"Every Time You Vote against Net Neutrality, Your ISP Kills a Night Elf"

Ok, so I am totally in love with this report. It's written by Tony Greenberg and Alex Veytsel from a firm called RampRate that provides IT-related market and strategic analysis. I stole the subject line for this post from their report, and while I love cutsey titles more than life itself, the actual analysis in the report is even better than the title. They talk about the effect of the probable demise of net neutrality--the principle that your ISP and backbone net providers should not be allowed to discriminate between packets--on two industries, VOIP and MMOGs. I can't believe I didn't think of this myself, but net neutrality is all about jitter and latency and guess who cares a lot about these issues:

Companies like Blizzard Entertainment, Electronic Arts, Sony Online Entertainment, and countless others, have built a business on the fundamental assumption of relatively low latency bandwidth being available to large numbers of consumers. Furthermore, a large — even overwhelming — portion of the value of these offerings comes from their “network effects” — the tendency for the game to become more enjoyable and valuable as larger number of players joins the gaming network. With the permanent barriers that the removal of net neutrality will erect for these uses, the worst-case scenario includes three waves of change:

• One or more mainstream ISPs will introduce excessive lag that will effectively prohibit their users from participating in online games. The move will not be aimed at restricting usage per se, but rather to extract a fee from the game operator. However, as the Cablevision and YES dispute of 2002 showed us4, a fee disagreement between a cable company and content provider can effectively lock out the use of a popular service for over a year;
• As online gaming guilds, clans, and partners disappear into the rifts created in the Internet fabric, players that derive value from the community of the game rather than the playing experience per se will drop off. This vicious cycle of scarcity of users will lead to diminished enjoyment for existing users which will lead to still fewer users, until more games follow Asheron’s Call to oblivion;
• Hardcore users will write strongly worded messages to their ISPs, who will classify them as unreasonable malcontents using more than their share of bandwidth.

There is lots more. Honestly I just want to excerpt the entire report here, but obviously I won't. But go out and read this. It's really good.

Comments

[Ren Reynolds]

1.

Erm, I don’t get it. Sorry. If my ISP wants to differentially charge for different classes of service why should they not be able to do that?

In fact, this model has operated for years - it’s generally been (all other things being equal) on the number of routes that are advertised. More routes, the more expensive the connection rather than packet labels but at least one (no, not all) of the principles involved is the same. To get decent VoIP performance you have to engineer things differently and the way that it impact capacity can have quite an impact on overall performance, so why not get VoIP users to pay more for the packet prioritisation that they really need to get a good service. One could almost argue that they should pay a premium so that everyone else gets a good service.

Can some say why the report uses word like ‘evil’ when all we are talking about is segmentation of an ISPs product portfolio.

[Dmitri Williams]

2.

I'm skeptical that the loss of net neutrality would kill off MMOs because of bandwidth issues. What is essentially at point here is the introduction of market mechanisms into ISP service. We've seen a tiny bit of this with cost-per-download phone services and even a couple tiered broadband packages from big ISPs. But they are pretty crude--about the level of three tiers of cable packaging.

So consider cable as a parallel. You'd expect them to disagraggate their whole bundle into a la cart services to better exploit niches and match supply with demand. People would pay for what they use and nothing else. Yet they don't. The head lobbyist of the cable industry once told me (with a straight face) that it was technically impossible to go a la cart and that doing so would rob consumers. That was 6 years ago, but they continue the party line (see: http://www.ncta.com/IssueBrief.aspx?contentId=15)

Why the big lie? Because cable makes more money with an inefficient matching of supply and demand. They make everyone pay for the big hits (ESPN), even if all they want is Home & Garden. Disaggregating would let my mother in law pay $5 a month instead of $40. Efficient, sure, and a happy customer, but a loss of $35, so no dice.

Will ISPs go this route? As is, they have different bundling issues since there is no ESPN--just more or less bandwidth. But if you erase net neutrality, you allow for some services, providers, web sites, MMOs, etc. to become big ticket items. You reintroduce the bundling/disaggregation dynamic.

The report authors figure that means that we'll now see efficiency and charges applied for the heaviest users in accordance with their bandwidth. But the history and practices of the cable industry suggest that inefficient bundling is very much in the provider's best interests. Disaggregating and transparency in the system adds too much information to the transaction. That information would be used by consumers to the ISP's detriment. I would predict that they will stick with big dumb tiers of service, as aggregated as possible.

