[NNagain] On "Throttling" behaviors

[Sebastian Moeller]

Hi Colin,


I started composing a long reply to a previous post, but realized that due a philosophical difference that was not going in a productive direction, so instead, let's discuss our preconceptions first, as these clearly have bearing on our positions in the NN discussion? Or not, if you prefer not to go down this rabbit hole that is also fine ;)


> On Oct 4, 2023, at 17:53, Colin Higbie <colin at higbie.name> wrote:
> 
> I suspect we are all after similar long-term goals – open access to as many people as possible with performance metrics that users appreciate (as opposed to raw bandwidth marketing claims that can mask buffer bloat problems).
>  
> What I fear many here are missing in this discussion is the damage regulations cause. There is an assumption in this response that humans can architect rational legislation that will make things better. The preponderance of evidence is that regulations don’t do that.

	[SM] I see a lot of evidence for harm caused by lack of regulations or lack of enforcement of regulations, e.g.:
a) the Enron scandal (at its base fraud and the attempt to hide that fraud)
b) the Lehmann brother sub-prime market scandal, "innovations" in debt reselling resulted in a massive almost global crisis where in many countries the tax payer ended up paying for the damage caused by under-regulated finance-product-innovation.
c) anti-competitive behavior of tech companies, be it purchase of potential competitors with attractive products only to let these whimper and die, be it collusion on hiring to artificially restrict the employment supply side, ...

In these cases either stricter regulations and/or stricter enforcement of regulations could have saved us (as society) a lot of cost... 

But I mighr be looking at this too single-minded, if you want, please give examples of where regulations strangled innovation... I will however likely not accept pure hypotheticals of the kind "with less regulation we would have seen more innovation", as that is simply not provable one way or the other.



> They serve as a starting point for a new bureaucracy that grows, taxes, and crushes innovation out of whatever is being regulated. Because politics.

	[SM] Unlike corporate policy, public policy is shapeable, that is people can vote and be voted for and hence influence local policies and if they convince enough folks of their positions also state or federal government activities. Now, laws are often made as compromises, which is both bad (compromises tend to not result in simple principled statements, but lots of seemingly unrelated give-and-take) and good (as long as it is not clear that one side is completely in the rough a compromise will end up being generally more tempered and acceptable to a larger fraction of the population, which is the conceptual difference that makes liberal democracy so distinct fron the alternative systems).


>  
> I agree that there are cases where regulation is needed, because the market can’t drive the same effect because it’s too easy to push the costs elsewhere (e.g., pollution). I agree that there are historic cases where regulations have helped make things better (like with some public safety measures). 

	[SM] The IMHO clear problem with markets is, that those in a fair market tend to try to change that into a lop-sided market (so into a supply or demand side oligo- or monopoly) to gain advantages over others and under such regimes markets do not operate as efficient resource allocation optimizers, but really as profit optimizers for the oligo- or monopolist. None of this is novel insight, and all of this is happening out in the clear and can be observed in real time.



>  
> What I urge is we recognize that there should be a very high bar for regulations. In this case, where there is negligible evidence they are needed, we are nowhere near that bar.

	[SM] I respectfully disagree, NN regulations popped up, exactly because some ISPs where openly abusing their position as natural gate-keepers; as so often regulation is reactive and only happens after somebody abused the freedom offered by the older regulatory regime. That is not to say that all regulations are perfect or that all historical regulations need to be carried ad infinitum, but simply that regulations do rarely happen pro-actively, and simply dumping regulations without first checking which problem they tackled and whether that problem still exists, is an act of ideology and not of good governance. (Nobody here proposed blindly dumping existing regulations, but that idea is out there in the US and the UK and also in Germany, and probably other places as well*)

*) This often ties into the paragraph above, those that feel unhappy in a fair market seek to get those regulations relaxed/removed they consider obstacles in getting more control over their market. The sad irony is that often the same corporations lobbying for relaxaion od regulations at the same time consider everything not explicitly denied as "fair game", and sometimes even more (see e.g. the libor scandal) where they simnply factor potential fines as "cost of doing business", but I digress again.



> Absent a critical need, we are better served, at least in the long-run, in letting the market fight it out. This will no doubt lead to some bad outcomes for some people, primarily investors who back the wrong horses, but also some customers who end up with less performant connections.

