"It's just a hard thing to unlearn, our brains literaly loose the language plasticity they have at birth."
And that's why citizens of countries that dub films rather than subtitle them get a serious disadvantage when learning languages. And why dubbing even cartoons for kids is a bad idea.
Of course, this entire problem is easily solved. Simply perform a serious devaluation of the dollar so it becomes cheaper with US labour than external labour.
If heat is a significant issue for you, you might consider underclocking. It's an extrordinarily efficient way to reduce heat. Reduce bus speed and core voltage and you can chop heat output to a third for no more than 25% performance loss. Cutting an Athlon XP from 1400 MHz down to 1000MHz with core down to 1.15V and you should be able run it at 21W with only a fairly large heatsink.
From what I've seen it's quite a bit more expensive to roll my own decent solution.
The advantage from my own linux based solution is I have a shitload of features ranging from being able to back it up, remote in via ssh to adding any number of capture cards and disks I wish into it.
The selling point of building it yourself is flexibility.
The disadvantages are price and, of course, having to know a fair amount about computers to get it working.
Well, more likely, the other side would send their eightyearolds with a modem and hack the battlefield units.
Of course, the entire concept may be overkill as it would make much more sense to hack the US election and make Fido the Dog president with a whopping 120% of the votes.
The combination of "computer security by sales pamphlet" combined with "computerized weapons and voting systems" makes for such a fun future.
And these changes are rather pointless. Controlling the usage of a name should be done with trademarks, which is by far the most efficient method of doing that, and may be very difficult to do otherwise, even with a clause in the license.
And the acknowledgement is extrordinary pointless as you're not allowed to remove copyright notices _anyway_.
Because it possibly makes it incompatible with other licenses, for _no reason at all_. They state:
'The purpose of these changes is to strengthen the "except claim you wrote it" clause'
Brilliant. Except, of course, copyright law already _does_ this. Slap a copyright notice on it and it would be illegal to remove it. If they want a notice in the documentation, just slap another copyright notice in there. The clauses are completely pointless and handle a non-problem in a possibly damaging way. And the use of the name should be handled by trademark, which is the appropriate way to handle what you can and cannot use a name for.
Why not use MythTV? As far as I can tell ForceWare doesnt have any features myth doesnt have... (and MythTV has a few ForceWare doesnt appear to have).
"And attacking Linux on Mainframes is like hitting the broad side of a barn"
It's fairly tragic really. For these studies they've probably commissioned another dozen or so that have turned up the cost advantages of Linux, or they would have had some more relevant studies there.
What's next? An Earth based Windows Server 2003 is cheaper to buy than an x86 Linux based Mars lander?
What studies arent there really say more than the ones that are.
I think you've hit exactly on how it is usually solved. It's a good solution for large and/or industrial buildings where you have a real ventilation system and control the airflow through intakes and exhausts.
In ordinary houses or smaller appartment buildings you rarely have such a system, or you only have half of it, with an exhaust only and relying on intakes around windows or in vents in the walls.
So, yes, it's good, and fairly efficient. It's just quite expensive.:)
"Imagine if your house were perfectly insulated, then you would only need to suck out the heat added by the things inside it (200W per person, another 200W per computer)."
Well, and the CO2. And the water vapour. And whatever toxics that leak in miniscule amounts from materials inside.
You dont have to imagine it, it's been tried. It was found to profoundly suck, as people got sick and the houses molded or rotted.
The technology for building houses with perfect insulation has been here for a long time. Unfortunately, the problem isnt the insulation anymore, the problem is the ventilation. But come up with a highly efficient and cheap heat exchanger system and you could solve that too:).
Insulation isnt really the problem anyway. It's easy to make a house you could heat with a candle. Modern houses in countries with cold winters have triple glazing and good insulation, with negligable heat loss through windows and walls.
