Hacker News Reader

Prioritization is only necessary or useful when bandwidth is scarce and your endpoints would use more than you have if they could. The best solution for all of the applications listed is more capacity, not prioritization.

What, then, does their eagerness for prioritization tell us? It tells us that these companies would like to stop investing so much in increasing capacity, and instead squeeze as much money as they can out of their existing infrastructure.

They're going to make things worse (or let them gradually become worse over time) just so they can charge extra money to people who need something better.

Beyond that, prioritization is not in conflict with net neutrality. If my ISP wants to offer me a premium subscription that prioritizes my traffic over normal subscribers, that's fine. Where we run into trouble is when they start doing this for other people who aren't their customers. If they want to charge me more for better service, that's fine. If they want to charge Netflix more to provide me with better service, that's not fine.

Comcast has $5 billion USD on their balance sheet in cash. They can afford to add more capacity.

Of course. But they'd rather have even more cash if they can.

They also have about $120 billion USD in debt. Not exactly the healthiest company in the world financially speaking.

> Prioritization is only necessary or useful when bandwidth is scarce and your endpoints would use more than you have if they could. The best solution for all of the applications listed is more capacity, not prioritization.

This doesn't make sense to me. Sometimes there can be large spikes in network traffic. In theory, if your network has enough capacity to handle _all_ the traffic at low latency during normal load, then the prioritization has no effect anyways. It's only there for when you are low on capacity.

I'd love for someone to correct my understanding here.

>Either way, this is an important point and I'm surprised that the HN crowd doesn't discuss it more.

But right now all latency-sensitive applications should be operating just fine. I regularly play online first person shooters, and that is a latency-sensitive application if there ever was one. So that strongly suggests, seeing how I'm on Comcast / Xfinity's cheapest plan, that all lanes are fast lanes right now.

Accordingly, that means if you were to

>nuke [net neutrality] and let the markets figure it out

that's actually just letting a monopoly extract more profit from their network by intentionally degrading the quality of some connections unless paid more money.

I don't really think I need to say more, if this was a fair and competitive market where giant ISPs weren't lawsuit-happy in order to keep out competition, or they weren't able to do that successfully, sure, maybe the market would figure it out, and eventually the ISPs that offer true net neutrality would probably have traffic because no consumer is interested in paying more to be able to use the internet in specific ways.

As it stands, it's extortion being codified into law via bribery, disguised as 'the removal of heavy-handed Obama regulations in order to improve the network you've never had any issues with'.

A lot of games are low latency specifically because they worked with ISPs to reduce latency, or worked with a server provider that did so:

https://www.polygon.com/2016/2/17/11037220/league-of-legends...

As I understand, the monopolies like Comcast and Time Warner have almost no incentive to actually improve their infrastructure unless they have the ability to shake down the services that want speedier connections when they need them.

And this exactly how lack of net neutrality (combined with lack of infrastructure competition) messes up market signaling. A game then inflates its price, and all customers of the game pay a little bit extra because in some areas the ISP is shaking them down. The pricing pressure lands on the game company to resolve, in part with costs to customers which only have an indirect link to the transaction, not on the ISP that could actually resolve the problem with the direct parties involved.

Hi, what is market signaling?

It is how markets use prices as signals to decide what to produce and deliver efficiently. If the pricing of the good if ISP performance is actually hidden from the end consumer, then it muddies the customer valuation of both the game as a product, as well as an ISP as a product. By linking the prices of both in a hidden way, consumers don't have an effective way to make clear product choices which in turn signal to a company that it is producing the right thing (or the wrong thing). If the signals aren't clear, then the market becomes less efficient. The game company is no longer competes or is fully focused to make a good game, and the ISP is no longer competing to produce better network access.

That’s bullshit. I run a small telehealth company and latencies are not “just fine.” I spend much of my day dealing with internet connections for telehealth that are far from just fine, and it isn’t because of raw speeds, but the fact that “everything should be equal” is to ignore the fact that bandwidth is finite. Streaming YouTube is not as latency sensitive as telemedicine. Yet under NN — that traffic has to be treated equally. If I could pay a surcharge to have my traffic prioritized, I would, but under existing rules, real-time telehealth is if equal importance as some kid watching Kinder Suprise egg videos on YouTube, yet if an ISP wanted to prioritize real time traffic, under NN they couldn’t. Now, maybe ISPs could compete on offering things like “optimized for video conferencing or VOIP” or “optimized for large downloading needs” or whatever.