	[SM] Performance isn't the issue here (most end-users would likely judge: access rate == performance) it is wll established that different rate-tiers can be proced differently. What is at issue here is that those paying for transport services can expect that they will get the transport services they pay for. If I hire a trucking company to transport a shipping container from say NY to LA, I expect them to do that and charge me a price which allows them to make a living, what I do not expect is that in LA they start extorting the receiver (we will only deliver in time if you pay us extra, or worse unless you pay extra we will not deliver at all). Internet access service is IMHO a sort of transport service, and I see very little need for any fancy "innovation" here, sure faster more efficient techniques will be appreciated, but these really only affect the cost/price structure and change nothing on the underlaying contract structure.


> However, that free-for-all drives things forward en-masse and the poor options die off. That’s the product and service evolution, or Darwinism if you prefer, that drives optimal outcomes.

	[SM] Let's leave natural evolution out of this please, never a good analogy if looked at closely... (I am a biologist by education and profession)



> Faith in someone’s intelligent design (no matter how good you think it is) will not yield positive results, or at least not as positive in the long-run. Even if it starts out well, it will get twisted by political horse trading and bureaucratic control into something worse than you envision, never better.

	[SM] I hear/read this quite oftem, what you call "political horse trading" is essentially what policy is supposed to do, make compromises over a set of acceptable positions such that a majority can stomach the resulting hodgepodge... Now, this is not always perfect and often these compromises contain hard to swallow components not designed for the greater good, but that does not invalidate the process of finding positions that a majority can tolerate or even support.


>  
> Now, if you want to talk about forming a voluntary organization to make technical recommendations and perhaps standardize some language (like the Wi-Fi and USB consortiums), I think that’s fine and would support that fully. If done well (e.g., the latest USB C was not, with a lot of uncertainty around what USB C means and unlabeled cables more than the connectors determining power delivery capacity and bandwidth), the language around this and a standard means of testing and communicating bandwidth, latency, buffer bloat, etc. assists the better solutions to dominate in the market.

	[SM] I can only speak from limited experience in the IETF, and I am not convinced that the IETF process is in the same league as political consensus finding (with all its "horse-trading" and sausage making), heck it is not even playing the same game. IETF process has no real accountability and is based on the laudable idea that with enough good-faith acceptable compromises can be struck without politics and horse-trading. The problem is that in reality that process can only work like intended if everyday acts responsibly with the "greater good" in mind. That might happen occasionally, but what I observe is closer to horse-trading, sausage making and consensus by attrition, but I digress. My point is the existing political process, for all its warts and potential for improvement is IMHO not as bad as you seem to imply.


> I also support legislation specifically around busting the regional and local monopoly contracts that cable companies use. Those currently prevent much of the competition that I embrace as the rapid driver for commercial improvement.

	[SM] That is not going to help, there is not a queue of folks that are just waiting for local monopolies to end to throw fiber cables into the ground and hook up whole neighborhoods... IMHO what would work better is to force local monopolies to carry bit stream traffic for competing ISPs at a price close to the incurred cost (as if the monopolist would need to hire its own access lines, with a smudge of surplus on top)... I base this on what happens in my own home market, where the incumbent is forced to offer wholesale products for other ISPs, with regulated prices that such competitors (if acting efficiently) can still make a profit when offering internet access below the price of the incumbent. This while heavy on regulatory intervention works reasonably well and means that in the incumbents foot print consumers can actually chose between a few different ISPs.