The problem is ventilation. Even apart from the issue that you'd suffocate, houses that are too insulated are almost guaranteed get mold problems. You need a constant airflow, and that's where you get the major heat loss. Of course, various techniques like heat exchangers exist to ameliorate this, but unfortunately the technology for 100% efficiency is not quite there yet.
Of course, the history of the lightbulb is another great example why patents stink. Edison was, in fact, not the inventor. The working lightbulb idea came along in 1809, and various models were developed over the next 65 years. The patent was filed in 1875 by Woodward and Evans, but they couldnt finance the product development. That ended up with Edison (read, the 'more established and well financed company') buying the patent, finalizing product development and reaping the benefits.
The original inventors didnt benefit, the lightbulb would have happened with or without the patent as the surrounding technology had reached the point of making it practical.
Sure. I just called my ISP the other day and told them I needed 10 new static IP's for my home network and they gave them to me at once and without charge.
Oh, wait, no, that was in a dream.
I can get as many static IP adresses as I can conceivably ever want by transitioning to IPv6 _today_ without even calling my ISP. That's compelling reason enough to switch for me.
The GPL depends strongly on the concept of COPYRIGHT for forcing the contributor to grant rights in exchange for being allowed to distribute derivative works.
The copyright of the original author is the legal basis, not any public domain tradition. And it's the same in the nordic countries as in most of the world, if you dont have permission from the copyright holder you cant distribute, and with the GPL you can get that permission by adhering to the terms of the GPL or by getting your own separate license from the copyright holder.
There is no legal problem. The GPL is as watertight as it's possible to get, and the only thing that could cause a problem for the GPL would be if some global agreement to remove the concept of copyrights came along. Which isnt particularly likely to happen anytime soon.
Yes, most connected devices are behind firewalls or proxys, and eventually we'll run out of IP-adresses for the firewalls and proxys. NAT decreases the need for adresses but it doesnt remove the need entirely. Servers, firewalls, every customer to an ISP, possibly roaming devices like cellphones, anything that needs real p2p connections, etc, will need real adresses. Today it's very difficult to get real IP adresses even when you have legitimate need for them from many providers. Because there is a shortage.
NAT is not a replacement for firewalling. Just because you have public adresses doesnt mean you let anything through to them, so the toaster isnt really affected anyway.
"C has a stable ABI. For a program that is supposed to work anywhere without recompilation this is a big deal."
This has been the single most deciding factor behind why I dont do C++ if I can avoid it. Every compiler generating incompatible libs and binaries, the complete hell with every library upgrade compiled with a newer compiler when nothing works, the absolute pain when trying to link vendor libs to C++ code, etc.
Of course, there are many other reasons, but that one is IMO, the one most annoying part of C++.
The GPL is never "viral". The restriction is that you cannot distribute GPL code together with code that isnt as free, or freer than the GPL code. If you want to use a BSD (w/o advertizing clause) license on your code that's perfectly fine. Anyone could take your BSD code and release it in a proprietary product, provided they then cease linking to the GPL library (wether statically or dynamically).
The combined work of a BSD application using a GPL library will be distributed under the _terms_ of the GPL, but remove the GPL portions and you have a pure BSD licensed codebase again. The GPL only governs the license of the GPL software, and wether or not you can distribute the GPL software. It never governs how you can license your own software (except so far as to disallow usage of the GPL code itself, unless you license in a GPL compatible way (BSD, public domain, GPL, whatever)).
Well, the manufacturer states 3 percent false negatives and 1 in a million false positives. But that's when it's used for access control, with presumably a small database of employees or whatever. But then you'd have the system to reject (it's not bad to have an employee have to try again, it's bad to let intruders in). That works well.
However, when used in the way they're trying to use it now you dont have the ability to simplify that way anymore. You neither want to get false positives or false negatives, so you cant tune it one way or the other. And if you cant do that you get stuck with the intersection between positives and negatives, and with the way those graphs usually look you'd get a few fractions of percent error rates.
