The internet is over!

[Undertoad]

Originally Posted by Happy Monkey

But how do the ISPs hook up? Don't they have to go through a wire-owner at some point?

Here is a list of backbone providers. Short version:

360networks
AboveNet
Ameritech
AT&T
Btnaccess
Bell Canada
BellSouth
Broadwing
Cogent
Electric Lightwave
Fiber Network Solutions
Genuity
GlobalNAPs
Globix
GT (Canada)
IDT Corporation
Level 3
Multacom
Mzima
Netifice
Oxford Networks
PPL Telcom
Quest (Asia)
Qwest Comm.
SAVVIS
Sprint Wholesale
Telcove
Teleglobe (VSNL)
TeliaSonera
Telstra Inc. (Asia, USA)
Time Warner Telecom
Verio (NTT)
WilTel (Williams Comm.)
XO Comm.
Xspedius

A smart ISP would get a circuit to one of these people. But they might just as well get IP service from someone who got a circuit from these people. Or they could get service from someone who got service from these people. Or they could get multiple services and implement some sort of routing redundancy.

And now the hundreds of leaf nodes become thousands and millions, and the latency penalty of additional routers is still plenty low for voice.

Notice which two names are not on the list.

[Undertoad]

BTW - I'm a Verizon FIOS customer - and traceroutes to myself show that Verizon is connected via at least three of the above: Level 3, AboveNet, and 360 Networks.

Skype lists its "Carrier Partners" on this page.

[tw]

Originally Posted by Undertoad

tw: After Skype hooks up a simple packet sniffer and evaluates the arrival times of different packet numbers during a call, and compares that data to the packets they themselves have generated from the remote location, they will have all the compelling data they need before lunchtime.

Good luck. Packet skewing looks like internet latency. Have you every done packet sniffing on large networks?

Furthermore, intermittent application of packet skewing would make detection using packet sniffers even more difficult if not impossible. Skype customers sometimes get bad service which is enough for them to quit Skype. Meanwhile, while you are looking for packet skewing, the IP provider is using the X technique - that is secret so we cannot define it. How are you going to detect the X technique when you don't even know what to look for? Under current law, the X technique to selectively distort service is also legal.

You claimed that IP providers would just provide basic services because if they did not, then market forces would force them to change. Demonstrated are numerous techniques - technical, legal, and secret - that demonstrate that assumption is seriously flawed. Even demonstrated are trends by big IPs to monopolize more of the internet - even blaming Google for earning profits on a 'free ride'. I don't have your faith that a free market alone will provide a fair market because, already, competitive DSL providers were all but driven out of business AND because Verizon and Comcast even got laws passed to stop all future mesh networks in Pennsylvania.

How do you reconcile that law with your original assumptions? How do you reconcile that Washington Post article that says large IP providers want more control and profits of internet business - even at the expense of Google. I just don't have your faith in their integrity and honesty - especially with the number of times they have already demonstrated intent use their large IP infrastructure and 'purchased politicians' to stifle competition.

[tw]

Originally Posted by Happy Monkey

But how do the ISPs hook up? Don't they have to go through a wire-owner at some point? ISPs don't generally own the copper or fiber. And they certainly don't own all the wire that your packets go through to get to their destination.

It’s called the 'last mile'. Only about 2.5% of backbone fiber was (and probably will be) used is due to limitations created by the 'last mile'. Who are your 'last mile' providers? For dialup, a phone company. For broadband, most only have a phone company and a cable company - or less. UT only has two choices. He can connect via Comcast or he can use Verizon (either DSL or FIOS). No matter who his ISP or backbone provider is, he still must use the 'last mile' infrastructure - Comcast or Verizon.

Earthlink, et al cannot provide a mesh network because it was made illegal in PA. A law passed because those IP providers wanted to respond to consumer demands? I don't think so even though UT disagrees.

So you tell me - who is sitting best in a position of power? The 'last mile' providers 'stuck it' to AT&T. They also stifled 1981 DSL technology for over a decade. They resisted ISDN for how long? They are now vying for more control - not just being IP providers. And they have legal, political, and technical power to do so.