Regards
	Sebastian


>  
> Cheers,
> Colin
>  
>  
> From: dan <dandenson at gmail.com> 
> Sent: Tuesday, October 3, 2023 5:41 PM
> To: Network Neutrality is back! Let´s make the technical aspects heard this time! <nnagain at lists.bufferbloat.net>
> Cc: Sebastian Moeller <moeller0 at gmx.de>; Colin_Higbie <CHigbie1 at Higbie.name>
> Subject: Re: [NNagain] On "Throttling" behaviors
>  
> I know we're well off on a tangent here, but it's all somewhat related.  The competitive landscape really dictates how much regulation needs to be involved.  The less the competition, the more the government needs to take a roll to keep that small pool or single vendor in check.  Government really must take a regulatory rule here in my opinion because there are bad actors and so much of America relies on them for mediocre services.
> 
> The who's and how of previous monopolies really do matter because we can't take a true lesson if we don't have the facts.  I continue to stand by my stance that the virtual monopolies in many markets is why we really need NN or near-NN and soon.  I do really appreciate everyone's perspectives though.  Lots of great stuff here.
> 
> The problem I see is that from my perspective, we need a 2 pronged attack.  As Gene says above, presenting a post analysed solution to the powers that be might be helpful.  
> prong 1 is to clarify what NN is in legal terms and build transparency in.  If a service is NN they can choose to say so, if they are not they have to say they are not.  I might lose this battle and all services may have to be NN, IDK, I'm just presenting my thoughts.  That's the main purpose of this thread but I think a lot of members of the mailing list also have an interest in the second:
> prong 2 is to identify 'what went wrong and how to fix it'.  Why do so many people have poor service when we have so much infrastructure?  No access.  As an operator in multiple states I assure you all that access to existing fibers that were built by government money is minimum, even at splice huts getting services can be impossible.  I literally ran fiber through my current service area on contract in the late 90's that I cannot purchase services off of today. it sits there dark.  I know it's there, I feel like I still have the blisters.  I would suggest the microIX somehow, maybe pushing for schools to have a microIX hanging off of them or city halls or something.  Heck, the schools could get 100G and x dark fibers delivered to the nearest IX and let them sell ports off of that for a tariffed rate.  I'm sure we could come up with 2 dozen all for the public good models to get microIXs all over the place to solve this.  I promise that access to high capacity data will spur competition like crazy.  We have a school in our area that we didn't bid out but has a 100M fixed wireless link from a competitor as their only service and word on the street is that it doesn't deliver.  It's crazy.
>  
> On Tue, Oct 3, 2023 at 2:26 PM Colin_Higbie via Nnagain <nnagain at lists.bufferbloat.net> wrote:
> Sebastian,
> 
> Good points and thanks for the conversation. I agree with you on the INTENT of the NN regulations as proposed and that most of the non-content provider supporters had what you stated as the goal. However, the money driving the politicians, paying for those ads and social media campaigns, etc. came from the content providers who wanted government support and protection. This doesn't make the carriers right (or wrong), just know that the big supporters were the content providers and for obvious business reasons – they wanted to get their content to customers for free.
> 
> There are plenty of good and valid arguments on both sides that issue, but history shows that in the long run, government interference in the form of controlling regulations on what business may and may not do with each other, outside matters of public safety and perhaps establishing and mandating some measuring standards, is always destructive to innovation. That in turn hurts consumers in the end. 
> 
> On the Google point, I respect that you disagree that Google's offering in my hypothetical where it's bound to their own services would not have been as good as open 20Mbps Internet access. I would probably feel the same way. But that's not the point. The point is that if Google came in OFFERING that, it would have been disruptive. As long as customers had THE CHOICE of 1G content-controlled vs 20M open (or whatever they had access to), then Google's offering is only beneficial in the long run. The entrenched competitors would need to up their game or lose at least some customers. They would either increase their capacity and available bandwidth or tout the benefits of their open access or something else or some combination. That's how customers win from competition – it's not just price, or just bandwidth, or just access freedom, or any one thing. It's the unpredictable freedom to innovate and find new niches that customers want to pay for. And the market is ALWAYS better at determining which is the better value for customers. None of us as individuals (as much as Xi Jinping might disagree) can out-predict the crowd-sourcing power and wisdom of the entire free market.
> 
> I completely agree with you that my saying "it might have spawned significant investment..." is speculative. The core of that is the fundamental point of economics though: individuals making choices for their own self-interest is what drives innovation and advancement in a way that ultimately helps everyone. In other words, the regulation would also have been speculative that it would help more than not having that regulation. And it would have assumed BOTH that innovation won't solve current problems AND that customers are too stupid to choose the service that's of value to them. 
> 
> Given a choice between speculating that future innovations might solve problems or speculating that we're doomed without government protection from companies who have not yet really done the bad-Google-hypothetical-thing I described, I would much rather err on the side of letting the market and innovation continue to run with things. After all, that's what took the Internet from a military and academic network into the most impactful economic force of the past decades, and one infused by a culture of innovation and entrepreneurship.
