Actually, changing to UDP makes things better. With TCP, you have very little control over the transmit pacing and congestion control. With UDP, you have much finer control and immediately change how you are handling one connection based on what's going on with another one.
It remains to be seen if the BitTorrent developers can pull this off though. Many people thing that with pacing and windowing under their control they can do a better job than TCP does. Few actually succeed, since TCP already embodies most of the good idea in this area.
Except the defense did establish innocence, on at least two independent grounds. The problem was that the jury of the citizens was willing to overlook the law and the facts because the person charged was so bad.
Are you seriously arguing that nobody can be wrongfully convicted because the process is perfect and never makes a mistake? If so, I'd love to discuss some famous cases with you, starting with Dred Scott.
The problem is, she didn't enter with the intent to commit a crime. If she had, she would have been charged with that crime. (Did you notice that she was not charged with the crime of harassment, the crime of inciting suicide, or any other crime that makes sense.)
The "crime" she was charged with was violating MySpace's ToS. The "victim" was MySpace.
The death was not a foreseeable consequence of the alcohol use. You might as well hold FedEx responsible for operating a shipping company full well knowing that some people will use it to ship child pornography.
FedEx doesn't know which packages contain child pornography, and shouldn't have to stop running a shipping company just because some of them must. They are not responsible for the cases where their services do wind up transporting child pornography because there is no practical way they can discriminate.
The same is true for alcohol companies. Yes, some of their beverages will wind up killing some people. But they have no way to tell which. They do not need to stop selling alcohol to anyone just because some people will die from it.
Perhaps I was unclear. By "near transit free" I mean they are transit free, except you might find something somewhere that somebody might call transit. (They don't have fiber to Smeknikistan, so they pay someone to carry their traffic to the 2 regional providers there, say.) I mean they are effectively transit free.
If they buy transit from a tier 1, they are a run-of-the-mill single-homed provider that might have a few peers. They are not "near transit free".
It's like if a foundation has "no significant cracking" that doesn't mean you can take pictures of the cracks. They just write that because somebody someplace might find something they call a crack.
"Agreed with the provisio that you should still find out who thier transit providers are and make sure that between them they have transit to at least two tier 1 providers."
They won't have transit providers. There are two possible solutions:
1) Buy from two national near-transit-free providers. This may be expensive but there are no significant caveats.
2) Buy from two regional providers. This is likely to be cheaper, but then you do have to watch out for them relying on common transit providers.
I'm suggesting that option 1 is superior from a reliability standpoint. For one thing, you don't have to worry about shifting peering and transit relationships that can leave you effectively singly-homed to an important national provider.
Buying from one national transit-free provider and one regional provider may be a good compromise, especially if the regional provider does not buy transit from the national provider you picked. However, you do have to worry that shifting relationships may leave you with less redundancy than you expected.
"On the other hand if you are NOT multihomed and want reliability using a tier 1 is risky and using a wanabe tier 1 that likes to get into peering disputes (cogent) will practically gaurantee problems."
So is using any other kind of provider. For example, WV Fiber is a regional provider. But shifting peering and transit relationships left them with Sprint as their only route to Cogent.
It's easy to say "well then don't pick WV Fiber if you care about this", but the problem is that you can't know where the next de-peering or extended outage is going to be.
You could pick two regional providers and then discover, when it matters, that they both just happen to have, at that moment, Sprint as their only route to Cogent.
"You consider it ridiculous to do some research to ensure your backup provider and main provider have between them upstreams to at least two different tier 1 providers? I would consider that basic due dilligence."
They did. Do you still not understand the example?
"Of course the university are being even more stupid for picking cogent as thier sole upstream. This isn't the first time cogent have got into this sort of dispute and i'm 99% sure it won't be the last."
Right, end users should multihome. That's a sensible argument.
"End users and small buisnesses are as you say caught between a rock and a hard place in many cases since often the only decent broadband provider in the localitity is a tier 1 . That makes it even more important for anyone hosting anything important to do thier end properly or risk having thier users (who are probablly also customers) cut off."
