth increases. Increased bandwidth does not mean that you won’t have interruptions. It just means that people would value QoS less.”

    This is basically a shovel with a rope handle way of engineering. You’re arguing that it would effectively suck less with more bandwidth (which is not always feasible or practical or cheap) instead of just conceding the logical point that QoS would simply make QoS work ALL the time and suck none of the time. Now try explaining “suck less” to a business person who relies on the kind of reliability his land line provides him. Why are you so obsessed about not having QoS? All it does is reserve a tiny fraction of bandwidth for tiny latency-sensitive streams. It’s almost like this thing called QoS killed your mother or something.

    This is the kind of illogical stubbornness that is so prevalent in the pro-NN camp and I’m afraid you’re falling ill to it. First they argue that QoS isn’t needed, and then they say they don’t actually ban it. If you’re willing to concede QoS is better, why keep arguing against it for a solution that sucks less when a perfectly good solution exists in QoS?

    “My point is about economic demand: A person would value more fixing a 100-minute interruption in service than a 10-minute interruption in service.”

    You ignored the part where I said BitTorrent might just keep going. You do realize that once you have more bandwidth, people will simply fill it up with HD video or even raw uncompressed DV. This is what one of the DPSProject proponents was angry about, that he can’t transmit his raw DV footage at 28 mbps over the Internet.

    There is only one way to build a good network and that is to implement QoS. This is done in all the good standards bodies. Even one of the guys that wrote the end-to-end paper is currently working on QoS technology.

    I also find it interesting how you’re willing to go out of your way to excuse EBay/Skype and their dumping behavior but you’re willing to pass new laws on hypothetical behaviors of the carriers that I’ve already shown to be completely bogus.

  22. George,

    I’m suggesting that in a world without QoS (which I don’t support yet and it seems unlikely that I ever will), more bandwidth would help time-sensitive applications suck less. In the post above, I was considering how bad it would be if S/D were passed and Richard was right that the ISPs would not provide QoS or would stop providing QoS once the router queues were overloaded to make QoS ineffective. Increased bandwidth (and maybe other things) would make that suck less.

    Contrary to your claim, I’m not making the argument that we should ban QoS because increased bandwidth would make time-sensitive applications suck less. I can see how you would call that “illogical.”

    To dispell any fears of bias, I have no affiliation with or interest in eBay or Skype and have never even used their services. (I’ve been to eBay’s website, but not recently.) My point is that free market competition is sometimes fierce and unpleasant, and it creates some waste and redundancy, but that doesn’t mean that social welfare would be better off under some other economic regime. I would be concerned if eBay/Skype were hurting competition (in the economic/antitrust sense) as opposed to winning over customers from a single competitor. But that is not what you have described.

  23. Well ChadB, I’m interested in making VoIP, games, and other latency sensitive applications not suck at all if possible with the least amount of work. Adding more bandwidth isn’t always possible or it’s very expensive and the results aren’t perfect when it comes to latency. Bandwidth usage also grows whenever you give people more bandwidth because they can always start downloading more movies at higher resolution. QoS on the other hand is a very simple solution that’s a simple software configuration. If you’re afraid QoS can be used as a weapon, you’re being illogical. There are much easier ways to mess with someone’s VoIP if they really wanted to like introducing deliberate jitter, but that would be illegal tampering.

    Like I said, better throughput and QoS is always welcome and having one is not a substitute for the other. But the arguments for Markey and Snowe-Dorgan full under the category of stupid and insincere because on the one hand they say they won’t ban QoS but on the other they argue how you don’t really need QoS.

  24. Yeah, but blocking or degrading VoIP is a risky move, both legally and from a PR perspective. The FCC has taken action against an ISP for blocking, and the ISPs would suffer bad press is they broker their promise not to degrade. QoS fees are a much more subtle (and for now legal) way of creating an advantage for ISPs’ voice services.

    I agree that QoS is a good solution for minimizing the jitter and delay for time-sensitive applications. But I still think there is a trade-off: The risk that QoS fees might leave us with little choice in VoIP, video-conferencing, gaming, etc. And we could lose innovation in those areas because the inability to compete and earn profits in those areas will deter investment in innovation.

  25. “QoS fees are a much more subtle (and for now legal) way of creating an advantage for ISPs’ voice services.”

    I think you’re stuck in a permanent brain lock here. I’m sorry to be blunt, but I’ve been patiently shooting down every devil’s advocate example you’ve made and explained how QoS works as best I could. But how many times do I have to explain this? There is ABSOLUTELY NO advantage for a carrier to give themselves QoS and leave everyone else intact at “best effort”. The only way they could force people to switch is if they broke the law and tampered with packets going to Skype, Lingo, or Vonage to produce artificially high latency. Otherwise most people won’t care and will keep using “best effort” to Vonage or Skype service. Most of the quality issues there are the multiple transcode issues. “Best effort” is good enough for most people. If they really cared about low-latency, they would rather buy VoIP service from their carrier without QoS than Vonage service with free QoS.