Now, if you compare 24 million people against a register of, say, 50000 you get a pretty horrendous failure rate. Quite likely in the range of several hundred thousands mistakes per year.
Biometrics are useful for some problems, but this isnt one of them. In this case it will just be a huge problem with the number of innocents getting hassled by misidentification outnumbering the real catches by the tens of thousands.
So, were the identities verified, or were these just false positives? Were the identities assumed, or was the system assuming the wrong identity?
You dont need to nail _the_ guy to 'prove' the efficiency of a system to the public, you need to nail _a_ guy. Wether it's the right or wrong people doesnt really matter.
AFIS systems have an error rate around a few fractions of a percent, with more or less false positives or negatives depending on settings. Several hundreds of thousands false matches should be expected every year from this system.
Of course it wasnt pointless. Now they have your print, which means that if anyone with a rather similar print goes on a murder spree they can easily nail you for it.
Nailing you is just as satisfying for the public as nailing the real culprit, and it serves the political agenda of the prosecution just as well.
This way they'll have such a large database that they can always dig up some poor sucker who matches well enough when they need a sacrificial goat. It's quite useless for any real security, but dont confuse that with being useless for everything.
And on the flip side of that, that doesnt really matter for Blockbuster, as everyone who actually wants to rent the movie in a country 6 months after the first release will have obtained it elsewhere, wether from imports or over the net.
In which case you can remove that code and do what you should have been doing in the first place if you didnt want to release the code; write it yourself.
This is no more viral than any other copyrighted code. If you place a small amount of Microsoft code, or a small amount of Apple code, or a small amount of IBM code in your project you can be forced to stop distributing the whole thing due to that 'small infraction'. If you wish to claim something is viral here, it's copyright law. If you violate copyright you can be forced to cease doing so. If you want to call that viral, feel free. I wouldnt consider it a very good term to use tho.
Slashdot Mirror
"It's just a hard thing to unlearn, our brains literaly loose the language plasticity they have at birth."
And that's why citizens of countries that dub films rather than subtitle them get a serious disadvantage when learning languages. And why dubbing even cartoons for kids is a bad idea.
Of course, this entire problem is easily solved. Simply perform a serious devaluation of the dollar so it becomes cheaper with US labour than external labour.
If heat is a significant issue for you, you might consider underclocking. It's an extrordinarily efficient way to reduce heat. Reduce bus speed and core voltage and you can chop heat output to a third for no more than 25% performance loss. Cutting an Athlon XP from 1400 MHz down to 1000MHz with core down to 1.15V and you should be able run it at 21W with only a fairly large heatsink.
From what I've seen it's quite a bit more expensive to roll my own decent solution.
The advantage from my own linux based solution is I have a shitload of features ranging from being able to back it up, remote in via ssh to adding any number of capture cards and disks I wish into it.
The selling point of building it yourself is flexibility.
The disadvantages are price and, of course, having to know a fair amount about computers to get it working.
Well, more likely, the other side would send their eightyearolds with a modem and hack the battlefield units.
Of course, the entire concept may be overkill as it would make much more sense to hack the US election and make Fido the Dog president with a whopping 120% of the votes.
The combination of "computer security by sales pamphlet" combined with "computerized weapons and voting systems" makes for such a fun future.
And these changes are rather pointless. Controlling the usage of a name should be done with trademarks, which is by far the most efficient method of doing that, and may be very difficult to do otherwise, even with a clause in the license.
And the acknowledgement is extrordinary pointless as you're not allowed to remove copyright notices _anyway_.
Because it possibly makes it incompatible with other licenses, for _no reason at all_. They state:
'The purpose of these changes is to strengthen the "except claim you wrote it" clause'
Brilliant. Except, of course, copyright law already _does_ this. Slap a copyright notice on it and it would be illegal to remove it. If they want a notice in the documentation, just slap another copyright notice in there. The clauses are completely pointless and handle a non-problem in a possibly damaging way. And the use of the name should be handled by trademark, which is the appropriate way to handle what you can and cannot use a name for.