Do you really think these 'powers that be' will not exercise their power? They were the reason why multimedia did not happen in the early 1990s when national providers had installed fiber optic across the nation just for that purpose. The 1996 Communication Act was passed only because the 'last mile' providers (now called cable and telephone companies) would not innovate. And yet we trust them to comply with free market forces as UT claims? I seriously doubt it when their history is to do things despite those free market forces.

We are all at mercy to the 'last mile' providers who have so many options - including 'packet skewing - to manipulate the market, unfairly, to their advantage. In UT's case, they are called Verizon and Comcast. He has no one else to turn to. But still UT claims a consumer forces will make them comply ... when free market forces did not. When we even needed a 1996 Communications Act to make them respond to market demands. Why would they suddenly respond to free market pressure (as UT assumes) when they refused to previously?

[Undertoad]

You just route the packets you want to see to a place where you can watch them with ease. The only "techniques" they can actually use are introducing latency or dropping packets. Almost every network tool ever written to evaluate broken connections measures those two things in detail over time. (Including the granddaddy of them all: ping.)

The old motel is bypassed by the big highway. In a fit of pique the old motel digs up the entrance/exit ramp next to it. Unfortunately for the motel this dries up the last source of business and fails to hurt the highway one iota. The motel must lean that the highway has more power and the only way to survive is to work with it, not against it.

[tw]

Originally Posted by Undertoad

You just route the packets you want to see to a place where you can watch them with ease. The only "techniques" they can actually use are introducing latency or dropping packets. Almost every network tool ever written to evaluate broken connections measures those two things in detail over time. (Including the granddaddy of them all: ping.)

Your post assumes the latency is always same number in every ping. It is not, as was even demonstrated by web sites such as InternetWeather.com and others who have since taken up that task. That 'always changing' latency makes it simplistic to perform ‘packet skewing’ without detection. When 'packet skewing' is applied intermittently, then it is all but impossible to detect.

Furthermore, 'packet skewing' means pings travel normally whereas VoIP or other type packets are skewed. Just another reason why consumers would cancel Skype service and buy VoIP from the big IP provider.

And then we have techniques X, Y, and Z that are legal. How do you detect them? You don't even know what to look for. And if you detect them, well, so what? It is legal for IP providers to use such techniques. You are making assumptions not even based in technical reality.

UT. Provided are so many reasons legal, political, historical, and technical why IP providers - the 'last mile' providers - can and may manipulate their networks to maximize their products at the expense of competition. They have already done so previously. It was and is legal. Your claim that market forces would prevent this has repeatedly and historically been demonstrated a myth. Again, did we not learn from AT&T? Did we not learn that it took a 1996 Communication Act to get broadband provided? Where were these market forces that made the 1996 Communication Act unnecessary? Where were those market forces that protected AT&T?

Somehow, you still think Comcast and Verizon - your only two providers - will not unfairly manipulate the market using numerous political, legal, and technical techniques? They already have and no one complained? Where is a public up swell because a Philadelphia Earthlink mesh network is not permitted anywhere else in PA? Where is this consumer demand that UT insists will protect the market? Why then should we believe consumer 'free market' selection will keep those IP providers - the 'last mile' providers - honest? UT did not even complain when mesh network service was denied to protect Verizon and Comcast. History contradicts UT's claims. Technical facts demonstrate why even 'packet skewing' can be made all but impossible to detect.

[Undertoad]

You're over your head and looking bad. But if you're so certain, just make a prediction for the Cellar calendar.

[tw]

Originally Posted by Undertoad

You're over your head and looking bad. But if you're so certain, just make a prediction for the Cellar calendar.

Well if you are going to insult rather than answer technically, then that is proof you have conceded.

But please tell us why those IP providers - the 'last mile' providers - will be honest when so much history says otherwise. You ignore reasons political, technical, legal, and historical when you cannot reply? I suspect you had no idea why the 1996 Communication Act was created - which explains why you pretend I never cited it ... and so many other facts.

