It's nonsense for the reasons I stated. Your point about dynamic linking creating a derivative work (or not) is also potentially valid, but it's moot unless RTS is lying.
The summary questions Google's motivations for doing this, but I think it should be clear this isn't a Google project, really. 20% projects can't be totally random, personal things that have no relationship whatsoever with the business or possible business... but the link can be very tenuous, and the cooler the project is, the weaker it can be. All tech managers at Google are engineers themselves and tend to be just as able to geek out about cool stuff as the people they supervise.
Various other bits of obvious Google support for the project are also more incidental than planned. For example, Dany mentions that he built the machine in one of the on-campus workshops. Those workshops are there for "real" work, but they're also available for any employees to use on an as-available basis. Tech talks are also organized by and for the employees for their own interests, with basically zero "corporate" supervision. Most are actually job-related, but far from all. There are plenty of project talks and hobby talks (though this particular hobby/project talk is much cooler than most).
I imagine there was a cursory review required to get permission to publish the talk and the design, but such things tend to be handled on a "is there some really good reason we should say no?" basis. If not... go for it. Publishing cool, geeky things done by Google engineers is pretty positive for Google's brand, and it makes the engineers happy, which is good for employee retention -- especially since the kind of employees who do cool stuff for fun is the kind Google most wants to retain.
Bottom line: It's very unlikely anyone at Google has a corporate strategy built around the release of this information. It's just an engineer doing something he thinks is fun and valuable (to someone) and the company providing generic support for such activities, and otherwise staying out of the way.
linking to gpl code also means your code needs to be gpl. The whole point of the lgpl is to have a license without this restriction.
Even if it is their own code, if they use (link) to gpl code they can still be screwed.
Nonsense.
Linking your code to GPL code does not change your ownership of the code at all. If you link your code to GPL code and distribute the result, you are required to give recipients of your code the right to use and redistribute it per the GPL. Period. You have to give them the source, you can't stop them from giving it to others, and you have to tell them that. That's the essence of the GPL.
But if you wrote it, entirely, it's yours, and you're perfectly free to license it under whatever other terms you like, in addition to the GPL, to include it in proprietary products, to cease distributing it yourself or absolutely anything else you like. In particular, you can modify it and close the modified version.
Now, this is not actually possible for much GPL code, because much of the code in successful GPL projects isn't owned by a single company or individual. If the contributions from multiple parties have been added, then each of them owns their pieces and no one can re-license the aggregate whole without the permission of all of the owners. In the case of something like the Linux kernel, tens of thousands of people have contributed to it and the ownership is so broadly-distributed (and often not known) that it will almost certainly never be legal to distribute it under any terms other than the GPLv2. But there are pieces of it which are owned entirely by individual people or companies, and they are free to do whatever they like with their property, should they choose to place it in a different context.
Wow. I guess that this is a new record, eclipsing even the FBI's failure from a couple of years ago. Have to say, I am impressed. Leave it to the Pentagram to do things bigger and worse than anyone else on the planet.
In fairness, in terms of employees the US Air Force is 10X the size of the FBI, and bigger than any but a small handful of corporations. In terms of assets and materiel, military forces have one or two orders of magnitude more than a comparable corporation. It's pretty much guaranteed that their expensive failed efforts are going to be bigger and worse than just about anyone's.
FWIW, I agree that unions are a net negative in most cases. But the point is they've already fought this battle, and lost it, so there's no reason to expect a different outcome this time.
In a more industrial context, the USSR tried to "optimize" production of all sorts of goods by doing exactly what you say the software industry ought to be doing: standardize on one design, build one big factory.
Funny, I'd read the exact opposite - that they insisted on local production of goods even when it'd make more sense to centralize them in one big factory, and that's what did them in.
They did early on, but by the 70s and 80s they were very much into the "one enormous factory" mode, precisely because the cottage approach didn't work. But even when they did try to localize production, they still attempted extreme standardization, or at least had extreme standardization -- it may have been the result of lack of incentive for innovation rather than a party directive.
People may don't want to "move to the cloud" per se, but the cloud can be a way to archive what they want.