Maybe we're in violent agreement. It's hard to tell.
But I think you're still learning the wrong lesson. Buying from two non-tier 1 providers ensures you don't have a direct hop to anyone important. This may protect you from disasters like de-peerings, but it will dramatically increase your vulnerability to more mundane problems like black holes, congested links, and the like.
Buying transit from two major national "near-transit-free" providers is probably your safest option.
I think you missed the example. You are multi-homed. One of your providers is Sprint. The other is a regional provider whose links vary from time to time, but at the moment, they only buy *transit* from Sprint. They have various connections to other networks, but their only route to Cogent is/was through Sprint.
Yes, once you know that the problem is SprintCogent, it seems pretty dumb that your only route to Cogent was through Sprint. But if you don't know where the next disconnection will be, it's awfully tricky to make sure it doesn't hurt you.
Well, we don't know the terms of the agreement. Perhaps the agreement said that Cogent would either have to terminate the peering or pay for it.
The idea is that Sprint wants Cogent to get the blame. So they will say that Cogent forced them to cut the peering.
We'll only know the truth if the terms of the contract are made public (fat chance) or there's a lawsuit that puts these details in the public record.
FWIW, my position is that enforcing peering traffic volume makes perfect sense. Sprint should not peer settlement-free with my personal home web server. But enforcing ratios make no sense. Your customers pay you just as much to receive traffic as to send it. If your customers want to send more than they receive or receive more than they send, then you'll have to live with that.
From a business standpoint, it's idiotic for Sprint to try to pressure Cogent to compete with them.
No, the Internet could not have routed around this. Ask yourself this: who would carry Sprint's Cogent traffic or Cogent's Sprint traffic and how would they be compensated for this service?
Multihoming doesn't fix the problem. It does double your costs.
For example, I know one site that was multi-homed. They had Sprint and a regional provider. The regional provider was de-peered by Cogent about a year ago, and the regional provider only buys transit from Sprint (they peer with many other networks).
Guess what, he couldn't reach the University he just executed a major contract with -- they are single-homed through Cogent.
It's a tricky question whether there was an acceptance of the license or not. Neither the GPL nor the Artistic License require explicit acceptance. So you can't argue there was no acceptance because acceptance was not explicit.
Certainly some copies were made of the code before other things were done with it. If these copies were made without violating the license, that could indicate acceptance of the license.
Pirates are prosecuted under copyright law because they do not have any license to copy or distribute the works. Copyright law prohibits copying or distributing without a license.
Suppose I leave a bicycle unlocked at the front of my house. A sign says "you may borrow my bicycle but have it back within an hour". My neighbor's bicycle is also unlocked but has no such sign. A person who takes my bicycle with the intent of having it back in two hours is doing something very different from a person who takes my neighbor's bicycle with the intent of having it back in two hours.
Copyright infringement and license violation are two very different beasts. The former exists only when there is no agreement that license an operation prohibited under copyright law. The latter exists when the terms of the permission are violated.
This case is about the latter, but the court treated it as the former. This is as absurd as treating someone who fails to return a car in the time period agreed the same as someone who borrows a car from a rental lot with no permission at all.
Let me try an example. Suppose I say you can borrow my car but you can't use it to leave the State. If you do leave the State, can I charge you with the same thing I would charge you with if there was no agreement at all and you took my car without permission?
There is a huge difference between taking a car without permission and taking a car with permission but exceeding the scope of that permission.
Same thing here. There is a huge difference between copyright without any permission at all (infringing copyright) and copying with permission but violating the terms of that permission (infringing the license).
Except you did have permission. You violated the terms of that permission, but you had permission.
Copyright law is for cases where there is no agreement between the parties. Contract law is for cases where there is.
As I already explained, copyright infringement cases follow different rules from contract violation, and this is precisely because there is no agreement between the parties. (This includes a presumption of irreparable harm and statutory damages.)
