Asymmetrical Information

Wonders never cease!

Private industry is asking for government regulation in order to allow for greater diversified business opportunities? Are they ice skating in Hell as well?

It sounds like carriers are trying to create a classic price discrimination scheme that will differentiate customers, rather than service, based on a willingness to pay.

In addition to differentiating among each other, the carriers will surely differentiate their own service lines, with a premium package for corporate and other inelastic customers, and lower levels of service for other customers. It's not hard to imagine that one development will be to make internet service worse by adding dis-features to induce people into the premium services. this would work much like making airport services less convenient to induce people into "VIP" type services.

If this goes into effect, buy as much Cisco stock as you possibly can, because Cisco will be the big winner. This scheme will require a massive overhaul of the Internet infrastructure.

The entire idea of Level of Service agreements for Internet service is absurd. No content provider can control who or where their end-user is, and therefore has no idea how many networks must be traversed to service the end-user. If I'm on dial-up over a scratchy connection, any LOS between video.google.com and Bellsouth goes out the window. Does anyone owe anyone else a refund then?

Also, there is no accessible universal concept of priority, quality of service, or bandwidth throttling on the many physical layers that carry the Internet, nor is there in TCP/IP. These are all constructs that must either be bolted onto IP, or already exist in the TCP or UDP (and other network protocols) above IP (see 802.1 P,Q for attempts to put this into IP).

If I want to watch a video over the Internet it's currently delivered an a "best effort" basis, just like most package shippers. That's the only way it can be until my XP box, my linksys box, my cable modem, the cable company's entire routing infrastructure, the backbone providers routers and my content provider all agree to accept my box's and my content provider's agreement on the priority these packets must have over all other traffic on the network.

That's these guys' dirty little secret. You can't do that in TCP/IP, but you could in ATM and you could bill for it too. They (the telcos) are excellent at circuit-switched systems, but packet-based data systems don't take advantage of that expertise, nor the incredible capital expense of building a circuit-switched networks (ATM switches run 2X-10X the price of IP switches).

Hah, "tagging content" just seems so easy. What's to stop us from all "tagging" our packets so they float around as high priority, too? How are they going to recognize their own content from mine without added filtering and added time? Implementation will be interesting...

Just as a contrarian opinion: what is so wrong with allowing a company to offer a differentiated service instead of lowest common denominator? There are a lot of technical issues involved, but the new MPLS equipment (routers and switches) allows a tagging for higher levels of service: defined bandwith, latency, and jitter. If some services require this: such as real time video transmissions, why not allow companies to offer this type of differentiated service to those clients where these specs are crucial, where the rest of us can still access email and web-sites on a best effort basis, as we do now? Remember, bandwidth is not the sole issue, the latency (delay) and latency variation ("jitter") are key attributes that most of us don't deal with often, but are critical to applications such as VoIP and video.

I was going to say roughly what larrydj said, only worse. Thanks Larry.

What effect would this have, if any on Spam?

resigned:

"What effect would this have, if any on Spam?"

None.

I think it's the spammers that will reveal the entire endeavor as pointless. Any technical constraint can be circumvented (witness DeCSS), and the spammers have a huge incentive to figure out how to get their traffic tagged as maximum priority, or at least avoid being tagged as junk packets. And if the network is flooded with mal-tagged traffic, the providers won't be able to back up their contracts and no one will buy.

The problem is not "tiered traffic will destroy the internet"; it's "legislators are voting on yet another bill about an issue they don't understand".

You don't understand. It isn't about video, voip, or extra service quality. It's about money.

The Bells want to be able to go to Google, Yahoo, and anyone else making money and say "give us a fraction of your revenue or 30% of your packets will be dropped".

It's basically like the telcos getting a cut of every transaction made over the phone.

The carrier companies are spewing nonsense to get government sanction for a business model that will have no value to consumers.

