If I understand the limitation of liability clause,
SCO refuses to indemnify licensees if they should prove unable to use the product or otherwise incur damages due to IP issues.
Not only is there the little problem of the GPL,
but we can't be sure that there isn't a little infringment lurking in SCO's past, can we? If indemnification is so important for open source vendors, how come it isn't for SCO? What's sauce for the goose is sauce for the gander.
If I understand the technique correctly, it relies on information specific to individual users. Unless there is a way for users to export their information, that means that the filtering can only be done after the email reaches its destination, not by the ISP or central mail server. So it may be helfpul to individual users, but unlike some proposed techniques, it won't cut down on total email traffic.
The scientists signing the letter do not
represent the Union of Concerned Scientists.
They are an independent group who are merely
endorsing the UCS report. Furthermore, they include
scientists who are not particularly left-wing,
such as H-bomb designer Richard Garwin and physicist
Norman Ramsey, both of whom served as advisers to
Republican administrations.
According to this news item, organizations opposing the Bush administration policy include: the National Academies of Science, the American Association for the Advancement of Science, the Federation of American Scientists, and the Association of American Medical Colleges. The opposition isn't coming from the left fringe; it is mainstream.
There's a good reason that the EFF isn't mounting a
campaign against the RIAA: some of the RIAA's tactics may be scummy, and the penalties provided by the law excessive, but they're basically in the right. People who post and download copyrighted songs that they haven't already purchased in another format without the copyright holder's permission are in fact infringing valid copyrights. The purpose of the EFF is to defend civil liberties on-line, not the "right" to rip-off and distribute other people's property.
There are at least two ways the airport can pay for this. One is to include it in the fees charged to the airlines: landing fees, rental of hangar and
counter space, various other services. Another is to include it in the rent paid by shops and restaurants.
The real story here is that companies and other organizations and institutions are setting machines up as servers and are too stupid to create an appropriate robots.txt file and/or keep their confidential information elsewhere. Google doesn't just drop in, even on networked machines. I have some sympathy for individuals who don't understand what they are doing when they make their machine a server, but surely any professional sysadmin, even one with limited training and experience, should know better than this. It's the same as leaving your briefcase on the front seat of an unlocked car.
What happened here is certainly appalling, but I'm not so sure that outsourcing is the main problem.
Outsourcing arguably increases the risk of problems
of this sort because an in-house programmer is more likely to know the rules of the game, but this seems to me to be a fine point. On the one hand, in-house IT staff are not necessarily going to be well-informed about privacy issues and the nature of the data they are working with. On the other hand, it is perfectly possible to make such constraints clear to contractors and to make them part of the contract.
It seems to me that there are several other issues here as well. For instance, why would any programmer be working with the whole, real database? I can see that if the job is convert an
irregularly formatted text file into a usable database, but that is about the only situation in which the programmer needs the real data. Otherwise he or she just needs to know what the data looks like. If sample data is needed, it can be a small subset, and critical information can be camouflaged. Of course, the same applies to the
programmer asking for help on RentACoder. There's no need for him to post his whole database.
It seems to me that the real problems here are:
the programmers shouldn't have been working with the full, real database in the first place
I'd be interested in knowing how important Indian programmers consider the availability of software localized for Indian languages (a) for their own work and (b) for the consumers of their software.
I'd also be interested in knowing how much progress they think has been made in this direction.
It seems to me that this piece conflates two issues:
Should the net be controlled by large corporations?
Should the content of the net be regulated?
and that it gets the priorities backwards. It only briefly addresses the problem of having a network controlled by large corporations and focusses on regulation. In my view, corporate control is dangerous, as is regulation.
The primary problem with corporate control is that the corporations will act in their own business interests rather than in the interests of users and people in general. So far things haven't been too bad, but it is easy to see what could happen.
We could get lockin to particular proprietary technologies, e.g. MS Windows and IE, including things like DRM and spyware. Furthermore, precisely because corporations are not governments, they are exempt from constraints on
censorship such as the First Amendment in the United States. They could censor content in their own interests. So I would like to see control of the net taken away from the big corporations.
However, transferring control to governments is
also a bad idea, precisely because that will facilitate regulation. The fact is, most countries in the world are not open and democratic. Many, probably most governments engage in censorship and would do what they could to censor the net.
There is a long-standing movement in the United
Nations for a "New International Communication Order". Some of the arguments for this reflect the legitiamte desire of less developed countries not to be dominated by rich, developed countries, but
the actual proposals that have been made periodically in the UN, particularly by UNESCO,
have clearly had censorship as their primary objective. The current political movement to transfer control of the net to governments is just the latest incarnation of this movement.