If the gravamen of the argument is that the contract's terms were violated, and the action would not have been a violation of copyright law had the contract's terms not been violated, then you are suing because the contract's terms were violated.
The exception would be if the action you are suing over exceeded the scope of the license *under* *copyright* *law*. For example, if the license allowed only publication and you created a derivative work.
I think you're missing the point. We agree on what this ruling held. I'm saying that this ruling is inconsistent with prior ruling and what the majority of the legal community believed that the law was. I've even explained why.
You can't point to the ruling to show that the ruling is consistent with prior rulings and prior understanding.
The prior understanding was that if you gave someone a license to do something normally prohibited under copyright law, you gave up the ability to charge them with copyright infringement and gained the ability to charge them with infringing the license.
And there's a very specific reason why things have to be that way -- charging someone with copyright infringement has a different set of legal rules than charging someone with violating a contract. The whole logic behind copyright law is that there is no agreement between the parties.
"If you grant me a license to copy if I agree to certain conditions, and I fail to meet those conditions, then the license is void."
No, that's precisely not true. If you fail to meet those conditions, then you have violated the contract that required you to meet them.
"The court was absolutely correct. When Kamind failed to meet the conditions of their license, it was revoked. From that moment on, they were distributing JMRI illegally."
This is precisely what every other court has held that you cannot do. You cannot use a license to create new violations of copyright law. If the license grants you permission to do something, then doing that something is not a violation of copyright law.
This is so for a very precise reason. Copyright violations are special. You can seek statutory damages. You can presume irreparable harm. None of these things make sense when the action is a mere violation of a contract.
Can you cite *any* other court that has held that a contract can turn failure to comply with license terms into a violation of copyright law?
The only cases I could find were cases where, for example, the license allowed publication but the licensee created derivative works. That make sense -- copyright law makes creating derivative works a distinct right from copying, so you can license one but not the other.
The license is deciding what copyright infringement is, and that's precisely what pretty much every court to consider the issue has found.
Copyright violation is when you act *without* a license and do something that copyright law prohibits. Violating the terms of a license is a mere contract violation.
Copyright violation is very special. It permits you to presume irreparable harm. It permits you to seek statutory damages. Mere violations of contract do not give you these extra powers.
If you have a citation, I'd love to see. But given that you posted anonymously, odds are you are as misinformed as your are cowardly.
You must exceed the scope of a license *under* *copyright* *law* to violate copyright law. For example, if you have a license to publish the book, and you publish it but don't pay me the amount agreed in the license, that's a contract violation. If you create a sequel to the book, that's a copyright violation (since the license never gave you the right to create derivative works).
Every other court has held that a license cannot transform any act other than those prohibited by copyright law into copyright violations. They become mere violations of the license agreement.
No, the District Court had it right. If your license says "you can copy this software provided you do X, Y, and Z", then you have a license to copy the software. No doing X, Y, and Z violates the license, but does not violate copyright law.
I can't make a license that says, "you can copy this software provided you don't pick your nose" and then sue you for copyright infringement for picking your nose.
Because lawsuits for copyright infringement have special powers (like statutory damages and a presumption of irreparable harm), we can't let people decide what's copyright infringement just by writing it in a license. Congress has to do that.
An encrypted connection to an unknown endpoint is no better, and in some ways worse, than an unencrypted connection. The suggested browser changes would seriously weaken security. Users cannot be expected to know which sites are "supposed to" have unauthenticated certificates.
Slashdot Mirror
Actually, changing to UDP makes things better. With TCP, you have very little control over the transmit pacing and congestion control. With UDP, you have much finer control and immediately change how you are handling one connection based on what's going on with another one.
It remains to be seen if the BitTorrent developers can pull this off though. Many people thing that with pacing and windowing under their control they can do a better job than TCP does. Few actually succeed, since TCP already embodies most of the good idea in this area.