The only reason they wouldn't would be if they haven't thought it through like NACTA has.

BTW, this dynamic is the reason I'm less hot on "protecting" net neutrality. It sounds great for consumer welfare, but the devil is in the details of the implemenation, i.e. the rules and enforcement that come along with it. I'm personally more concerned with the privacy issues that comes from a loss of neutrality.

[Cory Ondrejka]

3.

This was actually discussed briefly at AGC on the panel that Jerry ran, which is here: http://www.secondcast.com/?p=8

I have not read the report, but I have my own reasons for disagreeing with both Ren and Dmitri.

Business Reasons
if Microsoft or Electronic Arts is able to pay for differential routing, they can create an incredibly anti-competitive environment for online games of all kinds. EA, in particular, has had some false starts in the online space, so wouldn't they love to pay to ensure that their packets get their first?

Technical Reasons
Apologies to Ren, but this is not the same as paying for multiple backbone connections. The loss of net neutrality allows the someone to pay to adversely impact other people's packets. If I buy N routes, it simply makes it easier for me to efficiently route around net problems, not to generate artificial inefficiencies.

More broadly, the idea of making routing decisions in the middle tof the net -- as opposed to the edge -- shows a profound lack of understanding of how the tubes work. For any given packet, you either have effectively infinite bandwidth -- that is to say, the packet is not dropped -- or you have zero bandwidth -- the packet is dropped.

So, if we're going to prioritize your packets over mine, we have two options. Option #1: if our packets arrive at the router at the same moment, we drop mine. Except that Option #1 doesn't actually happen. Instead, we have Option #2: if we start saturating a link, we start preemptively dropping my packets. Except that doesn't give you the quality you want since you'll still run out of bandwidth when you surge, so we instead are forced to Option #3: we always throttle my packets leaving headroom on the links so that when you surge your traffic, it's OK. Thus, you pay to throttle my traffic. Thus, you pay to reduce the efficiency of the network. Dumb.

The place to do differential routing -- if you really wanted to -- is at the last mile. Most of us have either DSL or cable modems coming into our homes. If those modems were smarter, you could have them make decisions about TCP-window and train sizes to differentially throttle different types of packets. Of course, this is a less palatable options, since it is "Hello, customer? We're going to force you to purchase a modem that doesn't work as well for the things you use!" -situation which encourages other ISPs not to use it.

Net neutrality is important. I agree that it is a complicated and (often) more nuanced issue than either side portrays it as, but it is still very important.

[ren reynolds]

4.

Cory > Apologies to Ren, but this is not the same as paying for multiple backbone connection

Apology accepted, but I was not talking about multiple backbone connections and I was not saying that buying on the basis of packet labeling (or what ever mechanism is used to differentiate payload type) was just the same, I said “at least one (no, not all) of the principles involved is the same”.

To put it another way, I was talking about the different between buying a connection from a Tier 1 ISP vs say a Tier 3. At the geek level, there is a difference in the number of routes that are advertised on BGP, this translates to a, on average and all other things being equal, better performance. The technicalities aside, what I was saying was that ISPs do differentiate in terms of class of service and have done for ages, this is the common principle to which I alluded. Yes using different pack types is a different application of this principle, but at one level there is commonality.

We then get into core vs edge packet differentiation. I just don’t know, in practice, either would work on an internet scale, but I think the argument might be even more complex.

Cory > So, if we're going to prioritize your packets over mine, we have two options. Option #1: if our packets arrive at the router at the same moment, we drop mine.

This assumes that both packets arrive at the same router. What if we have, as it were, a fast lane and a slow lane. This sounds a bit bonkers and as if we are adding lanes into the free wheeling internet where traffic goes where it will. But it doesn’t. The internet is highly shaped and traffic runs across the backbone (or at least the individual ASs that make up the backbone) via effectively multiple VPNs made up of wavelengths on optical routers or other tech such as Frame PVCs or similar, why not have the ‘red’ ones for VoIP and all the rest for everyone else’s data. As I say, I don’t know that that would be better, but all kinds of counter intuitive things tend to make flow models flow faster, and separating and then fanatically differentiating some services might be one.

[wowgold]

5.

What is a full-bodied son?What again is to flavor son?What just is quite tasty==the wow let you enjoy full flavorgame

[Dmitri Williams]

6.

