Below you can find materials from CDM/CIPPIC's presenation before the CRTC in support of the SaveOurNet.ca coalition.

Oral Submission to the CRTC on July 9th, 2009.

Full Transcript Below:

3306 MR. FEWER: Thank you, Mr. Chair, and my thanks to the Commission for providing CIPPIC and the Campaign for Democratic Media with the opportunity to appear before you today.

3307 My name is David Fewer, I am the Acting Director of CIPPIC, the Samuelson-Glushko Canadian Internet Policy and Public Interest Clinic at the Faculty of Law at the University of Ottawa.

3308 We act for the Campaign for Democratic Media, which is a network of public interest organizations and people pushing for media democracy in Canada.

3309 I am joined today on my right by Stephen Anderson, co-founder of the Campaign for Democratic Media and SaveOurNet.ca coalition.

3310 I am greatly pleased that the following three individuals were willing to provide their time and expertise to our panel and to the Commission: on my left, Dr. David Reed, Dr. Andrew Odlyzko and Mr. Bill St. Arnaud.

3311 Dr. Reed is currently Adjunct Professor at MIT’s Media Lab. He played a significant role in the development of the underlying architecture of the Internet, contributing to the design and development of IP, TCP and UDP, the protocols central to today’s Internet. Dr. Reed is co-author of the seminal paper establishing the end-to-end networking principle.

3312 Dr. Odlyzko is a Professor in the University of Minnesota School of Mathematics. Dr. Odlyzko heads the Minnesota Internet Traffic Studies project, a research project that collects information about Internet traffic from a variety of sources that continuously monitor traffic on a variety of networks. The MINTS project is, we submit, one of the leading independent sources for rigorous tracking and assessment of global Internet traffic patterns.

3313 Bill St. Arnaud is the Chief Research Officer of Canada’s Advanced Internet Development Organization, CANARIE Inc., where he has been responsible for coordination and implementation of Canada’s next generation optical Internet initiative.

3314 We appear before you today one simple objective: to convince you of the value of the open Internet, an Internet that is neutral as to source, destination, content and protocol.

3315 The Telecommunications Act provides the Commission with the tools it needs to address traffic management practices that undermine the open Internet. The tests we propose in our comment permit the Commission to establish a normative and prospective framework that will do two things.

3316 First, this framework will guide ISPs in how to implement acceptable traffic management practices and, second, this framework will provide Canadian consumers and businesses and competitive ISPs, including those in the wholesale market, with trust and certainty that the Canadian Internet will continue to serve as a platform that supports innovation and competition.

3317 Our submission will proceed as follows.

3318 First, Mr. Anderson will provide a social context and a broad citizen’s perspective on the matters raised in this public notice.

3319 Second, I will provide brief comments on the manner in which the Commission has framed the relevant issues in its Schedule 2 and provide a substantive response.

3320 Mr. Anderson…?

3321 MR. ANDERSON: Thanks, David. And thank you for having me and for holding this important hearing. I’m thrilled that the CRTC has taken on the issue of traffic management.

3322 I am here as part of the Campaign for Democratic Media and I am the coordinator of the SaveOurNet.ca coalition.

3323 The SaveOurNet.ca coalition is a broad-based coalition of citizens, businesses and public interest groups fighting to protect our Internet’s level playing field.

3324 The coalition has over 115 member organizations as diverse as non-profit organizations like the Canadian Federation of Students and the Council of Canadians, to media organizations like Now Public, to businesses like Tucows Inc., to ISPs like Acanac and Teksavvy, to labour organizations like the National Union of Public and General Employees and CEP.

3325 I think it is important to acknowledge that this issue — this is an issue with a few Internet service providers and their partners on one side and nearly everyone else in Canada on the other. Businesses, civil society, cultural groups, everyday people, social, cultural and economic innovation and consumer choice are on this other side.

3326 The main issue at stake here is who will determine the way we use the Internet, will it be the users or will it be ISPs?

3327 You have received over 11,000 public comments in this hearing. That should tell you where the Canadian public stands. The values inherent in the open Internet are the values that are attracting widespread citizen support and I am here as one of those Canadians.

3328 The positions this panel puts forward correspond with what I have read from citizen submissions and what I have heard at the open Internet Town Hall events we organized in several cities.

3329 As you work for the Canadian people, I hope you will give these submissions due weight. I also encourage you to go through the citizen comments yourselves and come up with your own synthesis, which I’m sure will closely match our positions here.

3330 In this hearing Canadians are looking to the CRTC to establish guidelines for ISP traffic management practices that will ensure ISPs do not stifle innovation and that will safeguard emerging technologies.

3331 A Canadian citizen named Terrill put it succinctly in his submission and I quote:

“Canada has a long history of Internet-based innovation. It may not have been possible if it weren’t for the open, unbiased characteristics of the Internet.” (As read)

3332 The Commission’s policy objective in this hearing should be to ensure ISPs can manage their networks, reduce congestion and keep the Internet open at the same time. If we fail to create the right balance and ISPs are allowed to determine access conditions for applications, it could have unintended consequences for innovation and consumer choice. I would argue that those consequences are already being felt.

3333 If we don’t adopt the right rules, the main way — or if we do adopt the right rules the main ways ISPs will compete is by increasing their bandwidth offerings. This is what we want.

