A general rule of thumb for errors in counting in a sample is sqrt(n).
Err...If I remember correctly from my stats class the general rule for margin of error is actually 1/sqrt(n). Common sense says that as the sample size increases, the margin of error should decrease, so sqrt(n) doesn't seem right.
(IIRC, this is because we are sampling from binomial distribution (either an IP exists or it doesn't), where the margin of error in the normal approximation is given by z_star*sqrt(p_hat*(1-p_hat))/sqrt(n). Using z_star~2 for 95% confidence and p_hat=.5 in the worst case, this reduces to 1/sqrt(n)).
Anyway, a sample size of 150,000 is incredibly good, and I think margin of error will be so small that it's not worth calculating (yes I'm lazy). So a better statistical question is whether the IP addresses tested were a random sample of all possible IP addresses? (For example, I know that some addresses are reserved and may not be used, so it would be a mistake to sample such addresses.)
He may have some interesting things to say, but it looks like he didn't make many friends on the livid-dev list. Search for his name in the list archives (for example see here , here , here , or here ).
Many people seem to believe that the creation of another program named DeCSS will help the cause of playing encrypted DVD's by confusing the MPAA's lawyers, who want to crack down on the "real" DeCSS. This might initially look like a harmless, and somewhat humorous, joke, but if it becomes popular, I think it could end up harming the efforts of those who want to watch DVDs.
One of the main (and IMHO, the strongest) arguments of the defendants in the DeCSS cases is that people have a right to watch DVD's that they own, without any hindrance from big companies. Tactics like making a "fake" DeCSS destroy the crediblity of the defendants--if people really should be able to freely distsribute DeCSS, it shouldn't be necessary to create another program to obscure the real one. In other words, why should we have to cover up activities (such as distributing DeCSS) that should be completely legal?
Furthermore, a false DeCSS program, while annoying lawyers, will be even more annoying to those who actually want to use the program--consumers. What if we win the DVD case, and it becomes legitimate to use DeCSS; I don't want to wade through numerous copies of a program porporting to be "DeCSS" before finding the real thing!
In my view, people could better utilize their time by supporting and distributing the true DeCSS, rather than confusing the issue and potentially harming the effort to create a free DVD player.
Liebowitz initially oversimplifies the problem by selecting a worst-case scenario (for MS) where Microsoft is broken up into three companies, each producing Windows. There are various other reasonable possibilites for breaking up Microsoft.
But what totally invalidates his argument is the the assumption that the different versions of Windows would immediately become so different that they would require vastly greater development resources by software companies. He concludes, through an unscientific survey, that the breakup would cost consumers $30 billion.
Such a scenario is totally unrealisic; it is clear that market forces would prevent severe fragmentation. If one company tried to create an "incompatible" version of Windows, there would be few applications for it, and no one would buy it. What company would write software for a platform that caused it to lose money? Competition would prevent the segmentation that Liebowitz predicts. For example, why has Linux, which is distbributed by a variety of companies, not broken into incompatible factions? Liebowitz apparently provides no evidence to suggest that Windows would become segmented into incompatible versions.
Furthermore, he ignores the positive effects of the breakup. The benefits of competition would certainly drive down the price of Windows, and thus actually save money for consumers.
Considering the fact that Liebowitz's study was funded by "pro-Microsoft" organizations, a few small errors wouldn't be suprising. But his basic posulations are flawed, leading to a confusion of the facts. Coming from a MS supporter this "study" therefore appears to be nothing but (to use an over-used term) FUD.
is avalable at www.linux.org.uk. It talks about the impartiality of the tests, and how companies are starting to go out of their way to criticize Linux.
What I've found is that, by and large, Humanity is Lazy. Less than 10% of us create new content. The other 90% copy it, consume it, or ignore it, but don't contribute to it. So, the heck with them! Since our primary rate of exchange is money, then make the sheep pay for what they yearn for. If your only goal is to make money, then yes, you will probably want to force people to pay for your work. Not everyone shares your view, however; others value respect and freedom, which are also valid "rates of exchange". As a creator of new content, you may have the power to restrict consumers' access to your work; that does not imply that such restrictions are right.
I'm willing to bet that almost everyone who has anything, doesn't want to share it; and those that have nothing, want everyone who has stuff to be required to share it. In one sweeping generalization, you have dismissed not only the Free Software/Open Source phenomenon, but the entire basis of the Internet as well. While your statement may be often true for material goods, it completely breaks down when describing information, which may be freely shared without loss (other than perhaps money) to its creator. Yes, MP3 and Linux may utimately be a small part of history, but I would argue that the reign of the RIAA and Micrsoft will be even more ephemeral.
If you want to hoard your talents and restrict your work, and die with more money, to be buried in an unmarked grave next to Joan Baez, well, go ahead.
>Nobody owns a trademark on the term "Open Source".