Umm, no, that's not true. So long as you have 200 amp service, you can use 200 amps 24x7, any way you want.
Except the defense did establish innocence, on at least two independent grounds. The problem was that the jury of the citizens was willing to overlook the law and the facts because the person charged was so bad.
Are you seriously arguing that nobody can be wrongfully convicted because the process is perfect and never makes a mistake? If so, I'd love to discuss some famous cases with you, starting with Dred Scott.
The problem is, she didn't enter with the intent to commit a crime. If she had, she would have been charged with that crime. (Did you notice that she was not charged with the crime of harassment, the crime of inciting suicide, or any other crime that makes sense.)
The "crime" she was charged with was violating MySpace's ToS. The "victim" was MySpace.
Actually that's not true. Red wine is pretty darned close to a miracle food. In any event, you could make the same argument about skis.
The death was not a foreseeable consequence of the alcohol use. You might as well hold FedEx responsible for operating a shipping company full well knowing that some people will use it to ship child pornography.
FedEx doesn't know which packages contain child pornography, and shouldn't have to stop running a shipping company just because some of them must. They are not responsible for the cases where their services do wind up transporting child pornography because there is no practical way they can discriminate.
The same is true for alcohol companies. Yes, some of their beverages will wind up killing some people. But they have no way to tell which. They do not need to stop selling alcohol to anyone just because some people will die from it.
Perhaps I was unclear. By "near transit free" I mean they are transit free, except you might find something somewhere that somebody might call transit. (They don't have fiber to Smeknikistan, so they pay someone to carry their traffic to the 2 regional providers there, say.) I mean they are effectively transit free.
If they buy transit from a tier 1, they are a run-of-the-mill single-homed provider that might have a few peers. They are not "near transit free".
It's like if a foundation has "no significant cracking" that doesn't mean you can take pictures of the cracks. They just write that because somebody someplace might find something they call a crack.
"Agreed with the provisio that you should still find out who thier transit providers are and make sure that between them they have transit to at least two tier 1 providers."
They won't have transit providers. There are two possible solutions:
1) Buy from two national near-transit-free providers. This may be expensive but there are no significant caveats.
2) Buy from two regional providers. This is likely to be cheaper, but then you do have to watch out for them relying on common transit providers.
I'm suggesting that option 1 is superior from a reliability standpoint. For one thing, you don't have to worry about shifting peering and transit relationships that can leave you effectively singly-homed to an important national provider.
Buying from one national transit-free provider and one regional provider may be a good compromise, especially if the regional provider does not buy transit from the national provider you picked. However, you do have to worry that shifting relationships may leave you with less redundancy than you expected.
"On the other hand if you are NOT multihomed and want reliability using a tier 1 is risky and using a wanabe tier 1 that likes to get into peering disputes (cogent) will practically gaurantee problems."
So is using any other kind of provider. For example, WV Fiber is a regional provider. But shifting peering and transit relationships left them with Sprint as their only route to Cogent.
It's easy to say "well then don't pick WV Fiber if you care about this", but the problem is that you can't know where the next de-peering or extended outage is going to be.
You could pick two regional providers and then discover, when it matters, that they both just happen to have, at that moment, Sprint as their only route to Cogent.
"You consider it ridiculous to do some research to ensure your backup provider and main provider have between them upstreams to at least two different tier 1 providers? I would consider that basic due dilligence."
They did. Do you still not understand the example?
"Of course the university are being even more stupid for picking cogent as thier sole upstream. This isn't the first time cogent have got into this sort of dispute and i'm 99% sure it won't be the last."
Right, end users should multihome. That's a sensible argument.
"End users and small buisnesses are as you say caught between a rock and a hard place in many cases since often the only decent broadband provider in the localitity is a tier 1 . That makes it even more important for anyone hosting anything important to do thier end properly or risk having thier users (who are probablly also customers) cut off."
Maybe we're in violent agreement. It's hard to tell.