The argument for regulation made in the BBC piece is weak. It merely repeats tired old arguments that violent publications (whether on the net or on paper) foster violence and that there is too much porn. The evidence for this is incredibly weak. And in view of the very limited harm that certain kinds of content can be argued to do, as opposed to the very great harm that censorship would do, it seems clear to me that facilitating censorship is a bad idea.
According to this
AP item in USA Today, the ABA has already been opposing increased oversight of antitrust settlements by the courts. The appointment of a Microsoft lawyer as Chair of the Antitrust Section may not be so much a matter of the fox guarding the chicken coop as the recruitment of an experienced and committed anti-anti-trust lawyer to help the ABA pursue its agenda. It would be interesting to know whether the ABA is actually soft on anti-trust enforcement or whether it perhaps regards judicial oversight as improper
interference with the relationship between the two parties.
Taking the more specific point first, the fact
that assembly language is readable doesn't mean that binary machine language is. You seem to be conflating the two. With compiled languages, there are generally three main levels of representation
source (e.g. C)
assembly language
machine language
There may of course be more, since there can be one or more levels of preprocessing of the source
and one might distinguish between relocatable and absolute object code, and the machine code may be
translated into microcode. People still sometimes write in assembly language and sometimes read it, because the program was originally written in assembly language, because they are testing a compiler, or because they have disassembled some machine code, but they almost never read machine code directly. The example you give is of reading assembly language, not machine code.
People do occasionally write machine language directly - I've done it myself (in binary, not hex, entered through front panel switches), but reading machine language is extremely difficult.
Even in the days when much more programming and debugging was done in assembler and machine language, the few people who were any good at reading raw machine language were regarded as gurus. I don't know if anybody can do it today.
So it isn't at all misleading to describe source code as what humans use and machine language as
usable only by the computer.
For most software, the differences among Linux distributions are immaterial; if you port to one,
it will run on all. In fact, in most cases, so long
as the CPU is the same, the binaries will be compatible. For that matter, most properly written software will be portable, at the source level, among POSIX-compliant systems, meaning not only Linux but a wide range of other UNIX systems.
Except in the very unusual case in which different distributions use different versions of the kernel that differ in what system calls they support, and
where your software makes use of these system calls, the differences among distributions are entirely a matter of what versions of what libraries they come with, and what other software. That means that software that compiles and runs on one distribution can always be compiled and run on another; the difference will be that in some cases the person doing the build will have to install a library or a program that did not come with the distribution. That is generally not a big deal.
If your software requires something exotic, you can also provide statically-linked binaries that
incorporate it for those who don't have the necessary library.
I've never encountered a problem due to differences among distributions. I have been using
Mandrake on my own machines for the last several years. I have had no problem compiling software that I write on the Red Hat machines in our lab.
In fact, I rarely encounter any problem compiling
my software on our Suns. (When I do it is almost
always because I am using GNU extensions that
Sun libc does not support.)
The variety of distributions may seem confusing
and chaotic to non-Linux people, and at the level of the desktop, I can see how inexperienced users
would find the differences offputting. But it really isn't a problem for developing or porting
software.
I don't think what they did is treason, but it
certainly was criminal. What is even more offensive is that the Attorney General, John Ashcroft, lied
about what happened. I heard him interviewed on TV
and he said that Arar was detained and deported
because he had come to the United States to do harm to Americans. It is UNDISPUTED that Arar was in
transit and had no intention of leaving the airport. There was no need to deport him because his intention was to
leave the US for Canada. What the US did was immoral and in violation of international law, and
the Attorney General lied about it. He also said that they had received assurances from the Syrians that Arar would not be tortured. Either he's lying or he's too stupid and naive to be Attorney General. Both the affair and John Ashcroft are
a disgrace to the United States.
I hate to tell you this, but those of us who
predate TeX used *roff to write all sorts of things,
including my dissertation. Writing elaborate man
pages (e.g. with tables) is probably painful
for anyone who doesn't already know *roff well,
but most man pages require only a handful of
macros and no understanding of how to write them or any of the other fine points.
As I underestand it, if MS patents their file formats, that will prevent anyone without a license from generating files in MS Office formats, but it will not prevent people from displaying them or converting the information into other formats.
That's because such patents are for methods of "storing" information. I know this seems pedantic, but law is pedantic, and I'm thinking of the precedent of LZW compression. Without a license, you couldn't generate GIF images but you could display them and convert them. So, although I'm distrustful of Microsoft (and don't use their products), and opposed to software patents, perhaps these patents aren't as dangerous as they seem. Any lawyers know for sure?