From the SPI Trademarks Page: Open Source... [is a] Registered Certification Mark[s] of Software in the Public Interest, Inc....The Open Source trademark is managed by Eric Raymond on behalf of the free software community.
I think that a "certification mark" is still protected from improper use. Unfortunately, the status of the trademark/certification mark appears to be uncertain because of a dispute between SPI and OSI, which publishes the open source definition. (You cited "prior use", which only applies to patents. Multiple trademarks may be granted for one term, if they are in unreleated fields.)
There was an editorial on/. a while back that suggested that Linux could adopt the GNU project for use with its kernel. For example, it is my understanding that using the term GNU/Linux automatically gives credit to the XFree86 Project, because the XFree86 project has adopted GNU as its operating system. Thus, rather than the GNU project adopting Linux (which I doubt either RMS or Linus would agree to), Linus and the other kernel developers could decree that GNU is the official operating system for Linux, and call the whole thing Linux. Wouldn't this keep the Linux people happy, while giving GNU the credit it deserves?
In my opinion, this would be much more reasonable than the highly impractical boycott of GNU that some have proposed.
Writing editorials about it or bashing Stallman about it only feeds the fire.
There is a difference between writing a carefully thought-out editorial and simply attempting to start a flamewar. Granted, editorials can often lead to heated debates, and perhaps the linux-kernel mailing list is not the best place for such a discussion. However, you seem to be suggesting that people withhold their opinions in the interest of avoiding flamewars; to me, the cure seems worse than the disease.
I think that the editorial is an example of how one should express an opinion, rather than bashing one's opponents or repeating what has already been said.
The license must not restrict anyone from making use of the program in a specific field of endeavor.
This would probably prevent the proposed COSS system from being Open Source(TM) compliant; similarly, it does not appear to be free software under the DFSG.
I believe that Johnson is misinterpreting the "many eyeball" effect. In my opinion, that phenomenon is only a side effect of the Free Software culture; the most important thing is that anyone may modify or improve the code however he/she chooses, without restrictions. Attempting to artificially restrict that system, for whatever purpose, would undermine Free Software's benefits of peer review on which COSS attempts to capitalize.
For example, if a large company were forced to pay thousands of dollars for a COSS program, it is likely that the company would choose a propietary product instead, thus robbing the community of any improvements that may have otherwise been supplied.
As a practical problem, who would enforce the proposed rules of COSS? In the end, probably only the most honest users would pay their fees for using a COSS program, while others wouldn't carry their fair share of the financial load. Arguably, that exactly the system in place today with Free Software/Open Source.
Slashdot Mirror
Try here.
Err...If I remember correctly from my stats class the general rule for margin of error is actually 1/sqrt(n). Common sense says that as the sample size increases, the margin of error should decrease, so sqrt(n) doesn't seem right.
(IIRC, this is because we are sampling from binomial distribution (either an IP exists or it doesn't), where the margin of error in the normal approximation is given by z_star*sqrt(p_hat*(1-p_hat))/sqrt(n). Using z_star~2 for 95% confidence and p_hat=.5 in the worst case, this reduces to 1/sqrt(n)).
Anyway, a sample size of 150,000 is incredibly good, and I think margin of error will be so small that it's not worth calculating (yes I'm lazy). So a better statistical question is whether the IP addresses tested were a random sample of all possible IP addresses? (For example, I know that some addresses are reserved and may not be used, so it would be a mistake to sample such addresses.)
Comments/corrections are appreciated.
Wil
He may have some interesting things to say, but it looks like he didn't make many
friends on the livid-dev list. Search for his name in the list archives (for example see
here , here , here , or here ).
Many people seem to believe that the creation of another program named DeCSS will help the cause of playing encrypted DVD's by confusing the MPAA's lawyers, who want to crack down on the "real" DeCSS. This might initially look like a harmless, and somewhat humorous, joke, but if it becomes popular, I think it could end up harming the efforts of those who want to watch DVDs.
One of the main (and IMHO, the strongest) arguments of the defendants in the DeCSS cases is that people have a right to watch DVD's that they own, without any hindrance from big companies. Tactics like making a "fake" DeCSS destroy the crediblity of the defendants--if people really should be able to freely distsribute DeCSS, it shouldn't be necessary to create another program to obscure the real one. In other words, why should we have to cover up activities (such as distributing DeCSS) that should be completely legal?
Furthermore, a false DeCSS program, while annoying lawyers, will be even more annoying to those who actually want to use the program--consumers. What if we win the DVD case, and it becomes legitimate to use DeCSS; I don't want to wade through numerous copies of a program porporting to be "DeCSS" before finding the real thing!
In my view, people could better utilize their time by supporting and distributing the true DeCSS, rather than confusing the issue and potentially harming the effort to create a free DVD player.
Liebowitz initially oversimplifies the problem by selecting a worst-case scenario (for MS) where Microsoft is broken up into three companies, each producing Windows. There are various other reasonable possibilites for breaking up Microsoft.
