There are lots of reasons why fair use might need the pure form of the work rather than a video recording of a TV screen.
What if you wanted to do a scholarly linguistics study of the evolution of language in movies and chose to use the closed captioning information? You could access that digitally, and do all sorts of easy lexical analysis, which would require difficult and unreliable OCR software otherwise.
Maybe you've got a machine with multiple DVD drives and you want to scan the works in parallel looking for select phrases and maybe correlate the occurances of those phrases with lighting, or music used in the film at that point?
The pure digital form makes detailed study of a work much simpler in many ways, and eliminating access to the pure form may prevent scholarly works which would have otherwise been possible. It's not that you can't do such studies, but that the time required to do them becomes prohibitively large.
I've been using debian for over five years and I have *never* had any serious problems installing on any machine I've ever used.
The installer is extremely flexible. At any point during the installation, you can stop what you're doing and go back to an earlier step of the procedure, or skip ahead to future steps. If what you want to do isn't supported in the menu, the installer can spawn an ash shell and get a command line on the machine even before you've partitioned or formatted a disk. You can handle just about any special case you can imagine by hand at the shell prompt. You could do the entire installation that way if you like! It's true the Debian's installer isn't the prettiest or the easiest, but it's one of the most powerful and flexible there is. You can't get much more flexible than a shell prompt!
If you read the article, it says that the images came from the Mars Global Surveyor, a NASA satellite orbitting Mars.
It's not all that freaky. You've got one agency with a lot of experience doing satellite photo enhancement to pick out small details and another agency with a lot of experience throwing things very far. It's hardly surprising that NIMA did the better job here. The only thing that's changed is that the satellite is orbitting a different planet, otherwise, it's right up their alley.
One way they are going to measure neutrinos other than electron neutrinos is to add salt to the water. I think that makes all the neutrinos interact.
Neutrinos interact with nucleons. Protons and neutrons. The neutrinos don't care what type of nucleus these nucleons are a part of. Neutrino interactions on salt will occur just as much as neutrino interactions on water, once you account for the relative numbers of nucleons.
Adding salt can help you see neutrino interactions that occur because Chlorine can undergo a process called neutron capture which can produce light which can aid in the detection of netrino interactions. It doesn't increase the number of neutrino interactions, it just helps you see them.
The point you missed, though, was that there are no muon type or tau type neutrinos coming from the sun. They're simply not there. The sun doesn't make them. All of the non-electron neutrinos come from cosmic rays, which are typically much higher in energy than solar neutrinos, and SNO is not well optimized to see these. It's too small, and you can't change the size by adding a little salt.
The experement you link to is a smaller scale version of the one in the main article.
Uh, no, it's not. SNO is a sphere with a 6 meter radius, and SuperKamiokande is a cylinder with a 20 meter radius and 40 meter height. SNO is the small one.
The experiments aren't really comparable, though. The detectors use different targets. SuperK uses ordinary water, and SNO uses heavy water. They're designed to measure different energy ranges. In short, they see different neutrinos.
several facillities have measure the neutrino mass (actually, the mass difference between two types of neutrinos),
Actually, the difference in the squares of the the masses of two types of neutrinos.
no two experements agree
This is not true either. First of all, there are at least three types of neutrinos. If you take three types of neutrinos, and pair them, you can arrange them in three different pairs so you can measure three different mass differences which are all different but that's not a disagreement, you've just measured different things. Second of all, not all experiments disagree!
Also, the experemental methods have had several shortcommings, such as the inability of the detectors to see tau neutrinos, and low efficiency.
Low efficiency is not a shortcomming. There's no shortage of neutrinos flying around, so the fact that you miss most of them is a blessing, not a curse.
And some detectors can see tau neutrinos, such as AMANDA.
This experement, if successful should detect a large percentage of the solar neutrinos, and more importantly, all three types.
All solar neutrinos are electron type, so your statement makes little sense. SNO has no ability to distinguish neutrino flavor. It was designed and optimized to measure electron type solar neutrinos, and that's pretty much all it does.
SNO is a little different than SuperK because it's designed to see different types of neutrinos.
In SuperKamiokande, the target is simply water. Neutrinos come in, scatter off a nucleon, and produce a charged lepton (electron, muon, or tau) and this charged lepton produces cherenkov radiation. The works well for superk, since its goal is to measure neutrinos with energies above 5 MeV. Below that energy, the charged leptons just aren't energetic enough to travel far enough to produce enough light to be detected.