3334 If we allow ISPs to slow down certain applications, as they are now, we actually create an economic incentive for bandwidth scarcity. If you can slow down and control traffic you can make more money if the resource is scarce. It’s like allowing bottled water companies to control public water systems. Of course, they will have an incentive to keep the water scarce and expensive.

3335 This is not a good path to take if we want ubiquitous bandwidth and the economic, social and cultural innovation that comes with it.

3336 As Canadian citizen Paul aptly put it in his submission:

“Allowing discriminatory traffic shaping creates a disincentive to upgrade networks and provide faster, better service.” (As read)

3337 It’s part of the problem and not the solution.

3338 What’s more, with the U.S. and others supporting the open Internet, if we do not move in the right direction Canada could end up as a backwater of online innovation. In these tough economic times we can ill afford such a fate.

3339 If Americans have permission-less innovation, so must we.

3340 As a Canadian named Roland submitted to you, and I quote:

“As a small business owner/operator we rely on the Internet. I am deeply concerned that a non-neutral Internet will harm my ability to compete on a level playing field with larger companies. Net neutrality is what has made my company possible. Please do not allow the Internet revolution we have seen in the last decade to be stifled by the interests of big business especially at a time when President Obama has declared his intention to defend net neutrality in the United States.” (As read)

3341 So the vote from the Canadian public is in and they want an open Internet. With that, I will let David Reed and the rest of the panel detail our specific recommendations. Thanks for having me.

3342 MR. FEWER: Moving on to the substantive portions of our presentation, we offer three preliminary comments.

3343 First, this hearing ought to establish a forward-looking result. We ought to be — this ought to be a norm-setting exercise. It ought to set bounds on permissible ISP behaviour with the objective of providing ISPs with competitive security and Canadian consumers and businesses with confidence that they can rely on the continuing openness and neutrality of the Canadian Internet. It ought not merely establish a reactive remedy for violations of the Act.

3344 Second, contrary to the claims of some of the participants in this proceeding, establishing these rules for managing an open and neutral network does not amount to lawyers building the Internet instead of engineers. First of all, ISPs build their networks further to business decisions, not simply to engineering decisions. If the engineers were really in charge, then Dr. Reed would be somebody else’s witness, not mine.

3345 Third, there are public good aspects to the Internet. That is something that has been underlying a great deal of the submissions in this proceeding but it needs to be stated. The Canadian Internet is not simply the private property of retail ISPs, it’s more than that. It is greater than the sum of its parts. It is a platform that Canadian businesses, innovators, entrepreneurs and citizens rely upon every minute of every day.

3346 Now, Canada has made a policy choice early on to rely upon the marketplace, on private parties to facilitate the construction of this critical resource, but we can’t allow the private to trump the resource. We may talk of an ISP’s facilities but it is Canada’s Internet and I submit that this is reflected in the Telecommunications Act.

3347 Now, we have a few short comments with respect to the manner in which the Commission has framed the issues in Schedule 2.

3348 We believe that the objective as stated is a good one:

3349 - it shows a willingness on the part of the Commission to treat this process as a forward-looking process instead of simply a reactive one;

3350 - it acknowledges that the Internet is a network for the public good;

3351 - it recognizes that the Internet fosters innovation and creativity;

3352 - and while it does recognize that there may be legitimate management practices that ISPs can employ, it also limits these practices to protecting legitimate interests;

3353 - and finally, it recognizes that there are privacy concerns and legislative constraints that apply to invasive traffic management processes.

3354 It is a good start. We do, however, have some concerns about the stated definitions and assumptions.

3355 The definition draws a distinction between the public Internet and the private services. We worry a little bit about creating regulatory incentives for ISPs to carve out ever larger chunks of the public Internet for private services and provisioning the private at the expense of the public.

3356 We also recognize that this is an issue that goes beyond the scope of this hearing but we see it as an appropriate subject matter for a future CRTC proceeding. This is an issue that is not going away.

3357 We believe that the assumptions made in Schedule 2 are functional but miss key elements that would assist the Commission, we submit, in making a decision that better meets its objective.

3358 So the first assumption is essentially that unrestricted traffic increases lead to congestion and congestion leads to a deterioration in services.

3359 A simple economic truth: when there is an increase in demand without a matched increase in supply, there will be shortages. However, the assumption as stated completely ignores the supply side of the equation.

3360 The assumption would be better stated as: “Unrestricted increases in Internet traffic can lead to congestion in all or part of an ISP’s network if these increases are not met with adequate provisioning.”

3361 The ISPs argue that they are provisioning to the best of their ability and that this provisioning cannot meet the demands of all the new traffic being created by P2P applications.

3362 We argue that the evidence suggests differently. When Comcast, for example, was required to throttle traffic to meet this uncontrollable growth in demand, it found in fact that it was enough to throttle less than 1 percent of users for no more than 15 minutes at a time. This doesn’t seem like unmanageable traffic to me.

3363 Provisioning is directly related to the requirements for traffic management practices. Functional marketplaces meet demand with supply, not by squashing demand.

3364 Now, this is how the Internet has always worked. Historically other bandwidth crises have been met by building capacity. That is how today’s crisis will be met too. We invite you to speak to Dr. Odlyzko on that point.