Tiered service already exists and bandwidth is not scarce. In major metropolitan areas we have:
Tier 1: Modem dial-up
Tier 2: ISDN
Tier 3: DSL or Cable broadband
Tier 4: DSL or Cable with expanded bandwidth
Tier 5: T1 line
Tier 6: T3 line

Even rural or semi-rural folks have options:
Tier 1: Modem dial-up
Tier 2: Satellite broadband (eg: Direcway)
Tier 3: Satellite with expanded bandwidth

So, why is there a need for packet prioritization? If people need better service they have existing options. For example, I am unhappy with slow uploads using Direcway. Therefore, I will pay more money to secure twice the upload (and 40% more download) bandwidth. If I wanted more reliable transmission that was not affected by heavy cloud cover or high winds, I also can establish a dial-up modem account with Direcway or many other ISPs. If an Internet connection (albeit a slow one) was essential, then the extra $10-15 a month would be money well spent. Adding some time of priority packet scheme is unlikely to be beneficial.

The idea that the government has to be involved in any way here baffles me. If the carriers want to offer tiered service, why don't they just offer it and charge for it? What is the role for legislation here?

Buy some MPLS routers and start trying to get customers doing video streaming to pay extra for priority on video packets.

In fact, doesn't this exist? What the hell is a VPN but a way of reserving a pipe from point to point? What does Packeteer sell but bandwidth management boxes that let the buyers prioritize some packets over others? What does Akamai do but help ensure that video streams and other web content reach the user quickly across the legislation-free internet?

Am I missing something here?

This has nothing to do with the web and email, and everything to do with Vonage and Skype. The carriers will effectively lose their long distance service as VOIP services proliferate.

Any medium or large business these days isn't bothering with a PBX- they are going right to VOIP. This is the leading edge of the wedge. When the telcos lose their fat long distance subsidies, they lose a hell of a lot of operating margin.

The only way they can compete is by intentionally degrading their networks to VOIP traffic. It's interesting that they only started talking about this when google introduced Google Talk, which is effectively a VOIP service.

> What the hell is a VPN but a way of reserving a pipe from point to point?

Most VPNs reserve nothing.

The "V" stands for Virtual, as in "faked". They fake the security attributes of a Private Network by encrypting the bits sent by the end-point applications.

Folks can arrange for special traffic handling (read priority, bandwidth guarantees, etc) between two points, or between their sites and a provider's end-points, but that's orthogonal to VPNs.

> What does Packeteer sell but bandwidth management boxes that let the buyers prioritize some packets over others?

Those bandwidth management boxes only prioritize traffic across them. They have no effect on how said traffic is handled between them and the traffic's destination.

> What does Akamai do but help ensure that video streams and other web content reach the user quickly across the legislation-free internet?

Akamai puts copies of content "close" to consumers. Again, there's no guarantee that the copied data will arrive sooner (except for the small detail that Akamai tries to use "sooner" to determine "close").

> Am I missing something here?

A basic understanding of how the internet works.

There are several kinds of tiering out there. Some are good, others are bad.

Good: users paying more for higher ISP bandwidth.

Good: payment for heeding packet priority markings and/or reserved minimal bandwidth rates, the policy to only have effect when bandwidth is scarce; nothing happens when there's plenty because no packets of any sort get dropped.

Bad: setting a low bandwidth limit you're willing to see between your users and external websites (e.g., google). Packets above the limit get thrown away, as Andy Freeman said.

Worse still: setting a low bandwidth limit and expecting websites to pay a toll if they want users to get there happily. There's some evidence that this is what BellSouth wants to do.

If broadband were a healthy non-monopolistic service, this would be no threat..... As it is, since BellSouth has been paid to develop their quasi-monopoly network at high taxpayer expense, this seems to me to be an unfair threat to competition.

The "Bad" practice is already a fact of Internet life. Not sure if anybody expects ransoms yet.

Jim, I am with you. Why do they need gov't permission for this?

In theory, tiered service, whether on the provider or the consumer side, is a good thing. It encourages investment, improvement of services and priority of resources.

