Forgot your password?
typodupeerror

Comment: Re: Grade is on the curve (Score 1) 110

by JoelKatz (#47410935) Attached to: YouTube Issuing "Report Cards" On Carriers' Streaming Speeds

"Connecting directly to the destination network is typically free ... you may not get access to another network's entire network."

Exactly. When the costs naturally split fairly evenly, there's no reason for anyone to pay anyone else. If both networks do their fair share, and we presume the traffic benefits both parties roughly equally, there's no reason for anyone to pay anyone else. Any argument that X should pay Y equally argues that Y should pay X. Senders pay to send, recipients pay to receive. This is a pretty typical case, and it's the justification for settlement-free peering.

So why don't you get access to their entire network? Because for some parts of the network, bringing the traffic to the destination network is *not* anywhere close to half the work. When the work doesn't naturally divide equally, settlement-based peering is used. This is how it's been for decades.

In the case of traffic from someone like YouTube or Netflix to a customer of an ISP like AT&T or Comcast, the costs don't naturally divide evenly. YouTube and Netflix use a small number of sources located wherever the cost is least. AT&T and Comcast have a large number of destinations located wherever they happen to be. This is a direct and inevitable result of the business model companies like Netflix and YouTube have chosen. We presume the traffic benefits both sides equally and so each side should pay half the cost.

In the vast majority of cases, the sender should bear roughly half the cost of delivering their traffic and the recipient should bear half. The result of this kind of imbalance has always been settlement-based peering.

Comment: Re: Grade is on the curve (Score 1) 110

by JoelKatz (#47408723) Attached to: YouTube Issuing "Report Cards" On Carriers' Streaming Speeds

It's a lot easier to mock an argument than to address it.

It has been the norm on the Internet for decades to use settlement-based peering when the costs fall unequally on the parties. It is much cheaper for Netflix to generate high-volumes of traffic from a small number of sources placed specifically where the costs are lowest than it is for Comcast to deliver high-volumes of traffic to a large number of sources placed in their customers' homes. The argument that it's fair for Comcast's customers to pay much more than half the cost of delivering Netflix's traffic to them is the one that should be mocked.

Comment: Re: Grade is on the curve (Score 1) 110

by JoelKatz (#47408689) Attached to: YouTube Issuing "Report Cards" On Carriers' Streaming Speeds

Perhaps you're imagining some alternate universe with an alternate Internet that works some other way. But in this world, for decades, settlement-based peering has been used when costs fall unequally on the parties.

Are you arguing that everyone should be able to run a line to the nearest IXP, pay only that IXP, and have Internet access to everyone and everything? Are you aware of the many, many reasons that can't possibly work? Who would pay to carry traffic across the Atlantic?

Comment: Re: Grade is on the curve (Score 1) 110

by JoelKatz (#47408647) Attached to: YouTube Issuing "Report Cards" On Carriers' Streaming Speeds

That's stupid and inefficient. Why should I pay X to pay Y when I can just pay Y directly?

That's actually what Netflix used to do, paying Level3 to reach Comcast. This was awful for everyone. Netflix paid more than they needed to. Comcast's customers had to go over Level3's network to reach Netflix. Level3's peering to Comcast was overloaded. And Level3 and Comcast had to deal with the lopsided data flow that was Netflix's fault. The solution was obvious and simple -- Netflix should just pay Comcast directly, and cut out the middleman. Now Level3's customers have connections to Comcast that aren't overloaded and Comcast customers don't have to use Level3 to reach Netflix.

Comment: Re: Grade is on the curve (Score 1) 110

by JoelKatz (#47408601) Attached to: YouTube Issuing "Report Cards" On Carriers' Streaming Speeds

You are imagining some hypothetical Internet that is nothing like the Internet we actually have. You are correct that ISP's have to deliver data to their customers. But only from sources that do their fair share of the work. If I place a computer in Antarctica, Comcast doesn't have to run a line to Antarctica at no charge to me just because one of their customers wants to reach that machine.

The Internet we actually have grew organically by organizations each doing their part to interconnect with others for mutual benefit. When the costs divide evenly and fairly, settlement-free peering is used and nobody pays anything to anybody else. When the costs don't naturally divide evenly, settlements are used. It has been this way for decades.