3365 The second assumption fails to define what is meant by traffic management practices or integrity of the network.

3366 Now, in our comment we distinguish between acceptable traffic management practices that respect the open Internet and traffic interference, invasive practices that interfere with end-user traffic such as application-based throttling or RST insertion — injection.

3367 Traffic interference should be permissible only transparently as a last resort where finely tailored to target congestion, to target the problem we’re talking about here, and where implemented in a manner that minimally impairs the user experience and is justified by congestion metrics on an appropriately provisioned network.

3368 The failure to define integrity of the network is troubling because it glosses over the fundamental threshold issue, when may an ISP intervene in traffic.

3369 We would argue that traffic interference practices are not appropriate for ISPs to use in order to maintain the day-to-day integrity of the network. However, we do agree that in extreme cases what we have been calling the Obama inauguration type moments or Mother’s Day type scenarios that certain traffic management practices may be appropriate.

3370 The last written assumption states that in order to grant approval under section 36 the Commission is governed by the policy objectives in section 7.

3371 We agree that this is relevant. However, we don’t agree that the Commission should ignore the content of section 36 itself. The purpose of section 36, which is to prevent any influence by a carrier on content or the purpose of communication should be taken into account.

3372 The Commission should also then consider the degree to which content is controlled or the meaning or purpose is influenced and whether there is a less intrusive method for the ISP to accomplish its objective.

3373 Now, the unwritten assumption here. Before we move on we would like to address the non-inclusion of section 27(2) in any of the assumptions or the questions.

3374 At the very least we would like to see a standard developed for determining if a management practice is a violation of section 27(2). This will prevent an overly high hurdle for complaints under section 27(2) and provide greater guidance to ISPs going forward.

3375 Moving on to the substantive comments, we have three key points here:

3376 First, P2P-specific traffic management is discriminatory. It’s a prima facie violation of section 36 and it conflicts with the policy objectives.

3377 Second, we argue that P2P-specific traffic management is simply unacceptable.

3378 And finally, that the Commission must provide a principled framework to guide ISPs in the future.

3379 The framework that we have proposed in our comments will lead, we submit, to cost-effective solutions that are compliant with the Act.

3380 Our bottom line is this: Currently ISPs have no incentive to tailor their practices to the requirements of the Act. Let’s say an ISP has two options, each equally capable of achieving the ISP’s objective, option one interferes with telecommunications to a greater extent than option two. Currently there is nothing to push the ISP to pick option two, the less invasive practice over option one.

3381 Our framework will give ISPs guidance so that they can find new and innovative ways to meet their legislative objectives in ways that are sensitive to the requirements of the Act.

3382 The effectiveness of this paradigm, we submit, is evident from what happened in Comcast in response to the FCC ruling. In that response, Sandvine and Comcast together were able to create, within a matter of months, a solution that is far superior and at the same time cost-effective.

3383 In the Canadian context this solution is also far less invasive of section 27(2), 36 and the principles in section 7. It also preserves the carrier’s role of ISPs in that it doesn’t require them to make decisions based on information not traditionally available to them.

3384 CDM is confident that given proper guidance from the Commission, Canadian ISPs will likewise develop solutions that are cost-effective in line with their respective management philosophies and also tailored to the requirements of the Act.

3385 As a result of this hearing there may be some minimal short-term costs to some ISPs as they update their practices to comply with the Act. Comcast faced similar costs and did not find it necessary to significantly raise prices to defray related costs.

3386 In the future, ISPs and network equipment providers will know at the development stage to factor in the statutory requirements of the Act when crafting network management practices. Once these factors are accounted for in the planning stage, the costs associated with them will be even lower.

3387 The resulting benefits will be far more innovation, less discriminatory — a far more innovative, less discriminatory open Internet that is more responsive to the needs of Canadians.

3388 P2P-specific traffic management is antithetical to the principle of common carriage generally, and specifically with the principles enumerated in 27(2) and 36. It also strongly conflicts with the policy objectives.

3389 First of all, it is discriminatory. It discriminates against a class of applications. While slowing down a file transfer application doesn’t necessarily prevent the application from functioning, it does put any such application at a competitive disadvantage relative to a file transfer application that is not throttled.

3390 Proprietary P2P protocols like BitTorrent are also placed at a competitive disadvantage, as application developers will hesitate to rely on these protocols when deciding how to solve their own file distribution issues.

3391 It also discriminates against ISP users who wish to use such applications to view content.

3392 In these hearings you have heard from a number of groups who rely on such applications to distribute their content.

3393 Finally, it sets up incentives for ISPs to confer undue preference on their own distribution services, especially with respect to wholesale customers. This is a very problematic but difficult issue to monitor.

3394 Application-specific throttling is also a prima facie violation of section 36. It controls content to the message, and, more to the point, it influences the purpose and the meaning of the telecommunication.

3395 In the case of a peer-to-peer file sharing application, the purpose of the telecommunication is to transmit data as quickly as possible. Slowing down the rate clearly influences the meaning and purpose of that telecommunication. Doing so can easily amount to defeating the purpose altogether if it causes users to abandon peer-to-peer file sharing applications or the BitTorrent protocol in favour of other forms of file transfer.

3396 Finally, application-specific traffic management conflicts with the policy objectives, detracts from the ability of independent Canadians and other artists to distribute their content, and hinders any other future innovation based on the P2P file sharing system or the BitTorrent protocol. It provides ISPs with strong incentives to rely on traffic management instead of investing in provisioning, and this will further degrade Canada’s internet infrastructure.