In reality, I am sure that a number of the bandwidth providers will put forth a number of failed experiments. The concept should certainly be allowed, even if the market might say no.

But can anyone explain whose permission is required exactly?

One more item of interest. Mark Cuban, who is deep into the content side of the equation and has a few bucks, argues for tiered service:

http://www.blogmaverick.com/entry/1234000267073488/

Tim Howland is partially correct.

I've read a remarkably candid interview with a baby Bell CEO a few months ago that dealt with this exact issue. The baby Bells know the traditional telephone lines are dying and a complete switch over to VIOP is inevitable. The point is who's VOIP service you're going to switch to. The bandwidth providers want to use their control over the pipe to make sure their VOIP, soon to be deployed IPTV solutions, and what ever else they are selling runs better that their competitors. The Anti-Trust issues are staggering.

Private industry is asking for government regulation in order to allow for greater diversified business opportunities?

Aren't they asking for the *removal* of a regulation that is stopping them from doing what they want to do?

Hmmm.

1. A few people aren't understanding the issue. It's not the ISPs wanting to charge Joe Sixpack more money for a faster connection. It's the ISPs wanting to charge Google more money to deliver it's content to Joe Sixpack.

I.e. two competitors Amazon and BarnesandNoble. Amazon coughs up the money and gets a premium service, BarnesandNoble doesn't. Most people will gravitate to Amazon because the web experience is faster, smoother and more efficient. However most people will also not realise that it's not BarnesandNoble's fault but rather their ISP.

2. I'm frankly opposed to this because both entities are paying the ISPs already. Joe Sixpack is paying arounr $50 a month for a cable modem connection or DSL. Amazon is paying for all of the bandwidth it is using already.

Just what exactly isn't already being paid for? Is this nonsense going to be premium service above and beyond what exists? Or will the ISPs universally reduce bandwidth across the board and then offer a return to status quo ante for an additional charge?

3. How does this impact flat-fee pricing? If I'm paying for 1.5mb bandwidth at my home and the ISP intentionally throttles the bandwidth, how is this not injuring me?

4. Tiered pricing is isn't the endpoint, it's the starting point. The real destination is metered service charges. But to get there you first have to convince people that flat-fee pricing is no longer possible or else the customers will balk. Once the customers have bought into the idea that more on the internet requires more cash, then metered service charges will be the new new reality.

Additionally the mechanisms needed to implement metered service charges are also those that are needed for tiered pricing.

5. Cross-charging scams will be the new normal. The way this works is this. ISP "A" services Joe Sixpack. ISP "C" services Amazon. ISP "B" is an intermediary. When Joe Sixpack visits Amazon to order something his packets have to travel from "A" to "B" to "C" and back again. The issue is that ISP "B" can raise it's charges to carry the bandwidth of "A" and "C". Those ISPs will now pass along the charges to their customers with the excuse that it's not them raising prices, it's someone else. Now shift the configuration a bit so that "B" is going to "C" via "A".

In essence each ISP will be able to raise prices on carrythrough traffic without having to raise prices directly on their customers. However they will apply the rate change on the carrythrough traffic to their customers. So the ISP isn't raising prices, it's just the internet is getting more expensive.

6. It doesn't matter if you use satellite or cable or DSL. It's not necessarily how you connect but the fact that you do connect. Any connection is local but the traffic is international. Do you pay more for local calls or national ones? How would you like your cellphone provider to drop the whole "call anyone in the nation for one price" scheme and return to the tiered pricing for long distance a la Ma Bell?

This is a really bad idea that won't go away. It'll be defeated now but it'll come back every single year. The only way to really defeat it is to make sure that any politician that tries to pass this realises that it'll cost him his job.

I'm with Jim and Matt: The Internet is already an uncontrolled madhouse where you pay for exactly the level of service you want. Want a gigabit link with 99.99% uptime from your Santa Clara site to a Chicago network access point? Get out your checkbook.