The purpose of SNO is to push the energy threshold down lower, to around 2 MeV. This can be done by adding a chemical scintillator to the target. Chemical scintillators become excited by the passage of high energy particles and emit light, so a typical low energy reaction in SNO will produced more light than one in SuperK.
Lowering the energy threshold is import if you want to study what goes on in the Sun because the fusion processes taking place produce a very particular neutrino spectrum. Because the fusion of hydrogen into helium doesn't occur in a single step, you can distinguish neutrinos coming from different stages of the reaction by their enery. Most of the neutrinos emitted from the sun are actually around 1 MeV and aren't energetic enough to even be seen by SNO, but SNO does a good job of observing more of the spectrum than previously seen by an experiment on this scale.
The experiment with the artificial neutrino beam that you refer to is K2K and it's still on going. Although K2K's results so far are not inconsistant with neutrino oscillations, it's just too early to tell. K2K will run for another three years to get enough data to make a conclusive statement.
You've applied this equation incorrectly. Yes, P=V^2/R, but V is the voltage difference between the potential at one end of the wire and the potential at the other. If the power plant is producing 500KV, and the potential on the wire when it reaches the transformer is 499 KV, then you're only loosing 1KV^2/1Ohm, not 500KV^2/1Ohm, which would be the total power used including line loss plus the power used by the consumer which is not inefficiency.
We had 20 operators drive this truck with curtains on all the windows -- totally blind. They each drove a four-mile course with hairpin curves, right-angle curves and an S-curve, and we had only one small incident...
...when a small boy ran across the road and got run over by the plow. He wasn't on the map and kids are notorious for their small radar profile. They're practically invisible!
Well, it's really more than that. It's the difference between the concept of "free" and "copyleft".
Works placed in the public domain or under a BSD-style license are unquestionably free, but they are not copyleft.
Copyleft aims to take a work, make it free, and keep it free. The GPL is a form of copyleft because once code is under the GPL, you can't change the terms of the license, it's free forever.
The BSD license is not a copyleft because additional restrictions may be added which make it non-free. This is why the GPL does not allow relicencing under another license, even if that license is free, because if you could get from GPL to BSD, then non-free is just one more hop away.
License conflict occur when you mix two different, incompatible licenses in the same program. Not all licenses are incompatible.
The GPL requires that all works derived from works distributed under the GPL also be distributed under the terms of the GPL.
If I'm writing an application, and I want to include some code in it which I received under the terms of the GPL, I can do that as long as I license my work under the GPL. If I want to include some code which I received under the QPL, I can do that, but then I have to use the QPL or some other compatible license. I cannot satisfy both of these requirements at the same time, so I cannot distribute a work which is simulateously derived from code I received under the GPL and code I received under the QPL.
However, if I write two different programs, and one used GPL'd code and one uses QPL'd code, I can GPL the first, QPL the second, and distribute them both on the same CD if I like. That's allowed.
The ASP loophole exists because the ASP provider never distributes a copy of the application to the user. In general, copyright law covers the creation and distribution of copies of works, but for certain kinds of works, there is also a restriction on perfomance of the work.
Take for example the play "Cats". Cats as literary work is copyrighted. You can't make copies of the script and sell them unless you're licensed to do so by the copyright holder. However, you also can't put on an independent production of the play and charge admission, even though the audience would not be actually receiving a copy of Cats, they would be perceiving a performance of it. That's controlled by copyright law.
What RMS is trying to do is to equate ASP to a performance of the software. It's a good analogy. The consumer never gets a true copy of the work, but they perceive all that is important about it. If ASP can be equated with a performance, then it can be controlled within the existing framework of U.S. copyright law.
The current GPL makes it very hard to distribute GPL and non-GPL programs together
There is no restriction in the GPL against shipping works not derived from GPL code aggregated with works licensed under the GPL, unless the GPL'd work depends upon the non-GPL'd work. Simple aggregation of independent works covered under different licenses is perfectly legal.
Chances are pretty good that someone created that desktop in Macromedia Director. Computer screens are done this way all the time for film and television.
You're probably right, but it's also largely irrelevant. It doesn't matter so much what was really driving the display as much as what the director wanted to portray.
The point is that someone in Hollywood thinks that the uber-hackers of the world use Linux or something similar and chose to express that. That's more important than the methods by which the images were created.
This isn't video compression at all. This is geometric data compression. Comparing it to MPEG4 is really misleading. These things are apples and oranges.