3397 It is a privacy invasive solution, because it requires ISPs to consider which application customers are using, information that ISPs, again, traditionally, don’t have access to.

3398 Since peer-to-peer-specific management prima facie violates 27(2), 36 and 7, an ISP implementing such methods must justify such a practice. It must, first, target a legitimate ISP objective, and second, it must adopt a practice that is proportional to that objective by being rationally connected to the objective, minimally intrusive, and provide benefits that outweigh the detrimental impact.

3399 Peer-to-peer-specific management practices, we submit, fail to do this.

3400 CDM has stated in its comments and maintains now that, in the context of traffic management, a legitimate objective is one that targets congestion that cannot be met through provisioning alone, such as the Obama-type moments.

3401 Throttling to meet normal peak traffic is not a legitimate objective.

3402 ISPs have claimed that, in the near future, traffic growth will be so great that it cannot be met with provisioning, but that’s not yet occurring, and there is no reason to abandon provisioning based on these kinds of extremist projections.

3403 We would submit that these projections aren’t consistent with the independent evidence, in any event.

3404 Targeting peer-to-peer is not a rational response to this legitimate objective. Peer-to-peer traffic can be addressed through provisioning alone. It currently produces a disproportionate amount of traffic, and this is due to its popularity as an application and its newly emergent status.

3405 Some have said that peer-to-peer allows 5 percent of ISP users to generate 50 to 60 percent of traffic. It should be noted, first, that this figure does not directly reflect peak traffic periods created by that 5 percent. In other words, this is not a true correlation to congestion, the problem that we are supposed to be looking at in this hearing.

3406 Only peak period traffic contributes to congestion.

3407 More to the point, penalizing an entire application platform isn’t a rational response to the practices of a small number of its users.

3408 Targeting peer-to-peer is not tailored to the requirements of the Act. More tailored solutions are available, and others can be developed. The Comcast solution and various IETF processes mentioned in our comments are examples. There are others on the record in this hearing. These don’t single out specific classes of applications, and so are less discriminatory. Instead, they directly target users actually causing congestion.

3409 It is not peer-to-peer applications that produce disproportionate traffic, but specific users of peer-to-peer applications.

3410 That is a point that I have heard made over and over at this hearing.

3411 Secondly, these solutions interfere with telecommunications to a lesser extent. They operate only in the presence of actual congestion, and so are less intrusive of section 36.

3412 Finally, an application-agnostic approach doesn’t impact detrimentally on content distributors, it doesn’t hinder innovation based on targeted application protocols, and it raises fewer privacy concerns, as it need not track which applications customers are using.

3413 Given that peer-to-peer-specific traffic management isn’t minimally impairing, its salutary benefits can’t outweigh its judgmental impact.

3414 Some have raised concerns — this is at paragraph 76 of our submission — that the proposed justificatory framework will be too stringent in the context of private companies, as it was developed for the protection of rights.

3415 I don’t agree that this is the case. It has been applied in the private context to determine what is just, acceptable, or reasonable. It is merely a means of analysis, and its stringency is in the application. It’s a flexible framework.

3416 Others have claimed that the justification framework is overly subjective and will substitute value judgments for those of ISPs, or that it will force one-size-fits-all solutions onto ISPs.

3417 This misconceives the nature of the minimally intrusive requirement, and that requirement is a method for adjudicators to defer to the legitimate objective of others.

3418 It allows ISPs to choose from a range of options, only requiring that these be tailored to the legal principles found in the Act.

3419 With some guidance from the Commission as to what these principles require, CDM anticipates that most ISPs will, from hereon in, adopt methods that meet this criteria from the start, and will have no trouble gaining approval from the Commission.

3420 The proposed paradigm is an ideal method for the Commission to implement its statutory requirements under the Act, while respecting the policy directive.

3421 This test essentially requires the Commission to interfere with an ISP practice only when it is clear that market forces have already failed to ensure the policy objectives, or sections 27(2) and 36 of the Act are accounted for.

3422 When an ISP adopts a method that makes no reasonable attempt to minimize discrimination or to influence the purpose of a telecommunication, or to hinder innovation or privacy, that signals such a failure.

3423 This is what we have been seeing here with application-specific throttling. No effort has been made to develop targeted solutions. The incentive to do so comes from the statute, and the CRTC must make sure that that is respected.

3424 Those are our comments, and we look forward to your questions. Thank you.

3425 THE CHAIRPERSON: Thank you very much for your submission.

3426 I presume you were here when the previous parties were before us, and yesterday when this whole issue came up, to which we can’t get a clear answer, and since you have some eminent experts with you who have no stake in this debate, maybe I can get one.

3427 The whole idea is, when the ISP sells to a wholesale customer and, through them, further sells to the retail market, and there is congestion in the wholesale customer because, let’s say, for argument’s sake, using the jargon, most of its customers are bandwidth hogs, et cetera, and cause congestion clearly on that ISP’s network, is there an ability for the congestion to affect the rest of the network, so that the wholesale provider suffers the consequences, or is it basically the problem of the wholesale customer who provides to the retailers?