Not all traffic is equal and not all consumers have identical needs.

What the REAL issue is is local governments providing de facto monopoly power to ISPs thus preventing the sort of competition that would make a much more innovative and customizable product for consumers. My needs are more geared around real-time and less around large downloads, yet here in France where I live, all traffic is identically treated, supposedly.

There are unintended consequences with NN that I feel that people don’t understand. The internet isn’t the same as a phone call — some traffic has different requirements in order to be successful, while phone calls are all pretty much the same.

Like I said, the most latency-sensitive topic I can relate to is first-person shooters; in particular, I play the game CS:GO, where one shot can knock you out of a round. I play within a 40ms ping area, generally around 20-30ms of ping, and that seems to get me (from California) more or less the whole west coast of America.

I experience virtually no issues from my internet, and like I said, I pay $30 a month for this service. I don't use a lot of band-width, I simply use a consistent stream, and I see no reason why I should have to pay more for that right. It's an automaton wiring packets, not people scrambling to move data.

Back to you, though: what sort of issues arise from this extreme sensitivity to latency you have? Is 'telehealth' just a doctor skyping with a patient or do you have some kind of real-time devices that depend on communication over a distance? If the latter, how sure are you that your architecture isn't the issue, and some computations couldn't be done in a different link in the chain?

>the REAL issue is local governments providing de facto monopoly power to ISPs

I totally agree. Like I literally said in my comment, the whole calculus of this conversation would change if ISPs actually had to compete, but since they obviously don't, I can't possibly see how revoking net neutrality is a good idea. It simply lets them tighten their stranglehold in a fashion that costs everyone more money and extends the anti-competitive sphere they exist in into the whole realm of business that relies on sending traffic over the internet.

So sure, it's easy to say "telehealth has a real impact on lives and YouTube for children doesn't", but that doesn't mean eliminating net neutrality here and now is a good choice.

I hope this is an exercise in playing devil's advocate and not true naivety. Do you really think telcos care about these issues? They are a business and not a socialist government.

Other comments have stated their goal clearly: increase margins from existing infrastructure.

If fair repartition of resources was their goal, why not make that an explicit part of the laws they want to introduce? Why do it in a roundabout way? Presenting a wall of 'possibilities' only serves to obfuscate their true goal.

Would you let 'the markets figure out' your water supply?

It makes more sense if you promise me X Mb/s download speed then give that to me, if you are not able then don't promise that speed, I should decide for what I need the speed, I should not have horrible speed for X protocol/site/app that I use 99% of the time but you have great speed for Netflix but we both pay same price. I want things to be fair, customers, websites, apps to be treated the same.

QoS is not net neutrality, and confusing the two is a common tactic of anti-NN shills who can’t make actual points.

I'd love to hear more about the difference between QoS and "prioritization", as they are calling it.

The existing rules contain an exception to the no-throttling and no-blocking rules for "reasonable network management." Here is how the Commission has defined that term:

> A network management practice is a practice that has a primarily technical network management justification, but does not include other business practices. A network management practice is reasonable if it is primarily used for and tailored to achieving a legitimate network management purpose, taking into account the particular network architecture and technology of the broadband Internet access service.

It's debatable what exactly this means, but providing higher QoS for traffic that needs it--in a way that does not discriminate between providers of those services--would seem to be a strong candidate.

> in a way that does not discriminate between providers of those service

Exactly. Anti-NN (broadly) wants to be able to discriminate on the basis of provider, not service.

"Provider" and "service" aren't always distinguishable.

Does that excuse provider-based discrimination in the cases where they are distinguishable? I'm pretty sure those are the overwhelming majority.

I take "legitimate network management purpose" to mean improving some metric of overall usage satisfaction. This is tricky, and already leaves a lot of room to maneuver.

QoS is decided by the user/service. Paid prioritization is decided by the ISP.

Every connection can have a max high-priority bandwidth and a max best-effort bandwidth, so people won't benefit from just marking every connection as high-QoS.

Any sane ISP either ignores or replaces QoS tags on ingress to their network.

Currently, ISPs strip/normalize the TOS bits on incoming traffic. ISPs could enable DiffServ (especially at the CMTS/DSLAM) and charge content providers to respect their TOS bits, allowing traffic like VoIP and gaming to get lower latency/jitter. But they don't do that.