As I said, the devil is in the details. If the replacement for net neutrality is the ability for one firm's bits to supercede another's, that is a different system than one which allows merely for more traffic on the user's terminal (i.e. tiered bandwidth at the last mile). The system that allows superceding is, indeed, very bad. It fairly begs for vertically integrated firms to squeeze their competitors.

Still, the argument against the current one-size fits all system is an easy one to make: light users are subsidizing heavy users. Your grandma pays 3 times what she should so you can play WoW at your current costs.

The arguments get wonky (and manipulable) when we mix pure bandwidth filters with ones that filter particular ISPs. I think the authors are doing that, so I'm posting in that context and I think still agreeing in substance with Cory.

[ombwah]

7.

The argument that "Your grandma is subsidizing an MMO player" is crap. Just because you don't use the full capabilities of a system you subscribe to is no excuse for you to bitch that others do. You pay for a system that *allows* for a bandwidth usage of *up to* X bits per second, just deciding you don't want to use that capability isn't a logical or realistic reason to ask for lower rates.

The absence of net neutrality is a license for extortion. F**k telcos right in the tight place. If they want a piece of the MMO pie, then they need to CREATE content, not just restrict the already extant avenue for delivery.

[Cael]

8.

Oh please, not this drivel again.

It's really very simple. The prioritised streaming is all about Video on Demand. With prioritising that streaming, your video experience will be roughly analogous to YouTube, ie crap.

In order to keep up with dataflow, telcos have to make significant investments in core infrastructure. Is Blizzard going to pay for that? Is Skype? Is Google?

No. AT&T and BT will be paying for that and they know it. If they can charge a premium to deliver video over IP, it may deferr some of those costs.

At the backbone level, there is no efficient way of differentiating between a Second Life packet and any other sequential UDP data. You think ISPs want to start screwing up services like MSN? If you do, you're crazy. All they can do is uprate or downrate some very obvious pipe hogs like VoIP (actually, since that interferes with the telcos' core business model, i expect this to happen) and VoD.

Think about it this way - if started differentiating any further, they'd lose legal carrier status and slip into the content-provider bracket, which would make them legally responsible for all the pr0n your kids are downloading.

In other words, that won't happen.

This "net neutrality" is the biggest steaming pile of marketing horseshit i've ever seen supposedly intelligent people fall for - most of those commenting have zero idea about telcos actually do and how they actually do it, or even about how packet transfers actually work!

I strongly suggest that all those commenting above go read some RFCs.

[Ola Fosheim Grøstad]

9.

Cael:This "net neutrality" is the biggest steaming pile of marketing horseshit i've ever seen supposedly intelligent people fall for - most of those commenting have zero idea about telcos actually do and how they actually do it, or even about how packet transfers actually work!

Big words...
This is what is already happening (at a small scale):

1. Users think they are paying for a N Mbps connection, but they actually only get "up to" N Mbps... So ISPs are already screwing their customers by false marketing.

2. Big service provider X pay ISPs to get priority with a dedicated connection, and this is hidden from customers who never wanted X to squeeze out other traffic. So he doesn't switch ISP.

=> all non-X providers are forced to getting a slower QoS, and the customer has no knowledge of it. He just thinks the global network "is slow", or that non-X providers are low quality providers... So he doesn't switch ISP.

Most ISPs are already slowing down peer-to-peer services (torrent etc). Hacker-type customers are already getting pissed off.

ISPs might care enough about the big games like WoW to up their QoS to an acceptable level, but they won't care about small games.

Note: the slowing down (dropping packets) doesn't necessarily happen at the backbone, but might very well happen close to the end-user.

Big ISPs are in a position where they can extort money from medium-sized service providers. There is no doubt about it.

[Andrew Crystall]

10.

Dmitri Williams, complete and utter rubbish.

Most games, especially MMO's, are NOT that demanding in bandwidth terms. What matters for them is quite simply ping/packet loss for the bandwidth they DO need.

And that's why the extortion tactics will start there.

Cael, and they are "requests". Right

"You think ISPs want to start screwing up services like MSN? If you do, you're crazy."

Nope, in areas where they have an effective monopoly, that's PRECISELY what I expect them to do

[Peter Clay]

11.

"At the backbone level, there is no efficient way of differentiating between a Second Life packet and any other sequential UDP data"

Rubbish - port number and IP address allow for all the discrimination you need to start extorting money from Second Life or their customers.