3428 MR. FEWER: Bill, you are probably in the best place to take a swing at that one.

3429 MR. ST-ARNAUD: I wish there was an easy answer. It really depends on the technical details and the relationship between the wholesale customer and the wholesale provider.

3430 In some situations, where they have a dedicated pipe and port, the effect of the congestion on retail will not propagate into the wholesale network.

3431 In other cases, where they are just reselling DSL, for example, it may be possible that that congestion could propagate into the wholesale market.

3432 But without the technical details and the market relationship, it is very hard to say whether it will occur or not.

— Laughter

3433 THE CHAIRPERSON: That is not a helpful answer. It could, you say.

3434 MR. ST-ARNAUD: Yes, I know.

3435 DR. REID: Let me try to bring it down one more level, so we can dive into the depth of the technology, which is probably where you don’t want to go.

3436 THE CHAIRPERSON: Absolutely not.

3437 DR. REID: An access network, such as the network that the DSL wholesale customer is using to deliver its service, by whatever means — PPOE or whatever — it has some fundamental capacity within that access network, and what can bleed over is, if the wholesale customer’s customers — the retail customers — are using more aggregate capacity than that wholesale customer had purchased, or whatever, at a point in time, then that will degrade that access network.

3438 It is much less likely, depending on market share, as in the earlier presentation, that it will degrade anywhere upstream from that access provider, so it will dissipate relatively quickly.

3439 But because the arrangement, the business arrangement in that circumstance, doesn’t actually amount to purchasing capacity, but sort of purchasing the right to use capacity, there can be errors in that contracting situation.

3440 I suspect that a far better remedy — rather than just interfering with traffic, a much simpler remedy would be coming and saying: Hey, you are exceeding what you contractually agreed to limit your capacity to, or limit your consumption to.

3441 So the wholesaler would speak to the retailer and say, you know, “In aggregate, your customers are using more than you said they would use at this point in time, do something about it,” and they would either do something about it by implementing protocols at their end users, or they would be asked to pay more.

3442 And that would lead to more provisioning.

3443 As Bill said, it depends on the business arrangement and the particular technology that is underlying that arrangement.

3444 But, from a technical point of view, you could actually implement, entirely within the retailer’s resources that it has access to, a mechanism that would throttle its customers, so that it stayed within the bounds it had negotiated with its wholesale provider.

3445 THE CHAIRPERSON: Secondly, one of the principles that you are advocating to us is that there shouldn’t be any application-based throttling. If you do throttling, you throttle the capacity, or use other means, but don’t single out a particular application because, if I understand you correctly, you may punish innocent users of that application, as well as the bandwidth hogs.

3446 It may also have an asymmetrical impact, you know, impacting some wholesale customers quite differently from others, depending on what type of customers they have, et cetera.

3447 If I buy the assumption, and I buy your analytical framework, which is based very largely on what we do in terms of Charter evaluations — section 1 of the Charter — I guess, theoretically, there could be an instance where an application-based discrimination was justifiable, if you go through all of the steps of analysis.

3448 Or, are you going from the assumption that in no case ever should — by using application-based throttling, you are automatically offside?

3449 MR. FEWER: Offside.

3450 THE CHAIRPERSON: I’m sorry?

3451 MR. FEWER: I would say, using application-based throttling, that you are automatically offside, for the reasons that we went through in our submission.

3452 The real issue for me is rational connection and minimal impairment.

3453 In imposing application-based throttling, you are going overbroad. You are going to be throttling people who aren’t contributing to a problem.

3454 And you are not tackling the problem. The problem is congestion, and there can be many sources of –

3455 THE CHAIRPERSON: No, I understand. You are saying, a priori, no matter how you apply the framework, or what the facts are, you could never come up with a solution which would say yes.

3456 MR. FEWER: Yes. It will never be sufficiently tailored to the problem.

3457 And, really problematically, it has downstream network effects that are very problematic from an innovation perspective.

3458 THE CHAIRPERSON: Would it make sense, let’s say, assuming that we adopt your analytical framework and we apply it — we say, “That’s what we will do,” and that is, in effect, the method by which we look –

3459 And we are complaint-based, so our examination is always ex post facto.

3460 If you put an exception in, if you try to do something application-based, you need ex ante approval.

3461 According to you, you would never get it, but I am not going to make an absolute ruling. God knows what the technology is, et cetera, but because of the likely — the most likely outcome being unjust discrimination of anything that is application-based, you come ex ante.

3462 If you want to try it, be my guest and try it. Anything else, you do whatever — here are our principles. You have to be neutral. You have to be technology-agnostic, et cetera. However, if somebody complains, then we will apply this analytical framework to decide whether you are onside or not.

3463 Does that make sense to you?

3464 MR. FEWER: Yes, that is the framework we are proposing, and that would be the result that we would expect, that an application-based traffic management practice is going to have so much difficulty getting over that rational connection and minimal impairment hurdle, and there are so many other options available. Really, that is the key.

3465 THE CHAIRPERSON: Okay, thank you.

3466 MR. FEWER: David, did you want to respond to that at all?

3467 DR. REID: Just perhaps on a slightly different angle on the same question you asked. I think there is a specific response that has to do with the fact that we are talking about internet service, not just any communication service. Internet service has a fairly well defined meaning, which includes delivering all packets all the time.