    “The risk that QoS fees might leave us with little choice in VoIP, video-conferencing, gaming, etc”

    Again, you’ve got permanent brain lock here. Being a devil’s advocate is one thing and is sometimes a useful academic exercise in a debate/discussion. But I’m afraid to say you’ve gone over the deep end here.

    If anything, forcing zero QoS fees will simply mean the cost of QoS will be evenly distributed to all broadband and business users whether they need it or not. Speaking as someone who would love to have QoS, Markey and Snowe-Dorgan (if the laws were adjusted to allow for limits on percentage of priority marked packets each customer can send) could actually benefit me since it would force the majority of people who don’t care that much about QoS to subsidize my QoS fees since my costs would be distributed to everyone. So in that sense Snow-Dorgan or Markey would actually force people to buy QoS whether they want it or not. The fact of the matter is, EBay/Skype will do much more to reduce your choice than anything with their dumping practices. QoS is totally irrelevant for a carrier to gain an advantage because a carrier has the SUPREME advantage to begin with. If you’re not willing to accept this argument, then you’ve gone completely insane.

  26. Get a straightjacket…

    If the ISPs have the “supreme advantage,” why do Vonage and Skype have any customers? Could the ISPs use QoS fees to make Vonage and Skype raise rates so the ISPs could win over their customers (or at least expedite the process)?

    You suggest that Vonage and Skype don’t need to purchase QoS because most people (for now) are willing to live with interruptions. To that I respond: How long will that last once? Let’s pretend that congestion gets worse as HDTV moves onto the Internet. Let’s imagine that BitTorrent becomes a bigger problem. Let’s imagine that meanwhile Vonage starts using Akamai to reduce the ISP hop advantage. QoS therefore becomes really important and the most important difference between Vonage and the ISPs. Now, even those who were willing to live with some interruptions before demand QoS because without it service is really bad. Vonage would have to buy QoS, but the QoS fees mean that the ISPs will be able to undercut Vonage because the ISPs won’t have to pay QoS fees. What happens to Vonage? Who’s left offering voice?

    Which premise is wrong? Increased congestion? BitTorrent getting worse? The availability of Akamai? Or is it that the conclusion doesn’t follow from the premises? Until you attack the soundness or validity of the argument, your “brainlock” quips will not persuade me.

    BTW I think you might be right about others subsidizing your QoS consumption if QoS fees are prohibited. Whether that’s fair is debatable. (Do we subsidize taxis and trucks because our taxes pay for the roads?)

    There are guys in white uniforms at the door. I think they want me to go with them…

  27. This is really starting to get annoying. I’m starting to think you have permanent brain damage.

    “If the ISPs have the “supreme advantage,” why do Vonage and Skype have any customers?

    Because Carriers are NOT offering VoIP in droves and offering $10 flat rate calling to 30 countries like AOL Phoneline. That kind of service is probably going to lose money for carriers so it’s really something they don’t want to compete in. Even the rate that Vonage charges $35 for all of North America is something the phone companies haven’t gotten use to and probably can’t compete against. Phone companies are use to charging a per-minute fee. Internet companies like Skype make their money by having massive numbers of users and getting EBay to pay some insane amount of money for them which they will probably never recover. Everyone is trying to get massive head count right now because they’re trying to go for the get-rich-quick by selling for billions to some stupid buyer. You really don’t know what you’re talking about. I could humor your concerns for awhile from a devil’s advocate angle, but you’ve really gone mad.

    “you attack the soundness or validity of the argument”.

    I already have, and you have no point other than some irrational hypothetical fear. You clearly have zero undertanding of the VoIP market and the dotcom-like dynamics that are in play to capture that market while losing money.

  28. What is all this about ‘more bandwidth doesn’t solve anything’ ?
    If an ISP has ample internal and outgoing bandwidth, as in more than enough to provide for the sum total of their customers allocations, then where does that leave the latency issue? The only way packets are delayed or refused is because there is not enough capacity on the destination wire. If there is enough capacity then *where* is the problem? Customers are by definition limited in the amount of data (under any protocol) that they can send, so all quantities are known.

    Ok, customers should use QoS on their _internal_ routers and their gateway router so that use of bandwidth hungry programs doesn’t impact their voip or other critical services. They should then be able to trust that their ISP has purchased enough bandwidth to be able to carry them even if they use all that they have purchased, and the ISP should be able to trust its peer group to have enough outward bandwidth to carry the traffic that it has paid for.

    Maybe a cash strapped ISP should be allowed to use QoS on its internal networking and _its_ gateways but that’s only a temporarily neccessary accomodation for poor business sense. It should not be possible for a greedy customer to swamp out critical services because all customers are by definition limited in how much they can send/receive and if there isn’t enough capacity then there isn’t enough capacity. Problem defined.

    Where’s the mistake in that scenario? OK, the money in providing ‘stupid’ linkups will be minimal, but there will, by the law of supply and demand be (just) enough profit to ensure provision.

    George writes “Adding more bandwidth isn’t always possible” – Why ? If people want it then it will be paid for. If it’s too expensive then it won’t. Surely the only real snag is ISPs selling more bandwidth than they actually lease for themselves. If there isn’t enough, then to build more, or not to build more, depends on whether the customers will pay for it.