Cory Ondrejka's explanation of "I pay to throttle your traffic" is the best one I've heard so far.

I don't think it's "evil", but it would turn my current transparent internet pricing into a morass of plans and fees like cell phones, with a second tier of fees on the other end ready to crush popular internet applications that depend on either high bandwidth or low latency. The real secret is not that Voip or Vod won't work without additional infrastructure - they blatently already work well enough for most people (teamspeak, skype, youtube etc) that they are threatening telco and cable incumbents, and the anti-neutrality measures are an attempt by the incumbents to either strangle them or get a cut of the money.

[Ola Fosheim Grøstad]

12.

Rubbish - port number and IP address allow for all the discrimination you need to start extorting money from Second Life or their customers.

Uhm, he said "efficiently". It is much cheaper to filter content which has paid for high QoS and throttle all other demanding traffic that most customers won't cancel subscriptions over.

ISPs are interested in throttling traffic to the backbone (which they don't own) in order to save money, obviously, even if they have plenty of headroom. I doubt they will bother to differentiate between filesharing and online game downloads. They will throttle when they see a surge in traffic from a particular node.

Besides, Cael is right. You can hide your IP address (pooling it with thousands of other services) and use port 80. Stateful packet inspection isn't free.

The real secret is not that Voip or Vod won't work without additional infrastructure - they blatently already work well enough for most people

Most people just use e-mail, web and the occacional lores video. If ALL of them started to saturate their connections with highres video then you sure need solid infrastructure. BUT this should be cached throughout the network anyway, since most video is cachable. There is no problem in charging extra for this. Disks cost money. The problem comes when you establish a regime in which you have to pay for decent QoS => LIVE sports hires video squeezing out all other traffic on the ISP network.

VoIP can't be a problem for the backbone as it just moves the traffic from a "non-internet" channel to internet-channels...

[Brask Mumei]

13.

What Peter said. VoiP works just fine for me with the current infrastucture. So, incidentally, does VoD. So when you tell me that the Telcos need to remove my ability to do these things in order to upgrade my connection so I'll be able to do them with *their* services, I am unimpressed.

[Peter Clay]

14.

most video is cachable

In the US, there would be a serious problem implementing this as quite a lot of the video traffic is copyright-infringing...

You can hide your IP address (pooling it with thousands of other services) and use port 80

How, do I hide my IP address (allocated to me by my ISP using DHCP) from my ISP?

How do I connect to EVE Online (server running on port 26000 I believe) using port 80, without going through a lag-inducing relay?

(Does anyone remember the details of the recent inter-ISP peering dispute involving Level 3 which left lots of people unable to connect to bits of the internet?)

[Ola Fosheim Grøstad]

15.

Ah Peter, I thought we were discussing the future dynamics of the net, not the net-as-is. The ISPs can do a lot of bullying now, but my point was that cooperating serviceproviders can "hide in the fog" if it really matters. Customers can't do much, except switch to a different ISP.

[Dmitri Williams]

16.

My utter rubish is saying that some users (gamers, VoD, it doesn't matter) use substantially more bandwidth than others, yet pay the same rate. And pointing out that the arguments about which ISP gets filtered/leveraged will become confused with bandwidth, as it has already in a short thread.

I'm also offering the thought that ISPs would do well to follow cable's model for their own good. The rents that come from the current inefficiencies are captured by the ISP, not delivered back to the consumer as lower prices for lower usage like a common utility.

[Ola Fosheim Grøstad]

17.

Dmitri, I believe the real problem for the ISPs is what happens at peak hours. So low and high usage matters less..? WHEN is more important. And who wants a connection that is only available between 2300-1500? Cable is different because the main cost is content, not delivery. Or?

[Dmitri Williams]

18.

There is always an economic answer to higher traffic. You charge more for those who use it: a tax for the driver who drives at rush hour, drives in a certain area, or who drives more period (London does this, as do many transit boards). One problem is that people feel entitled to a flat rate, even if they are actually consuming more resources at someone else's cost.

The company regulating that traffic then has to decide if it's worth educating the consumer about the system and how a "fairer" system will impact them. That's such a tough nut to crack that it's usually seen as better to just build more freeway capacity and charge everyone the average operating cost. London can't do that, but ISPs can.