Why not use MythTV? As far as I can tell ForceWare doesnt have any features myth doesnt have... (and MythTV has a few ForceWare doesnt appear to have).
"And attacking Linux on Mainframes is like hitting the broad side of a barn"
It's fairly tragic really. For these studies they've probably commissioned another dozen or so that have turned up the cost advantages of Linux, or they would have had some more relevant studies there.
What's next? An Earth based Windows Server 2003 is cheaper to buy than an x86 Linux based Mars lander?
What studies arent there really say more than the ones that are.
I think you've hit exactly on how it is usually solved. It's a good solution for large and/or industrial buildings where you have a real ventilation system and control the airflow through intakes and exhausts.
:)
In ordinary houses or smaller appartment buildings you rarely have such a system, or you only have half of it, with an exhaust only and relying on intakes around windows or in vents in the walls.
So, yes, it's good, and fairly efficient. It's just quite expensive.
"Imagine if your house were perfectly insulated, then you would only need to suck out the heat added by the things inside it (200W per person, another 200W per computer)."
:).
Well, and the CO2. And the water vapour. And whatever toxics that leak in miniscule amounts from materials inside.
You dont have to imagine it, it's been tried. It was found to profoundly suck, as people got sick and the houses molded or rotted.
The technology for building houses with perfect insulation has been here for a long time. Unfortunately, the problem isnt the insulation anymore, the problem is the ventilation. But come up with a highly efficient and cheap heat exchanger system and you could solve that too
Insulation isnt really the problem anyway. It's easy to make a house you could heat with a candle. Modern houses in countries with cold winters have triple glazing and good insulation, with negligable heat loss through windows and walls.
The problem is ventilation. Even apart from the issue that you'd suffocate, houses that are too insulated are almost guaranteed get mold problems. You need a constant airflow, and that's where you get the major heat loss. Of course, various techniques like heat exchangers exist to ameliorate this, but unfortunately the technology for 100% efficiency is not quite there yet.
Of course, the history of the lightbulb is another great example why patents stink. Edison was, in fact, not the inventor. The working lightbulb idea came along in 1809, and various models were developed over the next 65 years. The patent was filed in 1875 by Woodward and Evans, but they couldnt finance the product development. That ended up with Edison (read, the 'more established and well financed company') buying the patent, finalizing product development and reaping the benefits.
The original inventors didnt benefit, the lightbulb would have happened with or without the patent as the surrounding technology had reached the point of making it practical.
Sure. I just called my ISP the other day and told them I needed 10 new static IP's for my home network and they gave them to me at once and without charge.
Oh, wait, no, that was in a dream.
I can get as many static IP adresses as I can conceivably ever want by transitioning to IPv6 _today_ without even calling my ISP. That's compelling reason enough to switch for me.
You can tunnel onto the ipv6 backbone and get your own ipv6 adresses without your ISP supporting it at all.
The GPL depends strongly on the concept of COPYRIGHT for forcing the contributor to grant rights in exchange for being allowed to distribute derivative works.
The copyright of the original author is the legal basis, not any public domain tradition. And it's the same in the nordic countries as in most of the world, if you dont have permission from the copyright holder you cant distribute, and with the GPL you can get that permission by adhering to the terms of the GPL or by getting your own separate license from the copyright holder.
There is no legal problem. The GPL is as watertight as it's possible to get, and the only thing that could cause a problem for the GPL would be if some global agreement to remove the concept of copyrights came along. Which isnt particularly likely to happen anytime soon.
Yes, most connected devices are behind firewalls or proxys, and eventually we'll run out of IP-adresses for the firewalls and proxys. NAT decreases the need for adresses but it doesnt remove the need entirely. Servers, firewalls, every customer to an ISP, possibly roaming devices like cellphones, anything that needs real p2p connections, etc, will need real adresses. Today it's very difficult to get real IP adresses even when you have legitimate need for them from many providers. Because there is a shortage.