People want to have access to their stuff anywhere, anytime, from any device, even one they're only using for a few minutes. Not only do they not want to know what a file is or where it's saved, they don't want to have to "save" it at all. They should just be able to get it when they want it and any changes they make should instantly be available wherever else they want to use it.
This is what the cloud offers, and people really DO want it.
Nature tells us that in the overwhelming majority of cases, mutations (aka diversity) do not give an advantage.
Have you ever looked at the massive level of diversity in nature? Even what we call "species" and choose to view as cohesive aren't, really... they're just a collection of diverse individuals who have barely enough commonality to be able to breed successfully... most of the time. And the sheer number of species is staggering -- and important, because there is strength in diversity.
In a more industrial context, the USSR tried to "optimize" production of all sorts of goods by doing exactly what you say the software industry ought to be doing: standardize on one design, build one big factory. Highly centralized, highly efficient, right? Except... no. It turns out that competitive duplication of effort is more efficient. The US destroyed USSR in terms of design and manufacturing, consistently producing products that were more advanced and less costly, and it wasn't in spite of the capitalists' wasteful diversity, it was because of the non-obvious but very real efficiency that arises from diversity.
With respect to browsers, just look at what a mess IE was when Microsoft was dominant. Consider what happened to the rate of progress when Chrome entered the picture, with its orders-of-magnitude-faster Javascript engine and streamlined UI. Look at the result of the huge waste of effort that was Konqueror -- except that it produced KHTML, AKA WebKit, which became the basis for Safari and Chrome both. Heck, go all the way back to when Marc Andreesen abandoned Mosaic development to uselessly duplicate its functionality in something called "Netscape Navigator", which later became Mozilla, and then Firefox. Look at all of the features pioneered by Opera and later adopted by the rest of the browsers.
In every one of those cases, apparently duplication of effort resulted in new ideas, new features, new approaches which then cross-pollinated to improve all of the browsers.
Choice is a illusion. Get me the number of people who actively choose to use IE over Chrome and Firefox.
Not the people who use it because it was pre-installed on their computer.
Irrelevant.
The key point is that the market share is well-divided. Web site authors can't code to a specific browser, and no one browser can dictate the course of the technology. Sites have to code to standards, and browsers have to implement standards, and collaborate to define new standards.
This means that choice is real enough to accomplish what it really needs to accomplish -- require the industry to focus on standards and cooperative competition, since no one is in a position to dominate. And therefore, choice is not illusory, it's truly real, because you truly can use whichever browser you want, with a reasonable expectation that it's going to work. It doesn't matter if some people don't know enough to exercise that choice, and it doesn't even matter if IE is only "chosen" by those who don't know better.
The natural wombs are pumping out product at a terrifyingly prodigious rate with no help from you.
No they're not. In Hans Rosling's words, we've already reached and passed "Peak Child". The absolute number of annual births worldwide has peaked at a little over 100 million per year -- ~2 billion per 20-year generation -- and is beginning to gradually decline, even though some areas of the world are producing far more than their share still. As those areas get wealthier and more educated (which is happening rapidly) the birth rate will continue dropping. The youngest generation now alive will replace itself, but not more. Because there are many more young people than old people now, as each successive generation replaces itself we'll eventually reach the point where the five living generations between 0 and 100 years of age consist of about 2 billion each, for a total of 10 billion, and that will be the maximum population, barring significant changes in life expectancy.
I don't know. I asked my wife and she said she really enjoyed being pregnant and wouldn't have wanted to miss out on it for anything. Except maybe the morning sickness and the last 3-4 weeks. I expect a lot of women feel the same. Not all, certainly.
When I say "always been powers of 10", I'm talking about going back long, long before any of Windows, Linux or OS X existed. The IBM 305 RAMAC, for example, was the first ever commercially available hard drive, introduced in 1956 with a storage capacity of 5 MB -- meaning exactly 5,000,000 7-bit bytes. Ever since then, drives have had power-of-10 sizes. It's not their fault that baby OSes coming along 30-40 years later misled their users.