But what totally invalidates his argument is the the assumption that the different versions of Windows would immediately become so different that they would require vastly greater development resources by software companies. He concludes, through an unscientific survey, that the breakup would cost consumers $30 billion.
Such a scenario is totally unrealisic; it is clear that market forces would prevent severe fragmentation. If one company tried to create an "incompatible" version of Windows, there would be few applications for it, and no one would buy it. What company would write software for a platform that caused it to lose money? Competition would prevent the segmentation that Liebowitz predicts. For example, why has Linux, which is distbributed by a variety of companies, not broken into incompatible factions?
Liebowitz apparently provides no evidence to suggest that Windows would become segmented into incompatible versions.
Furthermore, he ignores the positive effects of the breakup. The benefits of competition would certainly drive down the price of Windows, and thus actually save money for consumers.
Considering the fact that Liebowitz's study was funded by "pro-Microsoft" organizations, a few small errors wouldn't be suprising. But his basic posulations are flawed, leading to a confusion of the facts. Coming from a MS supporter this "study" therefore appears to be nothing but (to use an over-used term) FUD.
is avalable at www.linux.org.uk. It talks about the impartiality of the tests, and how companies are starting to go out of their way to criticize Linux.
What I've found is that, by and large, Humanity is Lazy. Less than 10% of us create new content. The other 90% copy it, consume it, or ignore it, but don't contribute to it. So, the heck with them! Since our primary rate of exchange is money, then make the sheep pay for what they yearn for.
If your only goal is to make money, then yes, you will probably want to force people to pay for your work. Not everyone shares your view, however; others value respect and freedom, which are also valid "rates of exchange". As a creator of new content, you may have the power to restrict consumers' access to your work; that does not imply that such restrictions are right.
I'm willing to bet that almost everyone who has anything, doesn't want to share it; and those that have nothing, want everyone who has stuff to be required to share it.
In one sweeping generalization, you have dismissed not only the Free Software/Open Source phenomenon, but the entire basis of the Internet as well. While your statement may be often true for material goods, it completely breaks down when describing information, which may be freely shared without loss (other than perhaps money) to its creator. Yes, MP3 and Linux may utimately be a small part of history, but I would argue that the reign of the RIAA and Micrsoft will be even more ephemeral.
If you want to hoard your talents and restrict your work, and die with more money, to be buried in an unmarked grave next to Joan Baez, well, go ahead.
>Nobody owns a trademark on the term "Open Source".
... [is a] Registered Certification Mark[s] of Software in the Public Interest, Inc....The Open Source trademark is managed by Eric Raymond on behalf of the free software community.
From the SPI Trademarks Page:
Open Source
I think that a "certification mark" is still protected from improper use. Unfortunately, the status of the trademark/certification mark appears to be uncertain because of a dispute between SPI and OSI, which publishes the open source definition. (You cited "prior use", which only applies to patents. Multiple trademarks may be granted for one term, if they are in unreleated fields.)
There was an editorial on /. a while back that suggested that Linux could adopt the GNU project for use with its kernel. For example, it is my understanding that using the term GNU/Linux automatically gives credit to the XFree86 Project, because the XFree86 project has adopted GNU as its operating system. Thus, rather than the GNU project adopting Linux (which I doubt either RMS or Linus would agree to), Linus and the other kernel developers could decree that GNU is the official operating system for Linux, and call the whole thing Linux. Wouldn't this keep the Linux people happy, while giving GNU the credit it deserves?
In my opinion, this would be much more reasonable than the highly impractical boycott of GNU that some have proposed.
There is a difference between writing a carefully thought-out editorial and simply attempting to start a flamewar. Granted, editorials can often lead to heated debates, and perhaps the linux-kernel mailing list is not the best place for such a discussion. However, you seem to be suggesting that people withhold their opinions in the interest of avoiding flamewars; to me, the cure seems worse than the disease.
I think that the editorial is an example of how one should express an opinion, rather than bashing one's opponents or repeating what has already been said.
This would probably prevent the proposed COSS system from being Open Source(TM) compliant; similarly, it does not appear to be free software under the DFSG.
I believe that Johnson is misinterpreting the "many eyeball" effect. In my opinion, that phenomenon is only a side effect of the Free Software culture; the most important thing is that anyone may modify or improve the code however he/she chooses, without restrictions. Attempting to artificially restrict that system, for whatever purpose, would undermine Free Software's benefits of peer review on which COSS attempts to capitalize.
For example, if a large company were forced to pay thousands of dollars for a COSS program, it is likely that the company would choose a propietary product instead, thus robbing the community of any improvements that may have otherwise been supplied.
As a practical problem, who would enforce the proposed rules of COSS? In the end, probably only the most honest users would pay their fees for using a COSS program, while others wouldn't carry their fair share of the financial load. Arguably, that exactly the system in place today with Free Software/Open Source.