Meanwhile, UT, you are fooling only yourself if you think 'packet skewing' and other IP tricks can be reliably proven by 'ping' type testing. You are fooling yourself if you think with only two 'last mile' providers, then market forces will keep them honest.

[dar512]

Originally Posted by tw

Good luck. Packet skewing looks like internet latency. Have you every done packet sniffing on large networks?

Even so, at levels in which it would be audible, you have to be losing a fair number of packets. Certainly enough to be subject to statistical analysis.

[Undertoad]

I would gather an hour's worth of data from point A to point B and then demonstrate what went wrong.

I'd ship other packets from the same point A to the same point B, during that same hour. I'd also ship from point A to point C and from point B to point C and collect that data. Perhaps I'd also ship a different protocol of packets, to see if the protocol made any difference. How about ICMP packets? Good choice, everyone routes them, and every network analysis tool will interpret them, along with the packets used for data transmission of voice. Any latency or packet loss introduced by hardware or most routing problems would affect both protocols. Thus, "ping type" testing - looking at ICMP messages - is a half-decent approach.

[tw]

Originally Posted by dar512

Even so, at levels in which it would be audible, you have to be losing a fair number of packets. Certainly enough to be subject to statistical analysis.

Let's say you are Skype. You find that your 'quality of service' using Skype on Comcast is inferior. So what do you do? Do you announce to the world that Comcast is a bad service? Of course not. No advertiser profits from such comments. Spin doctors will say that is a worst you can do. If you announce Skype service is poor on Comcast, then the consumer will use some other VoIP provider.

Will you change IP providers? Who? You only have two choices (in UT's case) - Comcast and Verizon. Others have even less choices. But both are doing things that may degrade Skype quality because - 1) it is legal, 2) they want Skype's customers, and 3) no one can for one minute claim a statistical poorer quality is intentionally due to what Comcast and Verizon are doing.

Yes, you can statistically measure a degradation of IP service - if you know what to look for. So what. That degradation also happens during normal internet operation. Statistical measurement becomes inconclusive if 'packet skewing' is performed intermittently. Furthermore, if you (Skype) complain to Comcast, et al, well, Comcast need not do anything but claim ignorance. You (Skype) have no legal options other than to build your own IP service network - from scratch.

Furthermore, we have only discussed service degradation using packet skewing. What about technique X, Y, and Z? How you measure for degradation by those other methods (X, Y, and Z) which are also legal and that you don't even know exists. Remember you must also prove such degradation is intentional and not due to inferior Skype design. And then how many years will you go about measuring quality of service everywhere? Remember, they can apply service degradation intermittently. You are assuming Skype is a large organization with money to burn on verifying quality of service.

The Baby Bell must provide minimal 'circuit switched' service quality. It’s the law. Unlike IP service providers - the 'last mile' providers - the circuit switched services have specific numerical targets that must be met - as stated in government regulations. IP service providers (ie Comcast and Verizon) are exempt from such standards. UT says they will provide good service anyway because the consumer will blame Comcast and not blame Skype.

UT says they will provide those standards due to consumer 'free market' choices. I say bull. IP providers are not required to, the competition does not exist, and manipulating those IP services for self serving gain is too easy, too difficult to detect, and too profitable. Furthermore the big IP service providers have already demonstrated that they will do such tricks to benefit their company at the expense of potential competition. Trying to prove they are doing so - even statistically - got those other victims squat. Why do you think you - Skype - doing a massive statistical analysis will be any bit more successful?

Even if you statistically detect service degradation, then what are you (Skype) going to do? Sue? Good luck. Consumers meanwhile will simply take the easy way out. Comcast and Verizon provide reliable VoIP service. Since consumers have even less understanding of what I have posted - the technicals - then they will simply shift to Comcast and Verizon for more reliable service.

But UT says those consumers will leave Comcast and Verizon instead - while still using Skype. Why would they? They - like some here - don't even understand these simple technical explanations. The consumer will first abandon Skype long before they will reread (to finally comprehend) what I have posted here. Doing a statistical analysis would cost too much, hopes you know what to look for, can identify such problems as intentional verses normal internet variations, AND assumes the results of that analysis will mean something to the consumer. Good luck meeting all those points.