The point is that the current system has winners and losers already. The winners are the ISPs and the heavy users. The losers are the light users.

Cable maintains such a system by charging everyone the bundle price. Yes, there are tiered packages, but it's in their best interests to not let the bundles get too transparent (a la cart).

Note that I'm talking here about bit capacity, not who's bits. That is a different issue, and Cory has already nailed it.

[Ola Fosheim Grøstad]

19.

I don't think Cory has nailed it, he barely touched it... Anyway, I don't really think the light users are loosing out as much as you think as they most likely use the net during peak hours. Heavy users are throttled during peak-hours and probably use the net heavily during off-hours when it doesn't hurt anyone. Many of them also pay for higher Mbps caps (ADSL2), thus pay more for the bandwidth they don't get during congestion. Let's not forget that there are different "types of bits", filesharing isn't hurt much by limited throtteling. With throttling one might claim that filesharing is simply filling out the empty gaps between "real" traffic. Why should they pay extra for that? They are simply using the road when nobody has any real use for it...

[Ola Fosheim Grøstad]

20.

Hold on, Dmitri, maybe I misinterpret. If you say that people who want very high QoS (high relibility, guaranteed responsetimes) should pay a premium, then I don't disagree. That's not the same as bandwidth though. ;-)

There is nothing unreasonable about charging the first-person-shooter guys for getting priority over your grandma. And yeah, the bundle is the way to go.

[Cael]

21.

Peter Clay: How, do I hide my IP address (allocated to me by my ISP using DHCP) from my ISP?

How do I connect to EVE Online (server running on port 26000 I believe) using port 80, without going through a lag-inducing relay?

Both of these can be managed with minimal delay, a winXP or linux box running IIS or Apache and four terminated lines of perl.

[Eric Random]

22.

Ramprate's customers are content-providers like Yahoo!, Microsoft, CBS, and Sony. The very customers that will be negatively impacted, and have less money to spend on middlemen like Ramprate, if Net Neutrality does not pass.

If ISP's are required to treat all traffic equally, all rate-sensitive services will suffer, and there won't be a lot the ISP's will be able to do about it, considering the amount of investment it would take to get the WHOLE Internet to the level required for rate-sensitive traffic.

Ramprate realizes that rate-sensitive traffic like real-time gaming will be affected, and they want to blame the ISP first, before you start feeling the affects of what net neutrality really will bring upon the average consumer.

ISP's are not talking about QOS in the core. They are talking about a whole new Internet. ISP's can see the end of the tunnel for the current technology, and to make the next leap, they need to build anew.

This is not the first time ISP's have have needed to do this, if you consider many of them used to exist on analog, packet switched networks. Even then, they built a brand new network, and plugged it into the old one. Even then, the big players paid for it first. This was the day when the content provider actually was the network provider, like Compuserve. We've been through this before.

The problem is now, the big players don't want to pay for it. They want to put the full burden on the ISP, and the end-user will pay for it. Not only out of their pockets, but through degraded services as well as a slowdown of technological advancement. It will be the unintended symptoms of a poorly legislated system rather than any intentional act. Ramprate just had the forethought to start the finger-pointing early.

Back then, in the early 90's, if you tried to use propaganda like Ramprate, people would have no clue what you are talking about as they barely even knew what the Internet was. Today, they still don't know, but they have a clue, and that's all you need for an opinion.

[The Internet]

23.

I used to be of the opinion that the telcos should be able to charge as they wish for the use of their network.

Then I came across this: http://muniwireless.com/community/1023

Which more or less states that they already got about 200b$ from various public sources, to lay out all kinds of stuff (fiber to every house, etc) and pocketed the money and never delievered.

So yes, they can and should eat the costs of building their new network.

[Ola Fosheim Grøstad]

24.

Eric Random: ISP's can see the end of the tunnel for the current technology, and to make the next leap, they need to build anew.

NEED? Where does this NEED come from?

[Andrew Crystall]

25.

Ola Fosheim Grøstad;

No, stateful inspection isn't "free". But I suggest you take a look at the shares of the companies who make the kind of devices which can perform the inspection.

[Ola Fosheim Grøstad]

26.

Here is a recent case of ISP extortion in Norway:

http://www.eirikso.com/2006/10/03/goodbye-network-neutrality-in-norway/

[Eric Random]

27.

Where does the need come from? Should we not advance our technology in the U.S?