This is like finding a better way for the Quake client and server to talk to each other, not a better way to stream The Matrix to your monitor.
Why do our elected officials feel it is necessary to pass laws which protect us from ourself? If someone wants to gamble away their money, that is their choice, and the only one they are potentially hurting is themself.
That's true, if and only if they're actually gambling with their money. Unfortnately, that's not always the case. Frequently, people will use their credit cards to rack up huge debts, be unable to pay and end up declaring bankruptcy. The credit card companies lose out and then pass that loss off onto other customers in the form of higher interest rates.
This is the kind of behavior that we need laws to help prevent.
This doesn't make any sense. Sure, the pads are random, you can distribute the pads, but you still need to distribute the information that combining certain pads in a certain way gives you a certain message.
If you could censor the delivery of the message, you could censor the delivery of the list of pads needed to create the message.
All you're doing is putting the information into a new form. It's the pad list which becomes the important piece of information here and it's precisely the pad list which is completely unprotected by this scheme.
Isn't evidence that is illegally obtained, not valid for use in a court of law?
That's true if it's obtained by the government (or an agent acting on behalf of the government). Private citizens have more leeway in how they can collect evidence.
Why would anyone in their right mind make a new design based on the ISA bus ?
Because there are a few motherboards out there which don't support PCI in the back-up BIOS.
I used to have one. A Tyan Titan III (S1468 I think). It was a standard Pentium mobo, with mixed PCI/ISA. The BIOS was stored on EEPROM and you could flash a new bios onto the board if you wanted. Lots of motherboards support that. But, there was also a back-up BIOS in real ROM, unflashable. If the flashed bios failed a checksum, then it booted with a very tiny backup bios. Because the backup bios was so small, support for the PCI bus was not included. The only function for this bios was to be able to boot off a floppy and flash a new real bios.
So, one day, something happened to my machine, and the bios got hosed. Checksum error. But, with my fancy PCI graphics card, I couldn't see anything. I had to get an old ISA card and swap it in temporarily while I flashed a new bios.
As long as your mobo has an ISA slot on it, then ISA is the way to go for something like this. It's far simpler to support, and, when things start going wrong with a system, the ISA bus will be more dependable than the PCI bus.
However, high end boards are now shipping PCI only, so, they really should have two versions of the card. That would be the best solution.
While I can envision some cases where having serial IO would be valuable so early in the boot process, the fact is that monitors and keyboards are now so cheap, abundant, and portable as to render serial IO archaic.
Pshaw!
I'm currently in Japan, but relying on my server in New York to keep running. If something goes seriously wrong with it, a card like this would be a lifesaver.
I could keep an emergency install CD in the CDROM drive, but keep the boot device as the hard disk. From half the world away I could completely reinstall the OS if I needed to!
Mightn't it be smart to mandate that modifications ARE sent back IF the updates are to be used in a commercial (vs. private or custom) distribution?
Why should that matter? And what is a commercial distribution, anyway? Is Debian a commercial distribution? What if Debian is resold by cheap bytes?
One of the problems with trying to dictate separate terms for "commercial use" is that commercial use is often a slippery term. You end up doing yourself harm becuase some people who you might want to allow to distribute the code will choose not to just to avoid potential legal problems. And what do you gain? Nothing.
a licence where anyone can download and work on the code and distribute it as much as they want, but have them send us back the modifications that they have done for us to decide whether or not to include them in the offical release.
I don't think it's in your best interest to try to force this with a license clause. In other words, don't try to force people to send you their modifications or not distribute them on their own.
For one thing, most people will do what you want anyway. Maintaining software can be a lot of work, and, if I have a modification I make to some code, it's a lot simpler and easier for me to just send it to you for inclusion than to maintain it myself or to start a parallel distribution. It just doesn't make sense to do that.
So, since it doesn't make sense for me to not do what you want, you don't have to legally mandate it in a license. Mandating that patches be sent to you would make your code non-free, using the open source definition, so, that would put off a lot of potential developers.
The GPL would probably serve you well and probably much better than anything you could cook up yourself.
Slashdot Mirror
There are lots of reasons why fair use might need the pure form of the work rather than a video recording of a TV screen.
What if you wanted to do a scholarly linguistics study of the evolution of language in movies and chose to use the closed captioning information? You could access that digitally, and do all sorts of easy lexical analysis, which would require difficult and unreliable OCR software otherwise.