The only reason to get the government involved is to make bait-and-switch legal. SBC and friends want to be able to advertise full-featured Internet access, but selectively deliver poor access to their competitors.

Winterspeak said "The unregulated internet is an anomaly in the media world and it's not surprising that carriers want that to change."

The Internet is extremely regulated (contract law, the Computer Fraud and Abuse Act, advertising law, etc.) What it is not is centrally controlled. Anybody can just add their own chunk to the Internet, just because they feel like it, and it doesn't even cost much.

Daniel hit the nail on the head. This is one facet of an effort by companies like the new AT&T to assert control over the Internet's infrastructure. They don't like that they can't charge an arm and a leg to get packets from Point A to Point B, so they'll try to make it so through the government. Expect attempts to impose new networking standards, take control of the DNS, and get draconian laws passed to prevent anyone from upsetting the apple cart.

Arnold Kling, though, questions the very notion of bandwidth being scarce. I'm with Arnold, I don't think bandwidth is scarce.

It's not. The going rate for bandwidth has gone down every year for the last decade or so. That's not what would be happening if bandwidth was scarce.

These days, many hosting outfits are paying more for electricity than for bandwidth.

I believe that carriers are agitating for this because they don't want to be dumb pipes competing in a commodity business.

DINGDINGDINGDINGDINGDING!!!

You win a prize, sir.

Telcos are in big trouble. The voice-landline business they built themselves on is going away, and they need to find another source of revenue in order to stay in business without getting a lot smaller.

They can't really differentiate themselves based on product- a bit that goes through Carrier A's network is no different from one that goes through Carrier B's network (if they are, there's a big problem)- and they can't really differentiate themselves based on price (selling bandwidth is already a low-margin business), so what's left?

That's the question telcos are trying to answer. Apparently the idea is to create a product via governmental regulation and then sell that product.

Speaking as a contributor to the problem (ditched the landline and went cellular years ago), if the telcos sold a product I needed, I'd buy it. That they don't is their problem, not mine.

I use Packet8 VOIP virtual PBX. It is perfect for our small business with 5 small offices in 3 states. Pick up the receiver and punch in an extension number, my colleague answers 1200 miles away; conference calls; big ticket voicemail features; unlimited local and long distance minutes. Same monthly price as a standard wired phone from AT&T with similar calling plan, but I get the PBX functionality for free, not thousands extra. My Telco (SBC/AT&T) makes zero dollars and Roadrunner (ISP) makes zero incremental dollars.

This, they will not let stand. They will look to the gov for a fix ... either pay up or they will degrade my VOIP packets.

Similar situation for streaming media which will explode in the next few years.

How will they get away with this? Create a crisis, call me a bandwidth hog, invite the gov to step in and fix the 'problem'. The fix will be that we will be forced to use our ISP's media and comm solutions, or pay more. They will eliminate competition in services. They want a vertical monopoly to be authorized and institutionalized through the FCC.

"Why do they need gov't permission for this?" What they want is regulatory cover for the sh*tstorm that will arise when they start deliberately degrading regular service. And more than that, regulatory obstacles to competitors entering the market when users realize what their telcos are doing to them.

All resources are scarce. Even bandwidth. Besides, it'd cut down on spam.

Hmmm.

Besides, it'd cut down on spam.

Sorry but it won't. What it'll do is drive spam into cooperating with criminal gangs that have large arrays of zombie computers. What'll happen isn't that the spammers will pay more for sending spam, instead it'll be the average computer-illiterate user who suddenly finds a $1,200 excess traffic charge on his monthly ISP bill.

I run a regional ISP that covers about a predominantly rural portion of a central US state, and have been involved in the technical management of Internet operations since 1993. I've worked with bilateral and multilateral exchanges, have been involved in the IPv6 Summit and administer a prioritizing MPLS network.