To provide end-to-end quality-of-service for the rate-sensitive and delay-sensitive services of the near and foreseable future, our Internet has to change. Best effort works fine with quality video the size of a business card, single channel audio links, and our low-rate game communications. You want to move into live HD-quality IPTV, high-res video teleconferencing, International aggregate monitoring, and high-responsive gaming, you have to go beyond a best-effort architecture. Because of the sustained bursting nature of these protocols, it can cause havoc in the current architecture. Even now, ISP's watch a link go from infrequent 30% peaks to sustained 98% peak utilization overnight, because of the rapid success of a killer web app that relies on these types of protocols. Even now, they make overnight topology changes to deal with known traffic explosions like Victoria Secret streams. Not to mention different applications have different peak hours, which regulate maintenance windows, which are different in each time zone, but can effect end-to-end multi-timezone communication.

What ISP's are asking for is a return to a committed rate structure. This is how it used to exist before the mid-nineties for content providers, but as ISP's learned more about traffic behaviour they were able to charge flat-rates because they understood how the probabalistic nature of asynchronous traffic effected the network. That behaviour is out-the-door with the synchronistic nature of new applications. The world has changed, and ISP's have to change with it. What content-providers are trying to do is keep them the same to protect their profits without really understanding, or perhaps, caring, considering the International nature of these corporations, about the state of communications in the United States.

You want best effort, it costs this much. You want committed rate or delay based service levels, it will cost this much, because that is more costly to provide. Some large ISP's have to do costly rapid build-outs just to relieve traffic pressures caused by a handful of content providers. That is why ISP's are starting to get concerned with what the Verizon CEO has, confusedly, called "free lunch" content providers. What about the little guy? ISP's are too busy trying to maintain service levels for the little guy while the big content providers take massive dumps on the network.

[Ola Fosheim Grøstad]

28.

Eric:Should we not advance our technology in the U.S?

Not if it means that bloated advertising is sent at a higher priority than 911 voip calls... I don't mind them charging for QoS (perceived performance). I do mind that they set up an near monopoly infrastructure where cost of entry is most likely going to be uneven. Charge for a particular Quality of Service, fine, but make this option available at the packet level as well for the same price (preferably regulated with a max price).

Eric: You want to move into live HD-quality IPTV, high-res video teleconferencing, International aggregate monitoring, and high-responsive gaming, you have to go beyond a best-effort architecture. Because of the sustained bursting nature of these protocols, it can cause havoc in the current architecture

Well, telco's did bet on people wanting to surf the web on their cellphones. Most people didn't... So now they struggle with finding ways to make people use all that surplus bandwidth (MMS etc). Case is, people don't really need all that tech, for the sake of tech. Telcos do, not people. Most people use regular calls and SMS, the rest is fluff.

Gaming doesn't need more than simple QoS for critical packets. Highres teleconferencing is only for business use, sure that and telepresence surgery are two valid applications where you need good QoS or even hard QoS, but that isn't really what Bell has in mind. What you'll get from the two-tier solution is flashy mpeg4-like advertising based crap. Not worth sacrificing net neutrality over. HD-quality video should be disted during off-hours and streamed from a server colocated at the local IPS! How many movies can you fit onto a petabyte diskarray? A lot, in fact, most movies made to date. What's the point in sending the same GBs of video down the same tube thousands of times per hour? Only telcos benefit from this.

Granted, I buy your argument about the current network depending on the well-behaving TCP, and that other protocols create problems. However, I trust that the current surge in research on router-algorithms and adaptive strategies will crack that nut, eventually. So, I'd rather see a temporary slow down during peak hours than end up with a permanently slow Internet where some big providers deliver flashy content in their little monopoly pool. I'd also like to see legislation that force ISPs to cooperate on creating effective packet-level QoS at fixed rate prices. (yesyes, there are technical challenges).

To be specific: I'd like to pay for a 1Mbps nonQoS 64kbps QoS subscription.

[Tony Greenberg]

29.

First of all, wow. Never thought we could spur such a great discussion. It's truly our honor to mix it up with the thought leadership that seems to breathe through TerraNova. Just wanted to do a quick run-down of points from various posts and our take:

1) Ren: obviously calling someone evil is a rhetorical device, but there is an element of destructiveness that accompanies a lot of the packet prioritization agenda (or the strawman created by the net neutrality agenda, depending on which side you’re on). It’s not about building better services, but rather getting someone else to lose $1 so you can make $.01 – a less-than-zero sum game that’s counterproductive economically. The rhetoric is often about “gee, Google / YouTube / Vonage is making money without paying us a tax” which seems envious and anything but constructive.