How many times did I show a statistical analysis would provide little useful information? Eight? Fifteen? I lost count.

[tw]

Originally Posted by Undertoad

I would gather an hour's worth of data from point A to point B and then demonstrate what went wrong.

I'd ship other packets from the same point A to the same point B, during that same hour. I'd also ship from point A to point C and from point B to point C and collect that data. Perhaps I'd also ship a different protocol of packets, to see if the protocol made any difference. How about ICMP packets?

UT - did you read the IEEE Spectrum article? Even simple packet skewing software will identify different packets from different type of VoIP providers; skewing only ones that an IP provider wants to degrade; leave all others undisturbed. While VoIP packets are skewed, those ICMP packets remain totally unaffected. Why would you even bring up ICMP which is not affected by 'packet skewing'? But then you knew that. So why are you trying to confuse others with ICMP? If I did not know better, I would suggest you don't understand the many services that are carried by IP networks. That is what your latest posts imply. Therefore I can only conclude you are trying to confuse others with irrelevant talk of ICMP.

Furthermore you only reply (partially) to technical facts. You completely ignore the legal, political, and historical aspects. I don't for one minute believe consumer 'free market' attitudes will protect Skype and other tiny companies from legal IP data manipulation. For it that was true, then AT&T and Covad would not have their problems even with regulated Baby Bells. Just one of maybe 20+ previous points I made. Point that you ignore to instead discuss irrelevant ICMP. Somehow you claim IP providers will be very responsive to consumer demands - even without laws requiring it. Your proof? Some irrelevant comment about ICMP.

[Undertoad]

So suggest a wager on the Cellar calendar, if you are so certain about how it will go.

You are assuming Skype is a large organization with money to burn on verifying quality of service.

It's not that hard or expensive, if one knows more than you do about network administration.

But, if you recall, I also suggested that a high price of entry would attract other competition. What if Google introduced a voice communication service?

Well as Douglas Adams used to say, you don't have to tax your imagination to hard, because Google HAS introduced a voice communication service.

[Undertoad]

Originally Posted by tw

UT - did you read the IEEE Spectrum article? Even simple packet skewing software will identify different packets from different type of VoIP providers; skewing only ones that an IP provider wants to degrade; leave all others undisturbed. While VoIP packets are skewed, those ICMP packets remain totally unaffected. Why would you even bring up ICMP which is not affected by 'packet skewing'? But then you knew that.

If you would have been able to comprehend my post, you would have understood that this was exactly my point. I would send both VoIP packets and ICMP packets, and then compare the two; since as I noted, a hardware or routing problem would affect both. If only the VoIP packets were affected, I would know the source of the problem, right? Ya follow buddy?

[tw]

Originally Posted by Undertoad

So suggest a wager on the Cellar calendar, if you are so certain about how it will go.

UT - again you are reading what I did not post. I predicted no specifics with certainty. I defined the big picture; so many ways that consumer choices would not force big IP providers to compete fairly? What is the benchmark - the exact wording - of a bet that is consistent with these 'big picture' trends? Do we bet on Skype going backrupt? I don't predict that. Do we bet that Congress forces regulation on the IP providers? Then state exactly what that law says. Do we bet on whether Comcast used 'packet skewing' software they have purchased? Even if they were, how would we ever know? Exactly what, from so many points and trends cited, do we bet on?

Clearly I did not say something specifically would happen. I stated the so many options that IP providers have because consumer 'free market' choices are just not that influential. But you tell me. In defining a big picture and by not listing a single specific prediction, what do we bet on?

Let's say Google does provide a VoIP service. Everyone is still stuck with the most famous part of every communication network - whether it is packet switched or circuit switched. You are stuck with those same two big IP providers - Comcast or Verizon - even if using Google VoIP. Those 'last mile' providers have so much power as to even stick it to long distance companies and to upstart DSL providers.