3468 Beyond whatever legal framework you have in Canada, it is accepted across jurisdictions that traffic will get through, and that there is a definition of internet service that supports all applications equally, or at least fairly among each other.

3469 So as long as you are talking about regulations that apply to someone who claims to be offering an internet service — and this came up in the FCC hearing, where I testified on this — there may be a distinction between internet service and other communication services that is quite important from a market point of view, which is that if you claim to be offering an internet service, you have to offer what you claim, and if you take the word “internet” out, you have a lot more freedom.

3470 So when someone claims to offer that, that could be a crucial distinction — an additional reason not to distinguish among applications.

3471 THE CHAIRPERSON: My whole question was based on — you are asking me to make an a priori ruling that application-based throttling is offside. I am really reluctant to do anything a priori, when I can see saying, “No, you need ex ante approval,” et cetera, because it is so contrary to the principle of internet, as your colleague just said. Therefore, you come first, et cetera.

3472 But you never know what the situation may be, so rather than making it a priori, I would categorize: those are ex ante, the other ones are ex post.

3473 And, from my understanding, you would see that that would be an acceptable or a logical way to proceed.

3474 MR. FEWER: Yes. I see now what you are getting at.

3475 You don’t have a complaint in front of you right now that a given ISP’s behaviour — that a particular ISP’s application-based throttling is a violation of the Act, so you are expressing your reluctance to, basically, rule that in this hearing. I understand that.

3476 Our view is that, by adopting the framework and saying, “This is the framework by which Canadian ISPs must be guided when considering and implementing internet traffic management practices,” it would be readily apparent that an application-based approach is going to be offside. It is going to fail that test.

3477 So ISPs would have the opportunity, then, to revisit their practices and get back onside the Act.

3478 THE CHAIRPERSON: But, also, for future — I mean, if you actually think you are in such a dire strait that you have to do an application-based approach, you don’t even contemplate, you need ex ante approval.

3479 I would sort of put that out, clearly, as a benchmark.

3480 MR. FEWER: Yes, I think, if you can get ex ante approval in those circumstances, then, go for it.

3481 I am just trying to think of what circumstance an ISP would be in that –

3482 THE CHAIRPERSON: None of us has a crystal ball.

3483 MR. FEWER: Yes, that’s true, you never know what is going to happen.

3484 THE CHAIRPERSON: Tim…

3485 COMMISSIONER DENTON: Good morning, gentlemen. If some of our questions seem a little focused, it is because, I think, there is broad agreement on some of the principles that have been advocated by several parties here, so I want to get into the issues of tests, criteria and what standards to apply, rather than broad principles.

3486 I notice, for example — and it is one of my micro nightmares that it might arise here — that carriers, users and internet types might all have exactly the same idea as to the principles to follow.

3487 It is illustrative, for example, if we look at the Open Internet Coalition, they had three principles: Does the traffic management practice in question further oppressing and substantial objective.

3488 Second, is the traffic management practice narrowly tailored, et cetera.

3489 And, third, is the traffic management practice the least restrictive to reach the objective.

3490 Now, this is remarkably similar to what you have come up with in your paragraph 18.

3491 Where do we go with this?

3492 As I hear you talk, and read what you have written, there seems to be a sense that some of the criteria and procedures developed in the Internet Engineering Task Force might be relevant guides to what we ought to be doing, or how we ought to be applying rules here.

3493 Can you comment, please, about the adequacy, pertinence, and applicability of IETF rulings, procedures and standards?

3494 DR. REID: I am happy to comment. I am not sure that my views are universally held to be truth, so this will be personal and based on my experience, and so forth.

3495 The IETF is a remarkably good organization at vetting technical alternatives. If you have a problem to solve –

3496 I think one of the principles that was mentioned here was the idea that you ought to subject it to scrutiny, try to decide whether it is the least invasive approach, whether it really solves the problem, whether the problem has been characterized properly, and so forth, and, as an organization, the IETF does a pretty good job, as do some other organizations like it in providing a very broad set of inputs.

3497 Its output is not regulatory in any sense, so the IETF has some standards that it produces that are technical standards that have been agreed to, and they tend to be fairly conservative.

3498 And then there is a whole collection of recommendations and other kinds of things that don’t hold any sort of legalistic weight.

3499 But what they do hold is, actually, a lot of weight because of their content and the fact that all of the parties who are involved in engineering the internet participate in that process, just as all of the parties that are involved in legal issues participate in a process like this one.

3500 So it does provide a counterpart to a regulatory process of surfacing issues, discussing them, and munching over them, and that sometimes takes some time.

3501 For example, the task force that started dealing with the problem that Comcast was having in the U.S., there has been a quiet and fairly active group trying to say: Well, what is really the problem?

3502 There was probably some problem that caused Comcast to spend some money to buy equipment and install it. Let’s look at the problem and get out of the political domain and see if we can solve it.

3503 And they, actually, have made a great deal of technical progress that has influenced Comcast’s decision, and so forth.

3504 So my sense is that it has a role to play. It’s role can’t be king of the internet, but it plays a very important role in providing a place for technically neutral discussion among parties who may have economic — who may be, actually, competitors, or otherwise have serious problems.

3505 COMMISSIONER DENTON: Would anyone else care to venture a view on this?

3506 MR. ST-ARNAUD: Following Dave’s comments, I think the Engineering Task Force has an important role in defining the technical issues here. In fact, there are meetings in Stockholm as we speak on network neutrality, and so forth.

