[shubb]

Can anyone think of any advantages to a non-neutral internet?

I can think of a few - Netflix and Skype would work better.

Most likely, we'd be able to pay for priority traffic, just like we pay for a large AWS instance. Non-priority traffic might be cheaper than current bandwidth.

It would be crazy to suggest all sites be forced to use the same size AWS instance...

Some types of traffic are bandwidth sensitive, like video. Others are cost sensitive, like Linux DVD images.

If you think that an end to neutrality will 'ruin the internet', don't you expect consumers to choose ISPs and services that don't do it?

[josho]

> Can anyone think of any advantages to a non-neutral internet?

Yes. In the early days there was a competing networking technology called ATM. It provided quality of service (QoS) aspects in the protocol. So, you could prioritize packets, e.g. protocols affected by latency could be prioritized (e.g. VOIP, gaming), while those that weren't could be given a low priority (FTP, email).

The beauty of ATM was that on a relatively low bandwidth connection you could utilize all of your bandwidth and services like VOIP would still work beautifully. TCP/IP still today struggles with this.

So, as a consumer, I'd be happy to pay for QoS so that my VOIP packets had an expressway on my 2Mbps connection. However, that ship (for many) has sailed. With my largely under utilized 50Mbps connection there is no reason to pay for QoS because we've largely solved latency by throwing bandwidth at the problem.

However, with the approach that the FCC/comcast/et al are taking, I see no benefit.

[Jtsummers]

But QoS at the protocol level is different than non-neutral. Non-neutral is a way for someone, not the user or application developer, to reprioritize content. You're not using Comcast VOIP? Fine, we'll slow down your Facetime chat. Don't want to use Mediacom Streaming Movies? Fine, we'll slow down Netflix.

That's what non-neutral means. QoS is different, that's applications behaving themselves. BitTorrent, for instance, led to the development of uTP (micro transport protocol). One of its nicest features for torrent users is that it will slow itself down now in response to congestion and play nice with other network-using processes on the client side.

Putting this into the underlying connection like ATM did just means that you at least have to pick a default QoS and hopefully applications/systems pick a sensible assignment for the traffic. Rather than the default being to treat every connection as equal.

[josho]

I understand the difference, my point was simply acknowledging that not all traffic is equal, so I have no problem to pay extra to ensure that the traffic that is important to me is given a priority. I do that today by paying for more bandwidth than I need. Alternatively, if bandwidth were a limited resource, then I'd consider paying to shape my traffic (either by paying for a better router to apply QoS rules in my own network or pay my ISP to take care of that on my behalf).

Comcast, however, wants to flip this around, so that even if I have bandwidth to spare they seem to be purposefully slowing down traffic (or under provisioning their own bandwidth) to force the Youtube's, Netflix's of the world to pay more.

TL;DR: all data is not equal; traffic shaping is 'ok' in theory; Comcast is evil, so please god don't let them artificially create slow lanes to force those willing to pay into the fast lanes.

[jonlucc]

The problem is that most places don't have choices. I have only one option (ATT) for ISP and when I move in a couple months, I will have 2 choices (Comcast and ATT).

[lutorm]

I feel so much better about my situation now: I have 3 choices: TWC, dialup, and some guys providing microwave links for $100/month. ;-)

[revelation]

You are burning an artificial technical straw-man. There is no technical capacity limit at play here. Networking technology has leapfrogged the bandwidth available to consumers many times over, and to boot, it is pretty much infinitely scalable.

So when theres no problem transferring all of Netflix, Skype and BitTorrent simultaneously, why slow any of it? Sure, at times hardware fails, fiber is damaged, and ISPs can feel free to prioritize traffic at that time, we certainly have the technology to do that.

But what is certainly not ok is slowing traffic because you are not willing to invest into your connectivity, investing not even enough to actually deliver all of the meagre bandwidth (100 Mbit is 1995 vintage technology, where in the US can you even get that?) you have sold to consumers.

[microtonal]

There is no technical capacity limit at play here.

This. The US market is ripe for disruption or regulation. Sure, if you live remotely in Arizona, it may be hard to get a fat pipe, but it seems that even most cities have outrageous prices.

I currently live in Germany. We have 150MBit downstream, 5 MBit upstream, plus phone and television for 40 Euro per month. When we lived in The Netherlands we were on 130 MBit downstream internet. Even in the stone age (2004) we had 20MBit downstream DSL for 20 Euro per month. Since downloading music and movies was legal in the Netherlands until recently, many families were saturating their connections. Netflix is not that demanding in comparison.

The current situation in the EU shows that it is possible to get high-speed low-cost internet with net neutrality.

[shubb]

Strikes me folks here are really complaining about a broken cable market, made up of local monopolies.

I googled the American cable market a bit, and some of what was described - the cable majors carving up the country so as not to compete, or aggressively blocking new competitors - that stuff sounds like anti-competitive practice.