Instead, ISPs are creating artificial congestion on their transit links (by leaving router ports unplugged) and then selling that bandwidth to content providers via paid peering to avoid the congestion. Basically "it would be a shame if massive packet loss were to happen... never mind, we already did it". So yeah, paid peering probably does have lower latency than congested and bufferbloated transit links, but it's not like ISPs are using any kind of QoS as the term is traditionally understood.

ISPs always talk about low-latency applications (think of the deaf people!) but then they don't do anything that would actually reduce latency/jitter.

QoS can't be implemented without payments to authenticate intent, but the argument would hold more water if the fees are merely transfer payments between users and Comcast gets $0 out of it, or ... Comcast has competitors.

Neither is true, so QoS basically should not be implemented and it's Comcast's own fault for being a rent-seeking monopoly.

People probably believe that the positive effects are on the more obscure or hypothetical side while the negative effects would be more severe and definite.

Sorry to the hearing impaired community for not prioritizing your latency needs, but Fox News buying up all the ISPs and controlling the news is worse.

The funny thing about this is: All of the negative effects are hypothetical. Nobody offers "fast lanes" now, is planning to introduce them in the near future, and nobody offered fast lanes before the regulation was put into place. And yet, it's the top net neutrality supporter panic call.

And yes, there's a whole list of individual token offenses ISPs have committed that people will spam out when you point this out, but it doesn't change the fact that the fears and horrors propping up this movement on a large scale do not exist.

> The funny thing about this is: All of the negative effects are hypothetical

No they aren't; many of them are things that the FCC responded to in if years of enforcing net neutrality principles through case by case actions, such as mobile and fixed broadband ISPs blocking services that competed with their offerings, and blocking other lawful services or content, sometimes for political reasons.

> Nobody offers "fast lanes" now, is planning to introduce them in the near future, and nobody offered fast lanes before the regulation was put into place. And yet, it's the top net neutrality supporter panic call.

“Fast lanes” have gotten blown out of proportion because of discussion of a possible weakness in an earlier draft of the 2015 order. They aren't actually a concern for the non-neutral net, they are a concern about a potential end-run around a net regulated for neutrality but with a loophole. Unfortunately, a popular distortion side that that draft was designed by enemies of neutrality rather than people trying to save it who might have designed the draft insufficiently tightly, and so “fast lanes” have become associated with the goals and plans of the anti-neutrality camp rather than just a potential weakness in a no-longer-relevant pro-neutrality draft regulation.

I mostly blame John Oliver for that.

Although, in point of fact, edge provider preference much like fast lanes exists in the form of zero rating, which is occurring in the mobile broadband space. It's not precisely the same thing, but had much the same issue.

Viewed with a more holistic approach, I understand a lot of those case by case issues, and have a lot less panic about them. It's easy to paint a pile of different events with a broad brush, especially if you already view a group of companies as inherently evil by default. (I get why the net neutrality propaganda campaign has been so successful, it's super easy to convince people to hate Comcast and Verizon, we all already do.)

But for example, the paid fast lane situation... does not currently exist. The whole "you might have to pay for the social media package to see this tweet" nonsense? Not only doesn't actually exist but has no chance of even remotely existing.

Paid fast lanes sort of do exist, but the content providers are the ones paying. This is worse in some ways because it is invisible to consumers but it's better in the sense that consumers don't pay for infrastructure they don't use.

I agree the bigger issue is lack of competition. Why isn't there equal outcry about that?

Comcast has a very slow lane for all BitTorrent traffic (they effectively blocked it back in 2007/2008) until the FCC died them to not do that.

You’re argument is probably false. The fears are real we’ve seen ISPs provide tired bandwidth in the past. Everything except BitTorrent traffic was in the fast lane and BitTorrent traffic was in the slow lane.

FWIW, Comcast's statement clearly indicated it would not throttle or block lawful content. If you're being honest, you should be able to agree that most BitTorrent traffic is not lawful. As of 07/08, I think maybe the most common use of legal BitTorrent was that some Linux distros offered their ISOs over it.

Today this might be a bit more interesting of a case since a lot of software patchers from major protocols now use the BitTorrent protocol to patch software. I believe Star Citizen does, Blizzard games may as well. Windows Update at the very least now uses a similar system, though I don't know if it's specifically BitTorrent based or not.