3507 But I think the issues facing the Commission here are much more business and regulatory issues on this topic. There are technical things that we could contribute, but I think the real problem is something that the Commission has to address.

3508 COMMISSIONER DENTON: It has been presented to us that a considerable portion of the traffic management problems that we are focused upon are sort of a creation of inadequate capacity caused by insufficient competition.

3509 I would appreciate your comments on this view.

3510 DR. ODLYZKO: It’s very hard for me to tell what’s causing it in particular since we’re talking about the Canadian situation which I’m not as familiar with.

3511 However, in almost all cases any congestion that arises is a result of kind of collision of two factors, supply and demand and kind of the supply not being adequate to the demand.

3512 And the question then is, kind of, how do you handle it, to what extent do you throttle the demand and to what extent do you pull the supply and to some extent that’s regulatory policy or a business decision.

3513 From standpoint of studies I have done I’ve observed that demand is not growing very rapidly, it’s growing at rates that are sort of comparable to the rates at which technology is improving.

3514 You can look at, say, the Notebook computer that you bought this year, it probably has four times the memory of the one you bought three years ago and you probably didn’t pay any more for it than you paid for a previous model.

3515 Similarly, the technology suppliers to telecom companies like Nortel or Alcatel, Lucent, Cisco and so on are boosting the capacity of their products, and as far as I can tell at the moment, maintaining a steady rate of capital investment is adequate to provide increases in capacity that handle the increasing demand that exists.

3516 What happens in specific cases for particular ISPs and so on, well, that really depends on local situation, I cannot really kind of address it, this kind of very large, very global kind of observation for the industry as a whole.

3517 COMMISSIONER DENTON: But, Professor Odlyzko, given that global observation, can you just repeat that second to last sentence I think it was, that you’re saying that a steady increase of capacity is what, sufficient to meet…?

3518 DR. ODLYZKO: A steady increase of capacity that’s provided by using the latest technology at the current levels of investment is sufficient to meet the current levels of increasing demand.

3519 COMMISSIONER DENTON: And I know your studies, but perhaps you could elaborate on how you have arrived at that conclusion.

3520 DR. ODLYZKO: Well, so the main part of the project that I’m running, MINTS Project, is to monitor trends in traffic growth. This is a part I can address most directly, have greatest amount of expertise on and we just observe that traffic growth has not — has been declining.

3521 Still we have very vigorous traffic growth, but it has come down from a level, say, a hundred percent per year say half a dozen years ago, down to a level of more like 50 percent per year in many places, in particular based on the data that was released by the Commission based on what was supplied by the Canadian ISPs. In Canada it’s down to around 35 percent, at least it was last year.

3522 If now I draw the conclusions from data from other sources, kind of what suppliers are quoting for their equipment or so, it appears that that level of growth in traffic can be handled by the increases in capacity of equipment that’s provided by the suppliers to telecom.

3523 In other words, a kind of — as far as I can tell at the moment there is no need to increase the capital investment of telecom sector, therefore, no need to increase the prices to consumers in order to accommodate the traffic growth that we observe in most places, including Canada.

3524 COMMISSIONER DENTON: Thank you.

3525 What I’m going to ask you to do is, in any order among Reed, Odlyzko and St. Arnaud, if you gentlemen had a message to leave us with, I would like you guys to give us each your particular message that you really want us to absorb in the context of this proceeding.

3526 Take it in any order you like.

3527 DR. REED: I was hoping to have a couple of seconds to think.

— Laughter

3528 DR. REED: That’s a very broad question.

3529 I guess the message that I would leave you with, because we’ve clearly spent a lot of time with David Fewer and other CIPPIC people and so forth and pretty much I would sort of look at the top line of a forward-looking process that doesn’t react so much to this point, the crisis of the moment, over react is probably where the Internet has been well served as have other sort of businesses that are based on technology growth and rapid adoption.

3530 There is a crisis perhaps in some ISPs, it may be a crisis partly of their own making, they didn’t anticipate demand or surprised by take-up in some area, but the general principle of sort of standing back and letting the — you know, helping the problem get sorted out at the technical level before, you know, taking a regulatory move to enable some new technique that sort of violates a lot of boundaries perhaps, is where I would try to stay.

3531 I think the whole engineering community is very interested in what is happening with video, for example, and there’s clearly been a shift in the delivery of what you might call asynchronous video from peer-to-peer networks where — that was largely because the main suppliers of video weren’t willing to play their content on the Internet at all, to services like Apple TV and Hulu and so forth.

3532 And I would at least argue that peer-to-peer will continue to play a role in video distribution, maybe not as large, and that all of those are traffic hogs.

3533 When my kids watch Apple TV every night and Hulu at their dorms and I know that’s the dominant part of capacity now, maybe more than the other things.

3534 So, I would say, you know, video may be a big problem, it may not be peer-to-peer that form takes and trying to focus on peer-to-peer to stand for the fact that video is gradually moving over to that space might be a premature solution and might privilege the video providers, the Hulus and the Apple TVs against, say, Canadian independent film makers to take a sort of an extreme example, who might not have access to Hulu or Apple TV and from a market point of view.