Comment: Re: Grade is on the curve (Score 1) 110

by JoelKatz (#47404729) Attached to: YouTube Issuing "Report Cards" On Carriers' Streaming Speeds

Each network pays for half of the costs of transferring the bits. The ISP charges its customers for its half. The data source pays part of its half to the customer's network because they do more than half the work.

Businesses like Netflix and YouTube necessarily emit large amounts of information from a small number of sources to a large number of destinations. This is always going to be much cheaper than delivering traffic to a very large number of destinations, like ISPs have to do. We assume the traffic benefits both networks evenly, so each side should pay half the costs. If not to the ISP, who should Netflix or YouTube pay the costs to? They don't bear them directly, because the traffic is necessarily much cheaper for them to carry than it is for the ISP.

Comment: Re:Where have I seen these claims before? (Score 1) 203

"Note that he claims he built a nuclear reactor when he was 14.."

Which he did. Nobody seriously disputes this claim. He held the record for the youngest person to build a fusion device until Jamie Edwards did so at 13.
http://www.dailymail.co.uk/sci...

Comment: Re:Not surprised (Score 1) 170

by JoelKatz (#47373187) Attached to: Privacy Oversight Board Gives NSA Surveillance a Pass

Equating those we voluntarily choose to associate with to those who we are forced to associate with is about as close as you can come to equating guns with arguments. If you don't like Google or Facebook, you don't have to use them. If you don't like the government, you can't exactly choose the other government.

Comment: Wow (Score 1) 170

by JoelKatz (#47373173) Attached to: Privacy Oversight Board Gives NSA Surveillance a Pass

"Because of the manner in which the NSA conducts upstream collection, and the limits of its current technology, the NSA cannot completely eliminate 'about' communications from its collection without also eliminating a significant portion of the 'to/from' communications that it seeks."

It sounds like this board completely fails to understand how oversight of surveillance is supposed to work. To government can *always* defend a dragnet on the grounds that it takes a dragnet to get the information they want. The purpose of oversight is to ensure they *don't* use dragnets, even where it's the only way to get what they want. The prohibition against general warrants is needed precisely because they can be so effective. The role of surveillance oversight is to prevent oppression, not inefficiency.

Comment: Re:those ARE a problem. Mechanisms, not results (Score 1) 263

by JoelKatz (#47291667) Attached to: The Supreme Court Doesn't Understand Software

You can't get a patent on a result if there are obvious ways to achieve that result. If the result is "rank web pages in order of quality", well anyone can think of a dozen obvious ways to do that. For example, you could have people vote on them or track their usage. I don't believe anyone argues that one should be able to patent a result if there's an obvious way to get that result. The scope of a patent's protection can't include anything obvious. When you file for a patent, you have to set out the legal scope of your patent's protection in the claims, and any claims that can be violated by a solution that would be obvious to people skilled in the art to which the patent pertains are invalid.

The main problem is that things that are obvious, like one-click ordering, are getting patent protection.

Comment: Re:those ARE a problem. Mechanisms, not results (Score 1) 263

by JoelKatz (#47291655) Attached to: The Supreme Court Doesn't Understand Software

The development of the mechanism is the invention. But if you're the first person to figure out a non-obvious way to obtain a result that cannot be obtained any obvious way, then you should be (and mostly are) entitled to protection of that result. The scope of the invention for patent purposes is not the common sense notion of "the invention", nor should it be. This is why patents have a section called "claims" that sets our precisely the legal scope of the invention. They can't quite claim results, but they don't have to claim specific mechanisms either. The law is, justifiable, a complex balance.

If patents lasted forever, your hypothetical about ball point pens would be correct. But patents don't last for all that long. So all that would happen is Birome, in exchange for innovating using a small ball bearing to deposit ink, would get a market lead and about 20 years of exclusivity. But, after that, would have to compete on a level playing field.

It's easy to say that 20 years with just one pen manufacturer is too long. But the alternative could be 40 years with none.

"And do you think (fop that I am) that I could be the Scarlet Pumpernickel?" -- Looney Tunes, The Scarlet Pumpernickel (1950, Chuck Jones)

Working...