> 
> Tying that together with your point on the government policies protecting local monopolies: Why did Verizon start to build FiOS? Because they thought they could attract customers and earn a strong profit by providing a better, faster Internet via FTTH. On paper, this was a slam dunk. Why did they stop building it out? Because they ran into too many localities who blocked them with legislative and regulatory hurdles. 
> 
> That FiOS example is one of the best case studies of exactly how government regulation primarily stifles and harms end user experience with respect to Internet access speeds.
> 
> I also agree with your concluding points that if carrier access were regulated, business would do their thing, adapt, and find a way forward. The world would not end. All true. No dispute on that. However, I would say that if we look 20 or 50 or 100 years out, the state of human communications and computing technology would be further advanced in unpredictable ways by not removing some of those axes of freedom from the entrepreneurs as they innovate. The gains are only in the very short term. Protecting my Netflix access today is nice, but getting 1Tbps access (or sub 0.1ms latency for radically different levels of computation interactivity, or something else we don't even realize matters yet) to be commonplace in 20 years might be even better.
> 
> Taking all of the above together, this is why if we're looking for a STARTING point (and to be fair to your points, maybe more would be needed after that, but let's tackle the biggest, most indisputable problem first), START with dismantling those government-protected monopolies. Maybe, that will be enough. Given that innovation-crushing regulations tend to grow and are rarely ever retracted, better safe than sorry on this.
> 
> By the way, I say this as a guy on the content-providing side of the problem. So, to the extent that I have business bias, it would be in favor of forcing equal access. However, I also believe that the same regulations that would force equal access today would also stymie the development of 10G and 1T networks and beyond in the future (not their engineering, but the actual commercial deployment). Advancement and innovative growth occur when innovators and entrepreneurs have the freedom of the open sky (with all the confusion and wrong turns and failures that admittedly come with that), not under a smothering tarp of regulation. 
> 
> Cheers,
> Colin
> 
> 
> -----Original Message-----
> From: Sebastian Moeller <moeller0 at gmx.de> 
> Sent: Tuesday, October 3, 2023 3:50 AM
> To: Network Neutrality is back! Let´s make the technical aspects heard this time! <nnagain at lists.bufferbloat.net>
> Cc: Colin_Higbie <CHigbie1 at Higbie.name>
> Subject: Re: [NNagain] On "Throttling" behaviors
> 
> Hi Colin,
> 
> > On Oct 2, 2023, at 22:34, Colin_Higbie via Nnagain <nnagain at lists.bufferbloat.net> wrote:
> > 
> > While product and service innovation often originates from pure R&D or work performed in academic labs, in virtually all cases, converting that into commercially viable products and services is the result of profit incentives. A company won’t invest in doing something new with attendant risks unless they can expect a return on that investment greater than the alternatives (or they believe it will provide strategic support to some other product or service). For that reason, we want to be extremely careful about regulating how companies can implement innovations, including the use of potentially distasteful business practices. None of us who want to see the Internet become better over time and more accessible should want anything resembling NN regulation.
> 
>         [SM] At its core NN regulations really just say that who is selling internet access services is supposed to do exactly that and not try to act as gate-keeper picking winners and losers. I might be insufficiently creative here, but I do not think a simple "do not discriminate" directive really restricts the space of potential innovations in any meaningful way.
> 
> 
> > The regulatory side of this is largely not a technical discussion because future innovation, by definition, may exceed technical considerations we can conceive of today.
> 
>         [SM] Indded, prediction is hard, especially predictions about the future ;)
> 
> 
> > It's easy to conceive of examples where an ISP wants to prioritize or penalize certain kinds of traffic. And while that may seem superficially bad, it’s an important part of the very competition that drives innovation and cost reductions over time. E.g., recall when Google Fiber had been willing to install Gbps fiber in places at a time when most of the rest of the country was struggling to get 20Mbps connections. If Google had wanted to limit that to Google services, that still might have been a boon to those customers.
> 
>         [SM] I respectfully disagree, that would not have been meaningful internet access. An unrestricted 20M internet access link has more general utility that even a 10G gate-keeper only link (who that gate-keeper is is irrelevant). (I am not saying the 20M would be without issues)
> 
> 
> > Further, it could have shown the uses and values of what was then considered limitless bandwidth for a home or small business user.
> 
>         [SM] Yeah, on that question we are still waiting even though >= 1 Gbps services are not all that rare anymore. As far as I can see it we still lack use-cases that strictly require fast links that go above simple "more parallel" or "faster".
> 
> 
> > Even though this would clearly have been in violation of the tenets of NN, it would have provided important data that might have spawned significant investment by others and advanced the state of connectivity across the board.
> 
>         [SM] This is purely speculative though, it might as well had shown nothing of that kind by the sheer fact that google fiber roll-out was so small as to be not representative of anything, no?
> 
> >  
> > I know the counter argument to this is that local ISP monopolies already break innovation, and those companies, especially the big cable companies, therefore have no incentive to provide a good service. I largely agree with that (there is still some small incentive, in that if they are too terrible, customer outcry will turn to voter outcry and demand breaking those monopolies, and they don’t want to risk that).