  29. George,

    Feel relieved–this will be my last post on this string. But I wanted to recap.

    You have argued that the ISPs have a “supreme advantage” in providing quality VoIP because they are connected to users (while rival VoIP providers are a few hops away), which means less chance of congestion for the ISPs’ VoIP users. You argue further that even if rival VoIP providers could reduce the number of hops, the ISPs have many other advantages over roval VoIP providers–you cannot stomp out all the inequities the ISPs get as owners of the facilities and network. You argue that until you can prevent the ISPs from blocking or degrading, you need not worry about the ISPs using QoS fees to squeeze rival VoIP providers form the market. Essentially, you argue it is futile to try to level the playing field between ISPs and rival VoIP providers.

    I responded that rival VoIP providers might be able to reduce the hop advantage the ISPs have via Akamai or caching stations. I also argued that the advantages ISPs have might not be all that great. As evidence, I pointed out that the ISPs have not pushed Vonage or Skype from the market yet. I also argued that their most readily available tools–blocking and degrading–could cause problems from the FCC and the public. Another tool to push out rival VoIP providers–QoS fees–seems like a possibility, especially since the ISPs have suggested that they want to charge QoS fees. I have recognized that banning QoS fees throws a monkey wrench into network management (the ISPs might not offer QoS despite its usefulness) and efficiency (users not needing QoS would be subsidizing those whose uses need QoS). But I suggest that there might be ways to reduce these bad effects, such as increasing bandwidth. In essence, I have suggested that there is a trade-off: Without S/D, rival VoIP providers get squeezed from the market via QoS fees so the only choice is the ISPs and we lose innovation at the network’s edges in VoIP (and other time-senstitive apps). With S/D, we might not get the benefits of QoS or, if we do, it would be allocated among users inefficiently.

    You responded with four points:

    (1) You responded that Vonage and Skype exist only because the ISPs do not offer cheap enough rates to compete. I didn’t respond to this because if the ISPs cannot offer cheap enough rates, it suggests that the ISPs’ advantage over rival VoIP providers might not be supreme. I think that statement actually hurts your futility argument.

    (2) You argued that bandwidth is not a substitute for QoS because of things like BitTorrent. I agreed that bandwidth was not a good substitute or a complete solution for preserving quality for time-sensitive applications. I suggested, for example, that increased bandwidth could only treat–but not cure–interruptions in VoIP by reducing the incidence or severity of high fluctuations in network congestion–the source of the problem. This treatment–increased bandwidth–would reduce the social welfare loss if S/D passed and the ISPs elected not to provide QoS.

    (3) You also claimed that the ISPs could not use QoS fees as a competitive weapon to squeeze out Vonage or Skype. You argue that Vonage and Skype do not need QoS because consumers are willing to live with lower quality service. I responded that congestion is likely to get worse as demand for Internet services increases. In a world with more congestion, the reliability of VoIP without QoS will likely become intolerable for most people, and Vonage and Skype will have to introduce QoS for their services, which means QoS fees will likely become a competitive weapon. (Also, if increased congestion hurt rival VoIP providers’ non-QoS service, the ISPs would have a disincentive to increase bandwidth to alleviate the problem because Vonage and Skpye customers would switch to the ISPs.)

    (4) You responded that the QoS problem is hypothetical. I agree, but I think it is a likely problem that should be addressed in advance if possible.

    Oh yeah, you also said:

    (1) that I am “mad,” “insane,” “illogical,” “silly” and have “gone over the deep end”;
    (2) that I make “silly” arguments and “completely contradict” myself;
    (3) that I have a “complete lack of understanding,” have shown “illogical stubborness,” and have “permanent brain lock.”

    There was one more thing you said but I can’t remember because of my “permanent brain damage.”

  30. ChadB says:
    “I responded that rival VoIP providers might be able to reduce the hop advantage the ISPs have via Akamai or caching stations”

    Chad you flat out don’t know what you are talking about. If VoIP can be cached, then you wouldn’t even need QoS. A cached VoIP model has more in common with time delayed voice mail than it does a phone call. Just stop Chad, this is getting ridiculous.

    It’s ignorant people like you who base their arguments on ignorance and too much free time on their hands worrying about hypothetical threats that scare me, because people like you try to get these whacked out laws passed.

  31. Just a fast thank you to George Ou for pointing me in the direction of a logical response to my puzzlements. I don’t know my way around the (or at least this) blog medium. Usenet is more my natural habitat.

  32. George, you’re right about that sentence you highlighted. It didn’t make any sense in at least two ways. (1) After re-reading it, it sounds like I’m saying VoIP data packets would travel faster if Akamai were used or they were cached. That makes no sense whatsoever. To my knowledge, you don’t cache a live phone call. (2) What I was thinking was that Akamai could reduce congestion, which would reduce the need for QoS, but that really wouldn’t help the hop advantage for voice services. On that point, my bad.

  33. Right– just have Congress add the “free money for everybody” rider to the Network Neutrality bill to help us all pay for Akamai use.

    And remember that Google *has* its own network to reduce congestion.

Comments are closed.