But I think you're still learning the wrong lesson. Buying from two non-tier 1 providers ensures you don't have a direct hop to anyone important. This may protect you from disasters like de-peerings, but it will dramatically increase your vulnerability to more mundane problems like black holes, congested links, and the like.
Buying transit from two major national "near-transit-free" providers is probably your safest option.
It's always fun to blame the victim by coming up with more and more ridiculous things they could have or should have done.
I think you missed the example. You are multi-homed. One of your providers is Sprint. The other is a regional provider whose links vary from time to time, but at the moment, they only buy *transit* from Sprint. They have various connections to other networks, but their only route to Cogent is/was through Sprint.
Yes, once you know that the problem is SprintCogent, it seems pretty dumb that your only route to Cogent was through Sprint. But if you don't know where the next disconnection will be, it's awfully tricky to make sure it doesn't hurt you.
Well, we don't know the terms of the agreement. Perhaps the agreement said that Cogent would either have to terminate the peering or pay for it.
The idea is that Sprint wants Cogent to get the blame. So they will say that Cogent forced them to cut the peering.
We'll only know the truth if the terms of the contract are made public (fat chance) or there's a lawsuit that puts these details in the public record.
FWIW, my position is that enforcing peering traffic volume makes perfect sense. Sprint should not peer settlement-free with my personal home web server. But enforcing ratios make no sense. Your customers pay you just as much to receive traffic as to send it. If your customers want to send more than they receive or receive more than they send, then you'll have to live with that.
From a business standpoint, it's idiotic for Sprint to try to pressure Cogent to compete with them.
No, the Internet could not have routed around this. Ask yourself this: who would carry Sprint's Cogent traffic or Cogent's Sprint traffic and how would they be compensated for this service?
Multihoming doesn't fix the problem. It does double your costs.
For example, I know one site that was multi-homed. They had Sprint and a regional provider. The regional provider was de-peered by Cogent about a year ago, and the regional provider only buys transit from Sprint (they peer with many other networks).
Guess what, he couldn't reach the University he just executed a major contract with -- they are single-homed through Cogent.
And what do end users do? Multihome?
"For something to be computationally infeasible, it needs to have 80 bits of entropy or more."
Except this isn't about computational feasibility.
The attack uses syn cookies. So your suggestion is to disable them on your end.
So if someone attacks you with a gun, the fix is to put down your gun?
Did you really think this through?
It's a tricky question whether there was an acceptance of the license or not. Neither the GPL nor the Artistic License require explicit acceptance. So you can't argue there was no acceptance because acceptance was not explicit.
Certainly some copies were made of the code before other things were done with it. If these copies were made without violating the license, that could indicate acceptance of the license.
Pirates are prosecuted under copyright law because they do not have any license to copy or distribute the works. Copyright law prohibits copying or distributing without a license.
Suppose I leave a bicycle unlocked at the front of my house. A sign says "you may borrow my bicycle but have it back within an hour". My neighbor's bicycle is also unlocked but has no such sign. A person who takes my bicycle with the intent of having it back in two hours is doing something very different from a person who takes my neighbor's bicycle with the intent of having it back in two hours.
Copyright infringement and license violation are two very different beasts. The former exists only when there is no agreement that license an operation prohibited under copyright law. The latter exists when the terms of the permission are violated.
This case is about the latter, but the court treated it as the former. This is as absurd as treating someone who fails to return a car in the time period agreed the same as someone who borrows a car from a rental lot with no permission at all.
Let me try an example. Suppose I say you can borrow my car but you can't use it to leave the State. If you do leave the State, can I charge you with the same thing I would charge you with if there was no agreement at all and you took my car without permission?
There is a huge difference between taking a car without permission and taking a car with permission but exceeding the scope of that permission.
Same thing here. There is a huge difference between copyright without any permission at all (infringing copyright) and copying with permission but violating the terms of that permission (infringing the license).
Except you did have permission. You violated the terms of that permission, but you had permission.