> >  
> > Therefore, the legal issue to address is NOT how they treat or prioritize data, whether by content or protocol – which they should be allowed to do, EVEN WHEN IT’S BAD FOR CUSTOMERS – but, at least referring to the U.S. specifically with our federal/state system, to put federal limits on durations of regional monopoly durations. I believe this is within the scope of what FCC can mandate (some would debate this and it may take the courts to sort it out). These need not be purely # of years, they can be a function of time to recoup deployment costs. If a company negotiated a local monopoly as part of covering their deployment costs, I would personally say that they should be given an opportunity to recoup those, but then after that, they need to open up their lines for use by competing firms, similar to what happened with the RBOCs and the old telephone lines.
> 
>         [SM] The problem is that access networks often are not legal monopolies, but natural monopolies where if company A has a high-speed capable network deployed it becomes economically unattractive for other companies to deploy their own network (the competitor can torpedo such a deployment by lowering prices such that too few customers change to make the whole thing stay in the "loss" region for a long time). So leaving the access network to market players will always result in the incentive to monetize the gate-keeper role that is inherent in the network's structure. 
> One solution to this problem (not the only one) is to put the access network into the public hands, like other important infrastructure. The idea would then be like in Amsterdam, Zuerich and a few other places to have a local access network provider that in turn "concentrates" access links in COs local IXs where interested ISPs con connect to and then offer all end-users in that access network internet access services. That still leaves the natural monopoly of the access network untouched, but puts it under management of en entity that is not likely to exploit this (as fully as private entities are).
>         This is however pretty orthogonal to direct NN concerns, and I am sure not a generally accepted model. (Say if I would be operating a small ISP and would differentiate myself by how well I manage my access network, I likely would detest such ideas, and if I would operate a big ISP I would detest them for other reasons ;) so this is ver end-user centric and also relies on some modicum of faith in local government)
> 
> 
> > This is also the legal logic behind patents: give a company a 20 year monopoly on the invention in exchange for making it public to everyone and showing them how to do it (the patent must provide clear instructions). We deem the temporary monopoly worthwhile to incent the innovation, provided the inventor makes it public. This is the right philosophy to consider for something like bandwidth innovation, investment, and access.
> >  
> > In short, with ISP’s the open-ended government protected monopolies are the problem,
> 
>         [SM] Again these often are not legal monopolies where nobody else is permitted to build a competing network, but natural monopolies where the expected return of investment falls with the number of already existing networks, while the cost stays constant. AND the number of ISPs tgat might actually bite the bullet and set diggers in motion is still so small that in the end, we might change from a monopoly to an oligopoly situation, bith are regimes in which the free market does not really deliver on its promises.
> 
> 
> > not the providers’ ability to overcharge customers or prioritize some data over others. Competition will fix that over time, as long as competition is allowed to occur. And while it may be faster to force it through regulation, that has dangerous long-term consequences with respect to future innovation.
> 
>         [SM] Yes, meaningful competition could help, but IMHO an oligopoly likely would not result in meaningful enough competition. This is where the access network in public hand ideas comes in, it makes the cost to enter a market for ISPs relatively cheap, they really only need to pull/rent fibers to the local IX and maybe deploy OLTs/DSLAMs/CMTSs there (depending) on the local network tech, and can start offer services, without having to deal with the access network.
> 
> > Starlink is one example of innovation. FTTH is another. Cellular-based Internet is another.
> 
>         [SM] All of which are orthogonal to NN regulations, neither depended on violating the "do not discriminate" rule, no?
> 
> 
> > Simply buying bulk access on existing lines and repackaging it under different terms could be yet another. Those all seem obvious, because they’re the ones we know. The real danger in unforeseen consequences is the dampening effect NN-style regulations have on yet-to-be-seen innovations, the innovations that never come to fruition because of the regulations.
> 
>         [SM] I claim that rules and regulations always set the stage for which business decisions are acceptable/profitable and which are not, that is true whether we add the NN mandates to the mix or not, so I really do not see how these will have a meaningful influence on future expected innovation (unless that innovation really is all about active discrimination, but in that case I see no real loss).
> 
> Side-note: The thing is "discrimination" is still permitted under most NN rules, as long as it is under active control of the end-users, not the ISP. So I am sure some end-users would appreciate an "prioritize vide conferencing and VoIP over video streaming and gaming under load" option offered by their ISP and might even be willing to pay a little, as long as the end user can toggle this option at will it will not be subject to NN regulations as far as I understand. This clearly leaves some innovation space available even for active discrimination.
> 
> Regards
>         Sebastian
> 
> 
> >  
> > Cheers,
> > Colin Higbie
> >  
> > _______________________________________________
> > Nnagain mailing list
> > Nnagain at lists.bufferbloat.net
> > https://lists.bufferbloat.net/listinfo/nnagain
> 
> _______________________________________________
> Nnagain mailing list
> Nnagain at lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/nnagain