Maybe you've got a machine with multiple DVD drives and you want to scan the works in parallel looking for select phrases and maybe correlate the occurances of those phrases with lighting, or music used in the film at that point?
The pure digital form makes detailed study of a work much simpler in many ways, and eliminating access to the pure form may prevent scholarly works which would have otherwise been possible. It's not that you can't do such studies, but that the time required to do them becomes prohibitively large.
I've been using debian for over five years and I have *never* had any serious problems installing on any machine I've ever used.
The installer is extremely flexible. At any point during the installation, you can stop what you're doing and go back to an earlier step of the procedure, or skip ahead to future steps. If what you want to do isn't supported in the menu, the installer can spawn an ash shell and get a command line on the machine even before you've partitioned or formatted a disk. You can handle just about any special case you can imagine by hand at the shell prompt. You could do the entire installation that way if you like! It's true the Debian's installer isn't the prettiest or the easiest, but it's one of the most powerful and flexible there is. You can't get much more flexible than a shell prompt!
If you read the article, it says that the images came from the Mars Global Surveyor, a NASA satellite orbitting Mars.
It's not all that freaky. You've got one agency with a lot of experience doing satellite photo enhancement to pick out small details and another agency with a lot of experience throwing things very far. It's hardly surprising that NIMA did the better job here. The only thing that's changed is that the satellite is orbitting a different planet, otherwise, it's right up their alley.
One way they are going to measure neutrinos other than electron neutrinos is to add salt to the water. I think that makes all the neutrinos interact.
Neutrinos interact with nucleons. Protons and neutrons. The neutrinos don't care what type of nucleus these nucleons are a part of. Neutrino interactions on salt will occur just as much as neutrino interactions on water, once you account for the relative numbers of nucleons.
Adding salt can help you see neutrino interactions that occur because Chlorine can undergo a process called neutron capture which can produce light which can aid in the detection of netrino interactions. It doesn't increase the number of neutrino interactions, it just helps you see them.
The point you missed, though, was that there are no muon type or tau type neutrinos coming from the sun. They're simply not there. The sun doesn't make them. All of the non-electron neutrinos come from cosmic rays, which are typically much higher in energy than solar neutrinos, and SNO is not well optimized to see these. It's too small, and you can't change the size by adding a little salt.
Mea culpa.
SNO doesn't actually use a scintillator. That's KAMLAND I was thinking of. SNO just keeps itself small and very clean to keep the background down.
The experement you link to is a smaller scale version of the one in the main article.
Uh, no, it's not. SNO is a sphere with a 6 meter radius, and SuperKamiokande is a cylinder with a 20 meter radius and 40 meter height. SNO is the small one.
The experiments aren't really comparable, though. The detectors use different targets. SuperK uses ordinary water, and SNO uses heavy water. They're designed to measure different energy ranges. In short, they see different neutrinos.
several facillities have measure the neutrino mass (actually, the mass difference between two types of neutrinos),
Actually, the difference in the squares of the the masses of two types of neutrinos.
no two experements agree
This is not true either. First of all, there are at least three types of neutrinos. If you take three types of neutrinos, and pair them, you can arrange them in three different pairs so you can measure three different mass differences which are all different but that's not a disagreement, you've just measured different things. Second of all, not all experiments disagree!
Also, the experemental methods have had several shortcommings, such as the inability of the detectors to see tau neutrinos, and low efficiency.
Low efficiency is not a shortcomming. There's no shortage of neutrinos flying around, so the fact that you miss most of them is a blessing, not a curse.
And some detectors can see tau neutrinos, such as AMANDA.
This experement, if successful should detect a large percentage of the solar neutrinos, and more importantly, all three types.
All solar neutrinos are electron type, so your statement makes little sense. SNO has no ability to distinguish neutrino flavor. It was designed and optimized to measure electron type solar neutrinos, and that's pretty much all it does.
SNO is a little different than SuperK because it's designed to see different types of neutrinos.
In SuperKamiokande, the target is simply water. Neutrinos come in, scatter off a nucleon, and produce a charged lepton (electron, muon, or tau) and this charged lepton produces cherenkov radiation. The works well for superk, since its goal is to measure neutrinos with energies above 5 MeV. Below that energy, the charged leptons just aren't energetic enough to travel far enough to produce enough light to be detected.
The purpose of SNO is to push the energy threshold down lower, to around 2 MeV. This can be done by adding a chemical scintillator to the target. Chemical scintillators become excited by the passage of high energy particles and emit light, so a typical low energy reaction in SNO will produced more light than one in SuperK.