2) Dmitri: agreed that the devil is in the details and cures can be worse than the disease, etc. Our piece is mostly about how messing with the status quo can have unintended (or intended but harmful to the majority) consequences and that one audience particularly – gaming companies and their clients -- should be striking a note of conservatism in this discussion because they have relatively little to gain and much to lose from change. But more directly to your point, I think we agree in the core of the argument: a bundle predicated on business parameters (which provider, which service) is not something an ISP should sell – they should bundle based on technical / traffic characteristics (how much latency, throughput, jitter is ok), and let the consumer decide the actual traffic they want to put through that pipe. And there is as much business / core competency argument in there as a “let’s regulate” argument, if not more. So there’s no problems with getting grandma on a different rate plan – but there is a problem with Game A working great and Game B not working at all based on which content provider gave the ISP a bribe.

3) Cael: without getting into the practicality argument too deep, most ISPs have found ways to filter some traffic at a fairly low level – if nothing else, then to black hole DDoS packets. In this case, as Peter pointed out, the only parameter that’s really needed is the origin / destination IP address – if it’s on the “good” list, send it on the direct route; if it’s on the “content providers who are overdue with their protection money” list, send it into a black hole or some detour via a congested pipe you’re renting out from a low cost provider. Obviously this is on the paranoid side of the potential outcome, but I don’t see why parasite-killing tools being developed for unfriendly traffic like DDoS (or spam) can’t be turned around to hunt services that are merely disliked.

4) Eric: Although not all consumer ISPs are also business ISPs, a good portion of them are. So there is a business relationship there and a balance of power that we deal with on a daily basis. That relationship has multi-million dollar contracts in place that actually do fund expansions of the backbone and better switches / routers and move the ISPs to a more stable, scalable network. As long as that relationship stays more or less equal, everyone does well. When any ISP with a large consumer reach has the nuclear option of killing a content service in their hands, the content provider becomes beholden to the ISPs that own the eyeballs, and you wind up having the dysfunctional ecosystem that we’ve seen in mobile. Your position is that the balance of power is all on the content provider’s side, but in our experience, there is more than enough payment and more than enough keys to that content provider’s heart and wallet for ISPs to get their fair share. What we don’t want is for them to get a bigger share not based on the services they provide, but on the services they take away.

[Ola Fosheim Grøstad]

30.

Greenberg: Obviously this is on the paranoid side of the potential outcome, but I don’t see why parasite-killing tools being developed for unfriendly traffic like DDoS (or spam) can’t be turned around to hunt services that are merely disliked.

Sure, they can kill off services based on statistical measures... If not and if the game is colocated in a big computing centre hosted by Sun or IBM, how will they know the IP address of a particular service? They can't if the infrastructure is set up to prevent it.

Besides, I doubt a company like Bell would like to get into another anti-trust... Which is why they want two tiers... Then they can just let the cheap tier detoriate. They don't have to ruin it, they can just not-fix-it...

[EdMcGon]

31.

I have a simple question: Is the Internet broken?

Certainly, there are peak times when everyone hops on and we all experience lag to different degrees. Sure, we would all like more bandwidth, so we can do more and do it faster.

But is there something wrong with the Internet now? I mean REALLY wrong? If so, I am missing it.

So why do we NEED net neutrality? As the old saying goes, if it ain't broke, don't fix it.

[Prokofy Neva]

32.

With all the agitation about preserving net neutrality, I never see any discussions about what the consumer would have to pay if we didn't have net neutrality. Can anyone point me to any actual cost analysis? Would it be $49.99 a month instead of $29.99 or whatever for your ISP if you also had an ISP that delivered all your packets?

Not knowing a thing about all this, when I used to try to upload my families to The Sims offline, and never got the uploaders to work for the family albums, I switched providers, because the route tracers on the first provider were constantly showing problems with the packets being delivered. Then I solved my problems for a time before it broke again.

The layman would be helped in this discussion if we could understand WHY a packet is dropped, or must be dropped, what is in the packet that makes it droppable.