NAT is not a replacement for firewalling. Just because you have public adresses doesnt mean you let anything through to them, so the toaster isnt really affected anyway.
"C has a stable ABI. For a program that is supposed to work anywhere without recompilation this is a big deal."
This has been the single most deciding factor behind why I dont do C++ if I can avoid it. Every compiler generating incompatible libs and binaries, the complete hell with every library upgrade compiled with a newer compiler when nothing works, the absolute pain when trying to link vendor libs to C++ code, etc.
Of course, there are many other reasons, but that one is IMO, the one most annoying part of C++.
The GPL is never "viral". The restriction is that you cannot distribute GPL code together with code that isnt as free, or freer than the GPL code. If you want to use a BSD (w/o advertizing clause) license on your code that's perfectly fine. Anyone could take your BSD code and release it in a proprietary product, provided they then cease linking to the GPL library (wether statically or dynamically).
The combined work of a BSD application using a GPL library will be distributed under the _terms_ of the GPL, but remove the GPL portions and you have a pure BSD licensed codebase again. The GPL only governs the license of the GPL software, and wether or not you can distribute the GPL software. It never governs how you can license your own software (except so far as to disallow usage of the GPL code itself, unless you license in a GPL compatible way (BSD, public domain, GPL, whatever)).
Well, the manufacturer states 3 percent false negatives and 1 in a million false positives. But that's when it's used for access control, with presumably a small database of employees or whatever. But then you'd have the system to reject (it's not bad to have an employee have to try again, it's bad to let intruders in). That works well.
However, when used in the way they're trying to use it now you dont have the ability to simplify that way anymore. You neither want to get false positives or false negatives, so you cant tune it one way or the other. And if you cant do that you get stuck with the intersection between positives and negatives, and with the way those graphs usually look you'd get a few fractions of percent error rates.
Now, if you compare 24 million people against a register of, say, 50000 you get a pretty horrendous failure rate. Quite likely in the range of several hundred thousands mistakes per year.
Biometrics are useful for some problems, but this isnt one of them. In this case it will just be a huge problem with the number of innocents getting hassled by misidentification outnumbering the real catches by the tens of thousands.
So, were the identities verified, or were these just false positives? Were the identities assumed, or was the system assuming the wrong identity?
You dont need to nail _the_ guy to 'prove' the efficiency of a system to the public, you need to nail _a_ guy. Wether it's the right or wrong people doesnt really matter.
AFIS systems have an error rate around a few fractions of a percent, with more or less false positives or negatives depending on settings. Several hundreds of thousands false matches should be expected every year from this system.
Of course it wasnt pointless. Now they have your print, which means that if anyone with a rather similar print goes on a murder spree they can easily nail you for it.
Nailing you is just as satisfying for the public as nailing the real culprit, and it serves the political agenda of the prosecution just as well.
This way they'll have such a large database that they can always dig up some poor sucker who matches well enough when they need a sacrificial goat. It's quite useless for any real security, but dont confuse that with being useless for everything.
"Hell its only a matter of time before people making mp3's are declared copyright terrorists and sentenced to lengthy jail terms in cuba :)"
Sentenced and jail terms? You must mean 'declared copyright terrorists, shipped off to Cuba and held indefinitely without a trial'.
And on the flip side of that, that doesnt really matter for Blockbuster, as everyone who actually wants to rent the movie in a country 6 months after the first release will have obtained it elsewhere, wether from imports or over the net.
In which case you can remove that code and do what you should have been doing in the first place if you didnt want to release the code; write it yourself.
This is no more viral than any other copyrighted code. If you place a small amount of Microsoft code, or a small amount of Apple code, or a small amount of IBM code in your project you can be forced to stop distributing the whole thing due to that 'small infraction'. If you wish to claim something is viral here, it's copyright law. If you violate copyright you can be forced to cease doing so. If you want to call that viral, feel free. I wouldnt consider it a very good term to use tho.