Lowering the energy threshold is import if you want to study what goes on in the Sun because the fusion processes taking place produce a very particular neutrino spectrum. Because the fusion of hydrogen into helium doesn't occur in a single step, you can distinguish neutrinos coming from different stages of the reaction by their enery. Most of the neutrinos emitted from the sun are actually around 1 MeV and aren't energetic enough to even be seen by SNO, but SNO does a good job of observing more of the spectrum than previously seen by an experiment on this scale.
The experiment with the artificial neutrino beam that you refer to is K2K and it's still on going. Although K2K's results so far are not inconsistant with neutrino oscillations, it's just too early to tell. K2K will run for another three years to get enough data to make a conclusive statement.
You've applied this equation incorrectly. Yes, P=V^2/R, but V is the voltage difference between the potential at one end of the wire and the potential at the other. If the power plant is producing 500KV, and the potential on the wire when it reaches the transformer is 499 KV, then you're only loosing 1KV^2/1Ohm, not 500KV^2/1Ohm, which would be the total power used including line loss plus the power used by the consumer which is not inefficiency.
We had 20 operators drive this truck with curtains on all the windows -- totally blind. They each drove a four-mile course with hairpin curves, right-angle curves and an S-curve, and we had only one small incident...
...when a small boy ran across the road and got run over by the plow. He wasn't on the map and kids are notorious for their small radar profile. They're practically invisible!
I believe until either this or last year, they didn't accept "Linux" as an answer for "Give an example of an operating system."
That's because the correct answer is "GNU/Linux".
Well, it's really more than that. It's the difference between the concept of "free" and "copyleft".
Works placed in the public domain or under a BSD-style license are unquestionably free, but they are not copyleft.
Copyleft aims to take a work, make it free, and keep it free. The GPL is a form of copyleft because once code is under the GPL, you can't change the terms of the license, it's free forever.
The BSD license is not a copyleft because additional restrictions may be added which make it non-free. This is why the GPL does not allow relicencing under another license, even if that license is free, because if you could get from GPL to BSD, then non-free is just one more hop away.
License conflict occur when you mix two different, incompatible licenses in the same program. Not all licenses are incompatible.
The GPL requires that all works derived from works distributed under the GPL also be distributed under the terms of the GPL.
If I'm writing an application, and I want to include some code in it which I received under the terms of the GPL, I can do that as long as I license my work under the GPL. If I want to include some code which I received under the QPL, I can do that, but then I have to use the QPL or some other compatible license. I cannot satisfy both of these requirements at the same time, so I cannot distribute a work which is simulateously derived from code I received under the GPL and code I received under the QPL.
However, if I write two different programs, and one used GPL'd code and one uses QPL'd code, I can GPL the first, QPL the second, and distribute them both on the same CD if I like. That's allowed.
The ASP loophole exists because the ASP provider never distributes a copy of the application to the user. In general, copyright law covers the creation and distribution of copies of works, but for certain kinds of works, there is also a restriction on perfomance of the work.
Take for example the play "Cats". Cats as literary work is copyrighted. You can't make copies of the script and sell them unless you're licensed to do so by the copyright holder. However, you also can't put on an independent production of the play and charge admission, even though the audience would not be actually receiving a copy of Cats, they would be perceiving a performance of it. That's controlled by copyright law.
What RMS is trying to do is to equate ASP to a performance of the software. It's a good analogy. The consumer never gets a true copy of the work, but they perceive all that is important about it. If ASP can be equated with a performance, then it can be controlled within the existing framework of U.S. copyright law.
The current GPL makes it very hard to distribute GPL and non-GPL programs together
There is no restriction in the GPL against shipping works not derived from GPL code aggregated with works licensed under the GPL, unless the GPL'd work depends upon the non-GPL'd work. Simple aggregation of independent works covered under different licenses is perfectly legal.
Can you clarify what you think is the problem?
Chances are pretty good that someone created that desktop in Macromedia Director. Computer screens are done this way all the time for film and television.
You're probably right, but it's also largely irrelevant. It doesn't matter so much what was really driving the display as much as what the director wanted to portray.
The point is that someone in Hollywood thinks that the uber-hackers of the world use Linux or something similar and chose to express that. That's more important than the methods by which the images were created.
This isn't video compression at all. This is geometric data compression. Comparing it to MPEG4 is really misleading. These things are apples and oranges.