I'm sure the defense would be able to subpoena the relevant segments of recordings, though, and the police will have some explaining to do if they're routinely not available.
Yeah, because that's been working amazingly well so far with the dashboard cams.
It does all the time. The dashboard cams often don't get anything of use because abusive cops are smart enough to stay out of their field of view, but that's a different issue entirely, and not one that is nearly as easy for bad cops to manage when the camera is mounted on their head.
UNIVAC, IBM, Burroughs and others all had power-of-10 memory sizes until the mid-60s, and some not until the 70s. The PDP line was the first group of commercially-available machines to deliver memory capacity in powers of two sizes. IBM didn't shift to powers of 2 until the system/360. Burroughs' stack machines never really did, but they had really weird memory architectures all around.
And we can see how that worked out. Why would this be different?
Because what's best for the policeman's union may not be what's best for the general public?
And the way it worked out is that any defense attorney or plaintiff's attorney can subpoena the dashcam videos, and will get them, regardless of whether the policeman's union likes it or not. What alternative would be better for the public?
I don't think the public will or should ever be able to see all the recordings at will. Police officers are often in places and looking at things that the public does not have a valid interest in seeing -- not because of the officers, but because of the rights of the public with whom they're interacting.
I'm sure the defense would be able to subpoena the relevant segments of recordings, though, and the police will have some explaining to do if they're routinely not available.
The jury won't be told that a recording ever existed. They still won't believe the police, but the lack of a video won't be admissible.
At the very least they'll know there should be a recording, and I don't see any reason the defense attorney would be barred from asking the officer on the stand "Were you wearing your department-issued officer-cam during the incident? Have you reviewed the footage from your camera? Was the footage consistent with your testimony here today?"
Slashdot Mirror
It's nonsense for the reasons I stated. Your point about dynamic linking creating a derivative work (or not) is also potentially valid, but it's moot unless RTS is lying.
Er, I meant "erase" or "lose" rather than the second "view". Sorry.
It's not necessary to stop the police from viewing the video just to be sure they can't view it.
My wife's kids, at least.
The summary questions Google's motivations for doing this, but I think it should be clear this isn't a Google project, really. 20% projects can't be totally random, personal things that have no relationship whatsoever with the business or possible business... but the link can be very tenuous, and the cooler the project is, the weaker it can be. All tech managers at Google are engineers themselves and tend to be just as able to geek out about cool stuff as the people they supervise.
Various other bits of obvious Google support for the project are also more incidental than planned. For example, Dany mentions that he built the machine in one of the on-campus workshops. Those workshops are there for "real" work, but they're also available for any employees to use on an as-available basis. Tech talks are also organized by and for the employees for their own interests, with basically zero "corporate" supervision. Most are actually job-related, but far from all. There are plenty of project talks and hobby talks (though this particular hobby/project talk is much cooler than most).
I imagine there was a cursory review required to get permission to publish the talk and the design, but such things tend to be handled on a "is there some really good reason we should say no?" basis. If not... go for it. Publishing cool, geeky things done by Google engineers is pretty positive for Google's brand, and it makes the engineers happy, which is good for employee retention -- especially since the kind of employees who do cool stuff for fun is the kind Google most wants to retain.
Bottom line: It's very unlikely anyone at Google has a corporate strategy built around the release of this information. It's just an engineer doing something he thinks is fun and valuable (to someone) and the company providing generic support for such activities, and otherwise staying out of the way.
Yep. That's exactly what I said :-)
linking to gpl code also means your code needs to be gpl. The whole point of the lgpl is to have a license without this restriction.
Even if it is their own code, if they use (link) to gpl code they can still be screwed.
Nonsense.
Linking your code to GPL code does not change your ownership of the code at all. If you link your code to GPL code and distribute the result, you are required to give recipients of your code the right to use and redistribute it per the GPL. Period. You have to give them the source, you can't stop them from giving it to others, and you have to tell them that. That's the essence of the GPL.
But if you wrote it, entirely, it's yours, and you're perfectly free to license it under whatever other terms you like, in addition to the GPL, to include it in proprietary products, to cease distributing it yourself or absolutely anything else you like. In particular, you can modify it and close the modified version.