That all said, there are a few observations I'd pass along regarding the requests of several of the carriers. There are issues today with traffic consumption by residential subscribers not paying for the capacity - peer-to-peer is a particular abuser. When the commercial rate for a 1 Mbps tier one or high-quality tier two provider is about $250 (bought in quantities greater than 1 Mbps obviously), and a residential cable customer purchases 8 Mbps download for $55/month and uses it, there is a problem.

Many have already implemented service grade tiers. We tag and queue traffic, providing different prioritization approaches by customer grade. We also identify as much P2P traffic as we can (always evolving) and rate-limit it to about 20% of the subscriber's purchased capacity (and detail this in their service agreement - they are free to purchase a commercial account for service not limited by protocol).

But this particular carrier petition is about rate complexity and is a red herring using traffic management issues as its cover. We all can address the traffic and pricing issues internally. Instead, many seek a complicated settlement and measured use arrangement which will not see the light of day if left to the market to establish it.

Consider if BellSouth cuts off Yahoo, Google and a few dozen sites that refuse to pay it for access to its eyeballs. Customers immediately drop the already marginal, underperforming DSL and switch to wireless or cable solutions. Only if regulatory structures force all providers into the content and/or use measurement model will BellSouth be able to go ahead.

And remember, complicated billing schemes always favor the bigger guy, especially when the billing platforms start at $5 million for software and millions annually for maintenance.

-sam

One additional observation (in the "those who forget history" category):

Much of what is happening is the continuation of a very old battle that started with Al Gore's NSFNET, believe it or not! As the NSFNET was dying in the early 90s and regional non-profits were already illegally shipping commercial traffic (such as MIDNET supplying a resale connection to Inacom against NSF rules and MIDNET's charter, using leased lines in Nebraska paid for by the Universities - thankfully both ethically challenged organizations are extinct today), the NSF board wsa under considerable pressure to provide a commercialization structure.

The solution was a hierarchial NAP (network access point) model that *assigned* NAP privileges to each RBOC, carried the traffic between the authorized NAPs over ANS (an IBM and MCI joint venture that had the NSFNET contract and is what gave notion to the concept of a single "backbone"), and then permitted interconnection at a NAP for last-mile distributors.

The NAP administrators (Ameritech would have been the one we would have had to work with) would be free to set rate structures, and were favoring a usage-based model that mirrored their feature group D model from voice telecom. They would also be free to discriminate against their own last-mile services and provide it a discounted rate - being both NAP admin and competitor to other NAP customers.

Faced with this absurd model, the existing commercial players (PSI, UUNET, NETCOM, Sprintlink and a few others - MCI wasn't even in the game at this time nor was AT&T) expanded the multilateral model through MFS's fiber networks and moved past the venerable CIX routers. Bilateral agreements between networks also became fashionable.

Despite Senator Gore's and the NSF's efforts, the BellNAP model died a quiet death, or so we thought. As the first completely commercial ISP (we were the first CIX member with zero NSF connectivity, this subjected us to quite a few lessons in the NSF's politics and ultimately took Senators from our region to force the NSF to let our school district customers have access to their "Bell and NSF Regional club-members only" network). Absent regulation forcing the unnatural model, it'll stay dead. Those who seek innovation and a US tech industry that can compete had better work to keep this fossil from returning to life.

-sam

I agree that the motivation is money, and in many cases, survival for the telcos.

But the technical excuse/reason that you will hear cited will be VOIP packet prioritization in order to control echo, if one or both end terminations goes into the circuit-switched network. The ATM network already handles this, and they will insist that they *have* to have packet prioritization to maintain a usable connection... which is, of course, BS.

With the massive increase of processing power available at the end point, echo cancellation, which used to be a high dollar item placed at concentration points in the voice network, is now easily done at the "telephone" or voip net interface point - on the voip providers nickle, of course.

This won't solve the "delay" issues that will occur when the net congests, but for that to become a serious problem we will really have to have a real bandwidth shortage, and until the vastly overbuilt fiber net - a legacy of the comm
and . boom - if filled, we're a long way from there.

email is human readable - aloud.