This is like finding a better way for the Quake client and server to talk to each other, not a better way to stream The Matrix to your monitor.
Why do our elected officials feel it is necessary to pass laws which protect us from ourself? If someone wants to gamble away their money, that is their choice, and the only one they are potentially hurting is themself.
That's true, if and only if they're actually gambling with their money. Unfortnately, that's not always the case. Frequently, people will use their credit cards to rack up huge debts, be unable to pay and end up declaring bankruptcy. The credit card companies lose out and then pass that loss off onto other customers in the form of higher interest rates.
This is the kind of behavior that we need laws to help prevent.
This doesn't make any sense. Sure, the pads are random, you can distribute the pads, but you still need to distribute the information that combining certain pads in a certain way gives you a certain message.
If you could censor the delivery of the message, you could censor the delivery of the list of pads needed to create the message.
All you're doing is putting the information into a new form. It's the pad list which becomes the important piece of information here and it's precisely the pad list which is completely unprotected by this scheme.
It sounds pretty useless to me.
Isn't evidence that is illegally obtained, not valid for use in a court of law?
That's true if it's obtained by the government (or an agent acting on behalf of the government). Private citizens have more leeway in how they can collect evidence.
Why would anyone in their right mind make a new design based on the ISA bus ?
Because there are a few motherboards out there which don't support PCI in the back-up BIOS.
I used to have one. A Tyan Titan III (S1468 I think). It was a standard Pentium mobo, with mixed PCI/ISA. The BIOS was stored on EEPROM and you could flash a new bios onto the board if you wanted. Lots of motherboards support that. But, there was also a back-up BIOS in real ROM, unflashable. If the flashed bios failed a checksum, then it booted with a very tiny backup bios. Because the backup bios was so small, support for the PCI bus was not included. The only function for this bios was to be able to boot off a floppy and flash a new real bios.
So, one day, something happened to my machine, and the bios got hosed. Checksum error. But, with my fancy PCI graphics card, I couldn't see anything. I had to get an old ISA card and swap it in temporarily while I flashed a new bios.
As long as your mobo has an ISA slot on it, then ISA is the way to go for something like this. It's far simpler to support, and, when things start going wrong with a system, the ISA bus will be more dependable than the PCI bus.
However, high end boards are now shipping PCI only, so, they really should have two versions of the card. That would be the best solution.
While I can envision some cases where having serial IO would be valuable so early in the boot process, the fact is that monitors and keyboards are now so cheap, abundant, and portable as to render serial IO archaic.
Pshaw!
I'm currently in Japan, but relying on my server in New York to keep running. If something goes seriously wrong with it, a card like this would be a lifesaver.
I could keep an emergency install CD in the CDROM drive, but keep the boot device as the hard disk. From half the world away I could completely reinstall the OS if I needed to!
There's nothing archaic about that!
Excuse me for just a moment.
Bwa ha ha ha ha
Okay, now that that's out of my system, please allow me to explain about the GIMPs #1 and 2 shortcomings
We're working on it. What are you doing besides laughing?
http://sourceforge.net/project/?group_id=1537
Mightn't it be smart to mandate that modifications ARE sent back IF the updates are to be used in a commercial (vs. private or custom) distribution?
Why should that matter? And what is a commercial distribution, anyway? Is Debian a commercial distribution? What if Debian is resold by cheap bytes?
One of the problems with trying to dictate separate terms for "commercial use" is that commercial use is often a slippery term. You end up doing yourself harm becuase some people who you might want to allow to distribute the code will choose not to just to avoid potential legal problems. And what do you gain? Nothing.
a licence where anyone can download and work on the code and distribute it as much as they want, but have them send us back the modifications that they have done for us to decide whether or not to include them in the offical release.
I don't think it's in your best interest to try to force this with a license clause. In other words, don't try to force people to send you their modifications or not distribute them on their own.
For one thing, most people will do what you want anyway. Maintaining software can be a lot of work, and, if I have a modification I make to some code, it's a lot simpler and easier for me to just send it to you for inclusion than to maintain it myself or to start a parallel distribution. It just doesn't make sense to do that.
So, since it doesn't make sense for me to not do what you want, you don't have to legally mandate it in a license. Mandating that patches be sent to you would make your code non-free, using the open source definition, so, that would put off a lot of potential developers.
The GPL would probably serve you well and probably much better than anything you could cook up yourself.
It seems a rare event these days when a company can actually publicly admit, in no uncertain terms, that they made a mistake.