Now, this is not actually possible for much GPL code, because much of the code in successful GPL projects isn't owned by a single company or individual. If the contributions from multiple parties have been added, then each of them owns their pieces and no one can re-license the aggregate whole without the permission of all of the owners. In the case of something like the Linux kernel, tens of thousands of people have contributed to it and the ownership is so broadly-distributed (and often not known) that it will almost certainly never be legal to distribute it under any terms other than the GPLv2. But there are pieces of it which are owned entirely by individual people or companies, and they are free to do whatever they like with their property, should they choose to place it in a different context.
Wow. I guess that this is a new record, eclipsing even the FBI's failure from a couple of years ago. Have to say, I am impressed. Leave it to the Pentagram to do things bigger and worse than anyone else on the planet.
In fairness, in terms of employees the US Air Force is 10X the size of the FBI, and bigger than any but a small handful of corporations. In terms of assets and materiel, military forces have one or two orders of magnitude more than a comparable corporation. It's pretty much guaranteed that their expensive failed efforts are going to be bigger and worse than just about anyone's.
FWIW, I agree that unions are a net negative in most cases. But the point is they've already fought this battle, and lost it, so there's no reason to expect a different outcome this time.
In a more industrial context, the USSR tried to "optimize" production of all sorts of goods by doing exactly what you say the software industry ought to be doing: standardize on one design, build one big factory.
Funny, I'd read the exact opposite - that they insisted on local production of goods even when it'd make more sense to centralize them in one big factory, and that's what did them in.
They did early on, but by the 70s and 80s they were very much into the "one enormous factory" mode, precisely because the cottage approach didn't work. But even when they did try to localize production, they still attempted extreme standardization, or at least had extreme standardization -- it may have been the result of lack of incentive for innovation rather than a party directive.
People may don't want to "move to the cloud" per se, but the cloud can be a way to archive what they want.
People want to have access to their stuff anywhere, anytime, from any device, even one they're only using for a few minutes. Not only do they not want to know what a file is or where it's saved, they don't want to have to "save" it at all. They should just be able to get it when they want it and any changes they make should instantly be available wherever else they want to use it.
This is what the cloud offers, and people really DO want it.
Chrome is a mess
How so?
Nature tells us that in the overwhelming majority of cases, mutations (aka diversity) do not give an advantage.
Have you ever looked at the massive level of diversity in nature? Even what we call "species" and choose to view as cohesive aren't, really... they're just a collection of diverse individuals who have barely enough commonality to be able to breed successfully... most of the time. And the sheer number of species is staggering -- and important, because there is strength in diversity.
In a more industrial context, the USSR tried to "optimize" production of all sorts of goods by doing exactly what you say the software industry ought to be doing: standardize on one design, build one big factory. Highly centralized, highly efficient, right? Except... no. It turns out that competitive duplication of effort is more efficient. The US destroyed USSR in terms of design and manufacturing, consistently producing products that were more advanced and less costly, and it wasn't in spite of the capitalists' wasteful diversity, it was because of the non-obvious but very real efficiency that arises from diversity.
With respect to browsers, just look at what a mess IE was when Microsoft was dominant. Consider what happened to the rate of progress when Chrome entered the picture, with its orders-of-magnitude-faster Javascript engine and streamlined UI. Look at the result of the huge waste of effort that was Konqueror -- except that it produced KHTML, AKA WebKit, which became the basis for Safari and Chrome both. Heck, go all the way back to when Marc Andreesen abandoned Mosaic development to uselessly duplicate its functionality in something called "Netscape Navigator", which later became Mozilla, and then Firefox. Look at all of the features pioneered by Opera and later adopted by the rest of the browsers.
In every one of those cases, apparently duplication of effort resulted in new ideas, new features, new approaches which then cross-pollinated to improve all of the browsers.
Diversity isn't just good. It's essential.
Choice is a illusion. Get me the number of people who actively choose to use IE over Chrome and Firefox. Not the people who use it because it was pre-installed on their computer.