Also think about what this would do to compatibility?

And the only reason that VOIP is so popular is because those same guys have been gouging the public for landline service for so long. My SBC landline, a dumb, slow copper hookup, costs me more than my cellphone. The RBOCs deserve to be smashed face-down in dug turds for the way they have treated the public all these years.

The fact that there's a lot of capacity in backbone networks doesn't change the fact that some applications (like streaming voice and video) are more sensitive to packet delay, jitter and loss than non-real time applications (like email and blogging). Prioritizing real-time applications makes sense if consumers want high-quality performance in those applications.

Also, whatever the capacity may be in the backbone, customer's "last mile" facilities are also subject to congestion. Prioritization in that part of the network may be key to providing applications like HD video over broadband.

It makes little sense to prohibit networks from prioritizing traffic. Antitrust questions may arise with respect to what they do with prioritization, but there may be good, consumer welfare enhancing reasons for prioritization. This should be looked at on an individual case basis.

> customer's "last mile" facilities are also subject to congestion.

That congestion is entirely under the customer's control and the appropriate place to shape the traffic is at customer's end.

If last-mile congestion was actually the issue, BellSouth would be announcing their "media friendly router/modem", aka CPE, and they wouldn't need any regulatory approval to do so. (Such equipment is already available from third party suppliers.)

Instead, they're asking to charge Google for responding to search requests. As I wrote above, what they're really asking for is a cut of every transaction. We don't give them a cut when we use the phone to conduct biz, why should they get a cut when we use DSL?

> The fact that there's a lot of capacity in backbone networks doesn't change the fact that some applications (like streaming voice and video) are more sensitive to packet delay, jitter and loss than non-real time applications (like email and blogging).

Except that they're not, unless we're talking real-time (which is sensitive to jitter). And even then, they're not nearly as sensitive as the above asserts. The issue is almost always bandwidth. When packet loss is an issue, it's typically so bad that nothing much works.

Note that VOIP is remarkably tolerant of jitter due to the fact that we already accomodate folks who pause while talking.

> Prioritizing real-time applications makes sense if consumers want high-quality performance in those applications.

Nice theory, but VOIP works today and the real demand is streaming video, not real-time, which is very jitter insensitive.

And then there's the fact that useful mechanisms to address those problems have nothing to do with the proposal, as I discussed above.

The proponents of the proposal have yet to explain why the pipes should get a cut of Google's advert revenue for searches, even though that's exactly what BellSouth is asking for.

Should BellSouth get a cut when I order take-out pizza by voice phone? Should the amount of money BellSouth gets depend on how many pizzas I order? Note that that's what BellSouth is asking for.

As Tim Howland and others have pointed out, there's definitely a survival battle going on. Thanks also to San and ed for clarifying some of the technical aspects of this debate.

I believe one should keep in mind that most of the carriers are "born and raised" in a hevily regulated monopolistic market. Regilation is what they know and as new technology such as VOIP proliferates, it cuts traditionally profitable margins. So, if you have a hammer, all problems tend to look like nails. A high cost barrier to entry have allowed carriers inflated margins for a long time.

More regulation could hamper new connectivity technologies such as internet over power lines or high-speed wireless networking to circumvent the existing technologies. More government regulation over the internet will ultimately mean that you will pay more. A long time ago, DSL was only DSL. Now, it's ADSL and DSL with different speeds. I don't believe this was done to enhance customer experience, but to increase the bills to the customer.

This just seems like another reason to prevent competition and tack on additional charges. Why are there, for example 5 to 7 billable line-items on the land line bill? Prioritized service, time, and megabits downloaded/served would certaily all qualify for extra charges if traffic is prioritized and regulated either by users or servers.

>That congestion is entirely under the customer's control and the appropriate place to shape the traffic is at customer's end.

This is true for upstream packets, but not for downstream. The customer end cannot do anything congestion in the last mile for downstream transmission.