> FWIW, Comcast's statement clearly indicated it would not throttle or block lawful content. If you're being honest, you should be able to agree that most BitTorrent traffic is not lawful.

Most BitTorrent traffic may not be, but that's a matter of content not protocol. Most of the people physically located in the death row wing of a particular prison may be condemned convicts but if a government says it's not going to kill anyone not duly sentenced to execution and just drops a bomb on that wing, killing mostly condemned criminals, but also a few correctional officers and service workers, the action would be clearly out of line with the commitment.

Sure, but again, the example given is ten years old, when there was almost zero legitimate BitTorrent traffic, and it was unreasonably difficult for ISPs to combat illegal traffic in other ways. (Of course, I suspect net neutrality supporters would be far angrier if Comcast were to inspect the contents of the BitTorrent content to determine if it's legal or not.)

If you're going to throttle a protocol to keep your network optimized for legitimate users, crippling a piracy protocol seems pretty practical, wouldn't you say?

Net neutrality supporters don't generally agree that ISPs have a proper duty to block even illegal content, only that it is not problematic (from an NN perspective, some may have problems in some cases for other reasons) if they do so provided that in doing so they do not block any legal content. If they can't discern illegal content they should just leave it alone; they are generally protected from liability for it specifically because it is not their job to police it.

What do you mean by token offenses?

"Fast lanes" as a general product aren't likely going to be offers on the company home page, they're going to be in the private peering agreements they have with larger companies. This has already happened. [Netflix had to pay Comcast](https://consumerist.com/2014/02/23/netflix-agrees-to-pay-com...), and [Comcast throttled their VoIP competitor Vonage](https://www.wired.com/2006/03/comcast_throttling_vonage_user...). Vonage and Netflix started working again but the why was never made public. It seems likely that paying for fast lanes was involved. These aren't the cold hard facts we'd like to have about what exactly happened, but they're clear violations of Net Neutrality that were resolved ambiguously.

As far as negative effects in general, beyond fast lanes, there's more than a few examples:

* [AT&T blocked FaceTime](https://www.savetheinternet.com/att-facetime) * [Verizon blocked Google Wallet](http://money.cnn.com/2011/12/06/technology/verizon_blocks_go...) * [Verizon throttles video](https://arstechnica.com/tech-policy/2017/07/verizons-throttl...) * And [a lot more](https://www.freepress.net/blog/2017/04/25/net-neutrality-vio...).

The violations are often subtle, and haven't always affected everyone, but the negative effects of the lack of NN are not hypothetical.

As I suggested, if you look at these incidents individually, there's very different perspectives you can glean:

Netflix gets free colocation from ISPs through Netflix Open Connect. (Hilariously, Netflix claims it "offers this for free to ISPs", whereas if you realize it is colocation, it is obviously Netflix should be paying for it.) Net neutrality cripples ISPs' negotiating position with Netflix (since it prevents ISPs from controlling their traffic in other ways), which protects their special treatment, ironically, since only Netflix has the market share to dictate terms to ISPs. If you made a startup video competitor, you'd have to pay for the service net neutrality guarantees Netflix gets for free. ISPs must host Netflix's boxes for free, or get hit by a deluge of upstream Netflix traffic.

FaceTime is an excellent example of a reasonable reason a provider may wish to get involved in content for performance reasons: Everyone switching to video chat would have a massive and immediate impact on that ISP's network traffic. With "unlimited data" models, ISPs don't necessarily get increased revenue to offset massively increased bandwidth use and network congestion. It's understandable why AT&T moved here. I actually prefer a pay-as-you-go data model, because it incentivizes carriers to provide neutral data as fast as possible: The more you use, the more money they make.

Verizon blocking Google Wallet stands in a long line of incidents where good PR has reversed the blame. ;) Google refused to allow competitors' payment systems from using the secure element in phones... a market Google has monopoly control of. Eventually, Verizon's hardline stance on being able to offer a competing service forced Google to rework Wallet to be more fair, and not get an exclusive security feature everyone else was prohibited from using. Of course, Google just moved to kill their competition another way: They bought out Softcard/ISIS and turned it into Android Pay. (And kicked Windows Mobile Softcard users to the curb too, hurting a second competitor in the process! Efficiency!) NFC payment competitors to Android Pay on Android phones like Samsung Pay only exist today thanks to Verizon's hardline stance.