Copyright law is for cases where there is no agreement between the parties. Contract law is for cases where there is.
As I already explained, copyright infringement cases follow different rules from contract violation, and this is precisely because there is no agreement between the parties. (This includes a presumption of irreparable harm and statutory damages.)
If the gravamen of the argument is that the contract's terms were violated, and the action would not have been a violation of copyright law had the contract's terms not been violated, then you are suing because the contract's terms were violated.
The exception would be if the action you are suing over exceeded the scope of the license *under* *copyright* *law*. For example, if the license allowed only publication and you created a derivative work.
I think you're missing the point. We agree on what this ruling held. I'm saying that this ruling is inconsistent with prior ruling and what the majority of the legal community believed that the law was. I've even explained why.
You can't point to the ruling to show that the ruling is consistent with prior rulings and prior understanding.
The prior understanding was that if you gave someone a license to do something normally prohibited under copyright law, you gave up the ability to charge them with copyright infringement and gained the ability to charge them with infringing the license.
And there's a very specific reason why things have to be that way -- charging someone with copyright infringement has a different set of legal rules than charging someone with violating a contract. The whole logic behind copyright law is that there is no agreement between the parties.
"If you grant me a license to copy if I agree to certain conditions, and I fail to meet those conditions, then the license is void."
No, that's precisely not true. If you fail to meet those conditions, then you have violated the contract that required you to meet them.
"The court was absolutely correct. When Kamind failed to meet the conditions of their license, it was revoked. From that moment on, they were distributing JMRI illegally."
This is precisely what every other court has held that you cannot do. You cannot use a license to create new violations of copyright law. If the license grants you permission to do something, then doing that something is not a violation of copyright law.
This is so for a very precise reason. Copyright violations are special. You can seek statutory damages. You can presume irreparable harm. None of these things make sense when the action is a mere violation of a contract.
Can you cite *any* other court that has held that a contract can turn failure to comply with license terms into a violation of copyright law?
The only cases I could find were cases where, for example, the license allowed publication but the licensee created derivative works. That make sense -- copyright law makes creating derivative works a distinct right from copying, so you can license one but not the other.
Put up or shut up.
The license is deciding what copyright infringement is, and that's precisely what pretty much every court to consider the issue has found.
Copyright violation is when you act *without* a license and do something that copyright law prohibits. Violating the terms of a license is a mere contract violation.
Copyright violation is very special. It permits you to presume irreparable harm. It permits you to seek statutory damages. Mere violations of contract do not give you these extra powers.
If you have a citation, I'd love to see. But given that you posted anonymously, odds are you are as misinformed as your are cowardly.
You must exceed the scope of a license *under* *copyright* *law* to violate copyright law. For example, if you have a license to publish the book, and you publish it but don't pay me the amount agreed in the license, that's a contract violation. If you create a sequel to the book, that's a copyright violation (since the license never gave you the right to create derivative works).
Every other court has held that a license cannot transform any act other than those prohibited by copyright law into copyright violations. They become mere violations of the license agreement.
No, the District Court had it right. If your license says "you can copy this software provided you do X, Y, and Z", then you have a license to copy the software. No doing X, Y, and Z violates the license, but does not violate copyright law.
I can't make a license that says, "you can copy this software provided you don't pick your nose" and then sue you for copyright infringement for picking your nose.
Because lawsuits for copyright infringement have special powers (like statutory damages and a presumption of irreparable harm), we can't let people decide what's copyright infringement just by writing it in a license. Congress has to do that.
This ruling is very wrong and very troubling.
An encrypted connection to an unknown endpoint is no better, and in some ways worse, than an unencrypted connection. The suggested browser changes would seriously weaken security. Users cannot be expected to know which sites are "supposed to" have unauthenticated certificates.
Yeah, go to an unsecured web connection and download a key that you're going to ultimately trust. That's a good plan.
If CACert doesn't provide a secure way to get their root key, they're not serious.