>they're asking to charge Google for responding to search requests.

I didn't defend such charges. My point was only that there may be legitimate, consumer-welfare-enhancing reasons to prioritize certain applications in some circumstances.

>Except that they're not, unless we're talking real-time (which is sensitive to jitter). And even then, they're not nearly as sensitive as the above asserts. The issue is almost always bandwidth.

Streaming video (or voice) are sensitive to jitter, delay and loss. I didn't any specific parameters around how sensitive, so the claim that "they're not as sensitive as the above asserts" doesn't make any sense. Also, since bandwidth in the last mile is fixed in the short run, bandwidth alone cannot always be the issue (or the solution).

>Note that VOIP is remarkably tolerant of jitter due to the fact that we already accomodate folks who pause while talking.

Pauses in speech may convey useful information, unless they're not actually pauses but artifacts caused by jitter.

>Nice theory, but VOIP works today and the real demand is streaming video, not real-time, which is very jitter insensitive.

VoIP quality is remarkably uneven. Sometimes it's better than what you get on a regular phone, other times it's almost unbearably bad. As I said above, streaming video is a big issue. As providers try to roll out services that include streaming HD video, it may be critical to have prioritization. That's why you see people like Mark Cuban supporting the idea of prioritization. He's not in the network business, but he does want to stream HD content.

>The proponents of the proposal have yet to explain why the pipes should get a cut of Google's advert revenue for searches, even though that's exactly what BellSouth is asking for.

The only reason BellSouth should get a cut of anyone's revenues is that because that entity wants to give them the cut. The government should not authorize BellSouth (or anyone else) to coerce payments from anyone. And the government should prohibit BellSouth from blocking as a way to coerce payment. But it shouldn't prohibit them from prioritizing traffic, unless it sees that they use prioritization to violate antitrust law.

> The customer end cannot do anything congestion in the last mile for downstream transmission.

Downstream traffic is caused by upstream traffic. If I've got a 3mb downstream channel and I request 15mb of stuff, I've got a problem that nothing upstream can solve. (It's actually worse than that - I may want the movie to watch next week but be waiting for an important e-mail.)

> I didn't defend such charges. My point was only that there may be legitimate, consumer-welfare-enhancing reasons to prioritize certain applications in some circumstances.

The proposal on the floor is to charge Google for supplying search results. Defending the strawman mechanism to implement said proposal is defending the charges.

> Streaming video (or voice) are sensitive to jitter, delay and loss. I didn't any specific parameters around how sensitive, so the claim that "they're not as sensitive as the above asserts" doesn't make any sense.

Streaming content is content that is consumed before it is completely received. (Yes, there is a difference between real-time and streaming. Unless you're interacting, real-time is unnecessary. That's why there's remarkably little real-time - even broadcast media runs on a "tape delay".)

If I start playing 3 minutes after the sending starts, I can tolerate 3 minutes of problems before noticing anything. (The observant reader will notice that I can use that technique to "stream" content that requires 3mb of play bandwidth over a 1mb channel.) If my downstream channel is a bit wider than my play bandwidth, I can even accumulate problem tolerance during playback. (Once again, traffic shaping can't put 15mb of traffic on a 3mb channel.)

>Also, since bandwidth in the last mile is fixed in the short run, bandwidth alone cannot always be the issue (or the solution).

The fact that bandwidth is fixed does not mean that it isn't the issue or solution. If the demand over a sufficiently long period is significantly greater than the supply, it's the problem. Note that time and bandwidth can be used to solve jitter, delay, loss problems.

> Pauses in speech may convey useful information, unless they're not actually pauses but artifacts caused by jitter.

Who said anything about useful information? We tolerate random pauses in speech because speakers pause.

> And the government should prohibit BellSouth from blocking as a way to coerce payment.

BellSouth wants to drop some packets to/from Google unless Google pays a fraction of its revenue to BellSouth.

That's the proposal you're supporting.

Note that said proposal actually wastes bandwidth because the packets will be retransmitted.