Irrelevant.
The key point is that the market share is well-divided. Web site authors can't code to a specific browser, and no one browser can dictate the course of the technology. Sites have to code to standards, and browsers have to implement standards, and collaborate to define new standards.
This means that choice is real enough to accomplish what it really needs to accomplish -- require the industry to focus on standards and cooperative competition, since no one is in a position to dominate. And therefore, choice is not illusory, it's truly real, because you truly can use whichever browser you want, with a reasonable expectation that it's going to work. It doesn't matter if some people don't know enough to exercise that choice, and it doesn't even matter if IE is only "chosen" by those who don't know better.
The natural wombs are pumping out product at a terrifyingly prodigious rate with no help from you.
No they're not. In Hans Rosling's words, we've already reached and passed "Peak Child". The absolute number of annual births worldwide has peaked at a little over 100 million per year -- ~2 billion per 20-year generation -- and is beginning to gradually decline, even though some areas of the world are producing far more than their share still. As those areas get wealthier and more educated (which is happening rapidly) the birth rate will continue dropping. The youngest generation now alive will replace itself, but not more. Because there are many more young people than old people now, as each successive generation replaces itself we'll eventually reach the point where the five living generations between 0 and 100 years of age consist of about 2 billion each, for a total of 10 billion, and that will be the maximum population, barring significant changes in life expectancy.
I don't know. I asked my wife and she said she really enjoyed being pregnant and wouldn't have wanted to miss out on it for anything. Except maybe the morning sickness and the last 3-4 weeks. I expect a lot of women feel the same. Not all, certainly.
Google engineers don't have descendants.
Ssshhh. My "kids" don't know!
When I say "always been powers of 10", I'm talking about going back long, long before any of Windows, Linux or OS X existed. The IBM 305 RAMAC, for example, was the first ever commercially available hard drive, introduced in 1956 with a storage capacity of 5 MB -- meaning exactly 5,000,000 7-bit bytes. Ever since then, drives have had power-of-10 sizes. It's not their fault that baby OSes coming along 30-40 years later misled their users.
I'm sure the defense would be able to subpoena the relevant segments of recordings, though, and the police will have some explaining to do if they're routinely not available.
Yeah, because that's been working amazingly well so far with the dashboard cams.
It does all the time. The dashboard cams often don't get anything of use because abusive cops are smart enough to stay out of their field of view, but that's a different issue entirely, and not one that is nearly as easy for bad cops to manage when the camera is mounted on their head.
No. RAM was pretty much always in powers of 2.
UNIVAC, IBM, Burroughs and others all had power-of-10 memory sizes until the mid-60s, and some not until the 70s. The PDP line was the first group of commercially-available machines to deliver memory capacity in powers of two sizes. IBM didn't shift to powers of 2 until the system/360. Burroughs' stack machines never really did, but they had really weird memory architectures all around.
And we can see how that worked out. Why would this be different?
Because what's best for the policeman's union may not be what's best for the general public?
And the way it worked out is that any defense attorney or plaintiff's attorney can subpoena the dashcam videos, and will get them, regardless of whether the policeman's union likes it or not. What alternative would be better for the public?
everything else is still powers of 10.
The files stored on said HDD aren't.
That's up to your OS.
And I didn't make those scenarios up. They are positions taken by police unions when dashcams were being fitted to cars.
And we can see how that worked out. Why would this be different?
I don't think the public will or should ever be able to see all the recordings at will. Police officers are often in places and looking at things that the public does not have a valid interest in seeing -- not because of the officers, but because of the rights of the public with whom they're interacting.
I'm sure the defense would be able to subpoena the relevant segments of recordings, though, and the police will have some explaining to do if they're routinely not available.
The jury won't be told that a recording ever existed. They still won't believe the police, but the lack of a video won't be admissible.
At the very least they'll know there should be a recording, and I don't see any reason the defense attorney would be barred from asking the officer on the stand "Were you wearing your department-issued officer-cam during the incident? Have you reviewed the footage from your camera? Was the footage consistent with your testimony here today?"