What struck me about the article is this:

"Unlike video and voice, where mainstream ISPs tend to have at least some competency depending on their heritage (cable vs. DSL), most access providers have no idea about what it takes to create and maintain a viable value proposition in the world of gaming, especially online, interactive gaming. To date, not a single top gaming company has been owned or funded by an ISP – and with good reason. The first tangible manifestations of an ISP’s gaming strategy9 are little more than casual gamer portals, dozens of which are already on the market10 with marginal branding and success rates"

What this suggests to me is that at some point, big media corporations will figure out they need to buy themselves telecoms and work ISPs as well as online communities; there's no readily identifiable reason for why telephone companies, which have always been seen as municipal services and funded by governments in many countries, should be at some walled-off place in the communications spectrum, and commercial games relying on subscriptions at the other end of the pipeline should be segregated. It's all about the pipeline. The pipeline *is* the conversation and the content and the game, the medium *is* the message.

[Ola Fosheim Grøstad]

33.

EdMcGon:But is there something wrong with the Internet now? I mean REALLY wrong? If so, I am missing it.

No. Not yet.

So why do we NEED net neutrality? As the old saying goes, if it ain't broke, don't fix it.

We have net neutrality. We don't want to loose it... The internet as is, is a best-effort network. I.e. datagrams (packets) will be delievered if possible, and dropped if not. It was considered cheaper/easier to have more bandwidth available than needed, than implementing advanced "negotiation" of what datagrams to send/drop. Even though IP datagrams contain a field that tells routers how important that datagram is, you just don't know if a router/switch will heed that or not.

Telcos now say: we give up increasing the bandwidth of the Internet to avoid congestion. Instead we create a "separate" network where we add more "unused" bandwidth. However, the only likely way to get people to pay for this "separate" network is to ensure that the current Internet is performing worse than the new one. "Worse" as is in: making sure that major applications are annoying to use. With this model telcos will have little incentive to fix the problems in the current Internet, and the new "separate" network will be a barrier to entry for smaller operators who'd like to fix the current network.

[Andrew Crystall]

34.

Eric Random,

So you don't mind paying $40 base connection, %10 for bit torrent, $30 for teamspeak and $20 for games?

Interesting. Because that IS where it's heading.

I have no problem with premium deals. I do have a problem with selectively degraded services. If a business wants premium-quality video streaming for conferencing, then yea they can PAY for it.

Plus, basically, you realise that a completely non-neutral structure won't be able to support very many ISP's, right? And when there's a lock in many areas and/or effective price fixing agreements, the costs then rise.

[ren reynolds]

35.

Prok > The layman would be helped in this discussion if we could understand WHY a packet is dropped, or must be dropped, what is in the packet that makes it droppable.

Standard connections tend to be ‘contented’ it’s like over booking an airline and packet drops are basically on first come first served, so there is nothing about the packet, just when it happens to arrive in respect of other packets i.e. if the port happens to be full at the moment it arrives. TCP/IP take care of re-sending, up to a limit, what you see is a slow connection as a lot less are getting through than you thought. The economics behind this is that it’s not efficient to have a port of size 10 X for 10 customers of X and it’s highly unlikely that all 10 will use X at the same time, problem is that you might get sold a, in reality access to a 2X port at the other end, which is what you paid for, just so long as no more than one other person is using it. Of coruse you can buy a 10X connection that 'really' is 10 X (at least at the port you talk to, if you are willing to pay (it's yet another way services are currently differentiated on quality but just not generally in the consumer market).

In the non-neutral things we talk about here, I think we are assuming that the network and various applications talk to each other and add labels to packets so that they are identified as say a ‘web’ packet or a ‘voip’ packet. Slightly technically this is using agreed bit’s in the packet headers.

You then apply the same logic as above but you assume that you bandwidth X is made up of say 20% a, 20% b, 60% c, where a b and c are some type of packet, the you just play the game above. But of course you can then start to charge for the size of a and say that ‘a’ packets will push aside b’s and c’s.

All, please free to correct me, my IP is a bit rusty.

[WOW]

36.

Domestic Powerleveling service supplies the lower price for gold sale and collects the MF,OF gold. We will create the first WOW territory and make the furthest WOW career for you on our website.game

[rikomatic]

37.

Here's my analysis of the new Snowe-Dorgan bill on Net Neutrality and its possible effects on gamers and virtual worlders:

http://www.rikomatic.com/blog/2007/01/what_the_dorgan.html