Shedding traffic helps only for congestion bursts. It can't help with long-term overcommits and actually makes things worse. That's why I wrote that "natural" packet loss is either in the noise or a virtual shut-down.

>If I've got a 3mb downstream channel and I request 15mb of stuff, I've got a problem that nothing upstream can solve.

But prioritization of your downstream packets can help. That is, if you have a 2MB video stream and the rest of it is stuff where you can tolerate delay, you may prefer that your provider prioritize your video packets.

>Defending the strawman mechanism to implement said proposal is defending the charges.

No, it's not. I again condemn attempts to coerce payment from Google by Bellsouth (assuming such attempts are made).

People may or may not want ISPs prioritizing real-time applications. We'll find out when they actually start doing it. My only point is that prioritization is not necessarily harmful to consumer welfare, and that there are circumstances when it may be beneficial.

> But prioritization of your downstream packets can help. That is, if you have a 2MB video stream and the rest of it is stuff where you can tolerate delay, you may prefer that your provider prioritize your video packets.

As I pointed out, the provider doesn't have the relevant information. (Video isn't always a higher priority than e-mail.) The customer end has both the relevant information and the ability to control.

Real-time won't ever dominate, so it's the wrong thing to optimize for. (Streaming isn't real-time.) And, even if/when it does, there's no reason to do so badly.

Why this fetish for an inferior solution to wrong problem?

> I again condemn attempts to coerce payment from Google by Bellsouth (assuming such attempts are made).

Is the parenthetical ignorance or chutzpah? BellSouth NAMED Google as one of the parties that should be paying AND said that the payment should be a fraction of revenue.

We've seen two things, the words "I condemn" and an exhaustive, albeit bogus, argument for the proposed mechanism for extracting that payment ("nice packets you got there, I wouldn't want to see anything happen to them").

I've said all along that I don't know whether prioritization will or won't enhance consumer welfare, but that it's conceivable that it could. You seem to want to convert this rather mild point into a defense of anything and everything that an ISP like BellSouth might try. Sorry, but I'm not defending those things.

I know that I would be willing to pay my broadband provider to prioritize my VoIP packets if they could show that doing so would make the quality more consistent. Maybe they couldn't do it. But if they could, I'd pay because I occasionally am frustrated by the service quality. Similarly, if and when I want to watch real-time video over my broadband, I would probably be willing to pay for prioritization of that video if feasible and service-improving.

> Similarly, if and when I want to watch real-time video over my broadband, I would probably be willing to pay for prioritization of that video if feasible and service-improving.

As I've pointed out more than once, you can do that yesterday by buying the appropriate CPE.

BTW - Not only is CPE that does the relevant prioritization available now, but Internet topology significantly limits the benefit that BellSouth can actually provide.

It's unlikely BellSouth owns/controls the whole network from content to its customers. That content probably lives on other networks that are, possibly indirectly, connected to BellSouth's. Congestion on those networks limits the prioritization that BellSouth can provide.

The TOS will reflect this. They'll basically say "we're charging you for extra priority, but you may not get any benefit".

Best effort for all traffic really is the most reasonable policy.

Two things are happening here, and bandwidth is not scarce-- if anything, it's likely to increase exponentially.

First, the carriers are trying to make the internet over in the image of a "rent-based" system. That is, they're looking for ways to make it a seller's market. That's the model of cable tv, big pharma, and as much of American business as can do it: make yourself the only possibility, and make the captive buyers pay big. In the business press this is highly thought-of and is usually called "pricing power."

The other thing is that this administration really wants to control internet traffic and content. Witness the Google fishing request.

If these two interests succeed in getting their way, we'll be like China, internet-wise. And that's certainly what the administration wants.

> The other thing is that this administration really wants to control internet traffic and content.

A Dem administration might use a different excuse, such as going after blogs, but they'd have the same aim. (Note that Clinton pushed Eshelon.)

Comments are Closed.