3535 So, you can have very big side effects in this space and I would say go slow and think forward.

3536 DR. ODLYZKO: So, one point I would like to make is it would be worthwhile to collect data on a continuous basis, for example, on things like traffic levels, also distribution of traffic among different applications.

3537 Much of the kind of hype about kind of the flaws of video swapping and has basically been hype, there’s been kind of no substantiation for it, goes contrary to what we observe is happening out there.

3538 On the other hand, there is growth, there’s continuing growth which I think is very healthy in many ways and there’s certainly potential for substantially greater growth in the future.

3539 I cannot kind of totally discount the concerns industry expresses about video files on the Internet. If all of the video that is now being watched over cable networks or over-the-air TV was suddenly to show up on the Internet, it would definitely, you know, crash the network.

3540 However, if you observe previous rates of adoption of other technologies, you seldom see kind of this big quantum jumps.

3541 I think, again, instead of anticipating things or kind of approving intrusive traffic interference methods, a priori, it would be better to have a continuing effort to measure, you know, to better measure what is going on to be able to make informed decisions.

3542 The other point is just kind of essentially to reiterate what I said before. One of the things actually that has caught me by surprise over the last few years is I’ve observed traffic growth slowing down.

3543 Back about 10 years ago when I first got into this game I kind of was expecting that we would see continuation of a doubling of traffic each year, which would have required architectural changes, increases in investment, et cetera, which I felt would, again, be healthy for the industry, but I saw sources of potential additional traffic which might feed this doubling of traffic each year that has been kind of — that prediction has been proven false, traffic growth has come down and I’m kind of somewhat puzzled by it. Again, reinforces the need to actually watch what it is that people do as opposed to reacting to some fears that are stirred by somebody without direct kind of data.

3544 MR. ST. ARNAUD: Yes. I just want to echo my colleagues’ points.

3545 One I think is not to focus on peer-to-peer. We’ve seen data, I think Sandvine reported here earlier that peer-to-peer traffic is around 20 percent now and there also is another study from the coalition — research group in the United States that peer-to-peer traffic’s dropped substantially on our networks.

3546 Now, there’s a number of factors to that. Some of it may be hidden under other protocols, but there’s also some suspicion that, you know, people now are buying legitimate music and videos, on the sites now they’re publicly available, now that the content providers are making it available on the Internet people are naturally gravitating to legitimate sites, and also probably some of the pressure of the RIA and MPA on piracy.

3547 So, I think it was concerns not to fight yesterday’s war and the new applications, video in particular is now the big driver, but to focus on — so to focus on application of peer-to-peer I think maybe it’s misleading us.

3548 The other thing is, in these new applications, particularly video and Canadian content and so forth, the early adopters are going to be the big users. So, you know, when you see individual users taking big traffic, this is not necessarily a bad thing, this means this is a new market opportunity and so forth and to recognize that — and to sort of encourage that type of development because that’s where, hopefully, some day we’ll have new growth in the traffic.

3549 And so, it’s really — it has conditions that do not restrict that innovation and adoption of new practices I think is very important.

3550 COMMISSIONER DENTON: Thank you, Mr. Chairman. Those are my questions.

3551 THE CHAIRPERSON: Len?

3552 COMMISSIONER KATZ: Thank you, Mr. Chairman.

3553 Just one question. Having just prominent university scholars in front of me forces me to ask this question.

3554 Universities in general have a reputation of consuming high band width for many reasons, academia, research, students, everything else as well.

3555 What are universities doing today, if anything, to control the use of their infrastructure, given that it has been growing as quickly as it has?

3556 DR. ODLYZKO: Well, it’s hard for me to tell about universities as a whole. There are some associations of university IT directors who might be able to provide you with such data.

3557 I can talk about University of Minnesota more specifically. We have a very large, very high capacity network. The part of it that goes to the dorms that is capacity controlled, there are rules against, you know, controlling peer-to-peer, students are all warned, you know, against pirated content and we have attorneys on our staff who respond to — take down notices from RIA or MPA constantly.

3558 And actually one of the problems is that, you know, they obtain a notice about some peer-to-peer application, some suspicious application running, they often have to go down and contact the users because in a large fraction of the situation it’s a totally legitimate scientific peer-to-peer applications.

3559 One of the problems — a little kind of aside — one of the problems of public discourse, peer-to-peer has been identified with pirated music and video. It might be true that the majority of peer-to-peer traffic is such, but in a university environment as ours that is not the case, there’s a lot of totally kind of legitimate scientific kind of or other approved users.

3560 So, in that case, university staff do have to worry about it and it’s kind of bothersome in terms of being…

3561 COMMISSIONER KATZ: There isn’t user prioritization or controls that are being put in place?

3562 DR. ODLYZKO: No, there’s no prioritization, other than throttling for dorms in general. For the academic unit, there is no prioritization whatsoever. So, the network is very lightly loaded, the connections to the Internet are lightly loaded and it’s actually a very good illustration of effect that demand does not jump and fill all of the pipes.

3563 And much of what you often hear from industry is a claim that you just boost capacity, then users will simply boost their usage. That doesn’t seem to happen. There are lots of counter examples to that.

3564 COMMISSIONER KATZ: Thank you.

3565 THE CHAIRPERSON: Okay. Thank you very much for a very useful and thoughtful commentary.

3566 Thank you.