Writing from a country with a functioning cable market, when I hear 'ISP will charge for X', I think 'Well I will change ISP then'. If you can't do that, I think net neutrality is the least of your problems.

[exelius]

There's no technical capacity limit, that's true -- but there are operational and financial constraints. Replacing a half million routers across the country to remove the previous technical constraints isn't an easy or fast thing to do: the hardware will be obsolete before the implementation is finished.

Think about everything that goes into this: they can't just pick a piece of hardware off of Amazon; they have to review proposals for companies to manufacture and support the hardware over their 10-year lifespan. They then have to train their employees on them, develop migration plans, rollback plans, schedule maintenance windows, etc. These migration plans often involve significant changes to CPE configurations, which also need to be planned, tested, implemented and trained for.

It's a huge infrastructure. It costs a lot of money to make any significant change. And you seem to be confusing Ethernet with a last-mile technology; it's more of a last hundred feet technology. A lot of the effort over the last 10 years has been spent moving the telecom-owned equipment closer to consumer homes so that faster speeds can be obtained over shorter cable runs. As the length of a cable run increases, so does interference and you have to dial down speeds as a result (this is even true of fiber, albeit to a lesser extent).

[revelation]

I'm painfully aware of just what lengths telcos will go to to press the last bit of (downstream) bandwidth out of the taxpayer funded land line infrastructure they have been gifted. Thats what I'm saying: they have no interest in investing. Who will wake them up and tell them that no, you can't bridge another 20 years of progress on fucking bell wire? Do we like send them postcards explaining the Shannon-Hartley theorem?

But the terrible state of last-mile technology in the field isn't even what this is about. Netflix servers and the intermediaries they peer with are not in a shed in Nebraska with data coming in over microwave. Telcos don't invest in the last mile where they would have to create actual infrastructure, they don't invest where high technology rules in the heart of data exchanges all over the world.

(Of course ethernet isn't the relevant benchmark here, but at that time it wasn't just about what you could do over a hundred feet of copper, but also at what speed systems could actually communicate.)

[Jtsummers]

Traffic priority should be set by the users and the application developers, not by the ISP. Streaming video/audio should have a way to signal that they're high priority traffic (because latency is noticeable quality of service absolutely essential). BitTorrent should be able to signal that it's low priority (unless being used for streaming like Popcorn Time) since it's often used in the background as a way to get large files or file sets. HTTP traffic should be able to sit somewhere between the two, with some services announcing themselves as high priority (services that stream media over HTTP or games or whatever) and the rest announcing themselves as "Guarantee me decent latency, I need to be there in a few seconds, but under a second is overkill".

> If you think that an end to neutrality will 'ruin the internet', don't you expect consumers to choose ISPs and services that don't do it?

I have 2 wired ISPs to choose from and several wireless (well, primarily via tethering with a cellphone) ISPs. My apartment has shitty wireless reception so Verizon, Sprint and AT&T are out. That leaves Cox Cable (who repeatedly disconnected me when they meant to disconnect a neighbor, costing me two days of leave to fix their fuckup) and Windstream (a DSL provider). So if both of my providers decide to play the non-neutral game, I, the consumer, am screwed. There are no options for me in your scenario. That's the reality of the ISP situation in the majority of the US.

[CamperBob2]

Can anyone think of any advantages to a non-neutral internet?

This question seems to be equivalent to "Can anyone think of any advantages to a centrally-planned economy?"

Yes, you can think of certain specific advantages, but the opportunity costs are enough to utterly dominate the long-term picture.

[marcosdumay]

> Netflix and Skype would work better.

They'll have a monopoly on "working better", thus thake "better" with a huge grain of salt.

> Most likely, we'd be able to pay for priority traffic

You (the customer of the ISP) can do that today. The difference is that they want to sell priority to the customer, and then only deliver it if the data provider pays it too.

[coldcode]

You forget that the customer would be forced to choose what sites they could access as pay per site. The ISPs (generally a future Comcast monopoly) want to charge both ends to connect. You want Facebook at all - pay $5 per month. Facebook, pay us $1 per customer. You want some random blog, sorry they didn't pay us. Or maybe they will connect you at such a low rate as the site is unusable. The end result is that only huge wealthy companies can be accessed. Everyone else will be so backwater they may as well use smoke signals. The effect of monopoly (at least in the US) will make it impossible for their to be an alternative. I have two choices, AT&T and TWC. Both will do this sort of thing if they can. Then what do I do?

[zacinbusiness]

I wonder if this will have an effect on web design? Less JS, fewer images, no web fonts. I can see a website going back to plain text. I realize that even plain text can be throttled to the point of uselessness, but would it help at all?

[danelectro]

How about an augmented form of plain text, where a SMALL MINORITY of special undisplayed characters invoke formatting options which can be utilized by consoles not much more powerful than dumb terminals. We could call it Hypertext . . . Could even give rise to a "markup language", we could call it HTML . . . Once you start going down that road it could be a slippery slope . . .

[zacinbusiness]

Do you have a real point or are you just being a moron?

[orthecreedence]

> Can anyone think of any advantages to a non-neutral internet?

Yes! People who are always crowing about how the free market will save us all and how regulation is the enemy will start bitching about their slow Netflix speeds. Some of them may even realize what a terrible idea an unregulated free market is.

> don't you expect consumers to choose ISPs and services that don't do it?

Most ISPs have a monopoly in their local area. Perhaps a state-run option would be a good alternative but it would probably eventually get completely weighed down with bureaucracy and red tape.

In other words, the free market can't save us and neither can our government. A healthy mix of the two seems to be a working solution.

[jordanthoms]

Keep in mind though, the ISPs got those monopolies from the governments in those areas. Not exactly free-market.

I don't have a problem with regulating them though - As far as I'm concerned, if you've been exploiting a state-granted monopoly for the last 40 years you don't get to demand a free market when someone might pass regulations less favorable to you.

However, the real fix here is going to be removing the monopolies and getting a competitive marketplace for bandwidth.

[nitrogen]

QoS is perfectly fine if it's set by the user, not the ISP. This is a red.herring anyway, because the Netflix problem has nothing whatsoever to do with prioritization.

[jsz0]

> Can anyone think of any advantages to a non-neutral internet?

It would give ISPs an incentive to spend the large sums of money required to upgrade their networks/infrastructure to offer significantly faster Internet speeds. I very much doubt that any amount of legislation/regulation is going to force ISPs into it otherwise.

[saganus]

How is it going to give them such incentive? If any, I think it will give them more incentive NOT to do so.

Why would they upgrade the infrastructure if they can just rise the prices until the demand for higher speeds either dies down or is enough at that higher price point that it will be cost-effective?

[hamai]

Wouldn't prioritizing packets at routers create slower queues for low priority packets?

Murphy's Law tells me there's a good chance these low priority queues will slow down exponentially at random times. Routing nightmare if you consider DDoS and what not.

[JoeAltmaier]

Not that simple. Queues can manage the latency of packets, not just by priority but by time-stamping packets and sorting them.

Priority is a sucky way to do anything - queues, threads, even email. What we usually want is control over latency.

[AnthonyMouse]

Latency isn't really the problem. Latency has three causes in practice: The speed of light, router packet buffers, and (generally congestion-related) packet loss. The first isn't usually a problem and there is nothing to be done about it anyway. The second and third aren't actually latency issues at all, they're latency as a side effect of a bandwidth shortfall. The only way to actually fix that is to add more bandwidth. The best you can do otherwise is to choose who gets screwed over by the lack of bandwidth, which is exactly the thing you don't want the likes of Comcast doing.

[JoeAltmaier]

Latency is the issue in routers when they categorize traffic, that was the idea. They don't simply prioritize packets; they age them and sort to try and ensure latency constraints. So its more complicated than priority.

[AnthonyMouse]

> Latency is the issue in routers when they categorize traffic, that was the idea.

That's what I'm contesting.

Suppose you have a router which is receiving data to be sent over a 1Gbps link. Over a one hour period there is an average of 600Mbps of streaming video, 400Mbps HTTP traffic and 300Mbps of FTP/BitTorrent/etc. traffic. There is more data than there is link to put it in; you are screwed. The streaming video is going to stutter or degrade to lower quality, web pages will be slow to load, people waiting for downloads to finish will have to wait longer, because there is not enough bandwidth. The latency will also be poor, if router buffers are set too large and create a large queue in front of new packets (bufferbloat), but that's the least of your worries in that situation. Fixing the latency wouldn't clear things up because you still have users trying to play video with a bitrate of 6Mbps through a connection with a 4Mbps throughput. And categorizing the traffic only changes who gets screwed -- if you put downloads at the bottom of the heap then you might make the streaming video and HTTP customers happier but the customer who paid for a 6Mbps connection and would have at least been downloading at 4Mbps is now waiting for a download which getting only 0.2Mbps. That's not a solution, it's triage.

Now suppose you have the same amount of traffic on average but the link is upgraded to 2Gbps. Now there is no packet loss. You might have short bursts where the traffic level exceeds the capacity of the link, but they get smoothed out by the router's buffers, which never get full so the queue length is never more than a few milliseconds. Solving the bandwidth problem solves the latency problem.

The point is, if you're trying to fix the latency that occurs as a result of router buffers getting full or dropping packets in the core of the internet, you're rearranging the deck chairs on the Titanic. The underlying problem is that router buffers are getting full or dropping packets in the core of the internet.