Are software patents worse for OSS?
on
Perens on Patents
·
· Score: 1
It seems to me that software patents are worse for
open source software because, in general, it is hard to tell whether a closed source program infringes or not. If the patent concerns the interface between the program and the outside world, then it is easy to tell, even if it is closed source. For example,
if a certain kind of compression is patented, any program that produces output using that kind of compression infringes unless it is licensed. But
if the patent concerns something purely internal,
it will often be difficult or impossible to tell.
For example, if a program uses a patented sort
algorithm, how can anyone tell without reading the source? Performance may provide a hint, but in many cases it will be difficult to tell whether the performance is due to a better algorithm in one area or an improvement of another kind. And even if, say, the complexity function looks like
that for a patented algorithm, who is to say that
the program is not using another algorithm with the same complexity function?
The push for software patents may not be
targetted at OSS, but it seems to me that they
pose much more of a problem for OSS than for closed source software.
The US Army doesn't seem to have the same take as SCO on the threat to US security posed by FOSS.
According to The Register, it has abandoned MSWindows and shifted to Linux for the Commander's Digital Assistant component of its Land Warrior program.
I've had plenty of experience with Immigration Canada and wouldn't characterize them as nasty at all. Like most other countries, Canada is careful about who comes to live and work there. Canada doesn't want criminals, and doesn't want people who won't
be able to support themselves. But in my experience Immigration Canada was never nasty and, as bureaucracies go, pretty reasonable, even, in several cases, accomodating. And it isn't true that only people with graduate degrees in urgently needed fields get in. Canada admits
substantial numbers of people as refugees and in the family unification category. You get a lot of points if you have a firm job offer, even if you're in a field that isn't urgently needed. The list of fields in which immigrants are desired is by no means limited to professionals. It includes a wide variety of occupations. Its true that if you're unskilled and have limited education you don't have much chance of getting in as an independent immigrant, but it certainly isn't critical to have a graduate degree.
Companies have been known to sue even when even just
one portion of the name is similar. There was another case recently that was similar in that it involved a big American company going after a tiny Canadian outfit. In this case, Starbucks (no doubt the provider of cafeine to many/.-ers) sued Haidabucks Cafe, a small cafe owned by Haida Indians in Masset, British Columbia. The names are obviously quite different, both in writing and in speech.
Fortunately, they stood their ground and obtained
the services of a top notch law firm and a web site designer, with the result that Starbucks backed down. That's a good thing, and not just for them: boycotting Starbucks is tough!
The importance of this decision doesn't lie only in its liberal approach to fair use. It is also important because it acknowledges that even a device that can, or even is, used in an infringing way should be permitted if it also has non-infringing uses. This issue comes up over and over again, e.g. in the attempt by DirectTV to treat all purchasers of smartcards as thieves.
Anything from a pry-bar to a debugger CAN be used
to commit a crime or violate a copyright, so the doctrine that the possibility of infringing use
doesn't justify prohibition or restriction is important for civil liberties in general.
Reading through the source isn't the only way in which people improve free software. Another way is by encountering bugs when they use it and fixing them, or by discovering that a feature would be useful and adding it.
This happens even with software that has been around for decades. To take a small example from my personal experience, some years ago I added a feature (the figname keyword) to GNU pic, the figure-drawing preprocessor for troff/groff and TeX. I found this feature very useful and eventually submitted a patch to the maintainers. They incorporated it, and it is now part of GNU pic.
I would assume that the main benefits would be of most use to (in order):
Other Semitic languages such as Arabic, Syriac, and Ethiopian
Other right-left languages, such as Farsi.
The Semitic languages written in the Ethiopic alphabet, such as Amharic and Tigrigna,
are written left to right.
Although unquestionably a forgery, the Protocols
of the Elders of Zion is taken seriously by more than a few cranks. It is widely circulated in the Arab world where it is used to stir up hatred of Jews.
According to the Egyptian newspaper al-Usbu', the new Alexandria Library, a showpiece with substantial international support, put an Arabic translation of the
Protocols of the Elders of Zion on display in the same case as the Torah, claiming that it is a central work of Jewish thought. After much criticism, they have now removed it.
This posting
on Language Log contains details and links. Similarly,
Palestinian Authority Educational Television
has repeatedly broadcast "educational" programs
about the Protocols in which it is treated as real. The most recent instance of this was on December 28th. Palestine Media Watch issued
this report, which includes an English-language transcript.
Before someone starts a flame war, this is not
a statement about the situation in Iraq, Israel,
Islam, Al-Qaeda or any other issue.
Slashdot Mirror
If I understand the limitation of liability clause, SCO refuses to indemnify licensees if they should prove unable to use the product or otherwise incur damages due to IP issues. Not only is there the little problem of the GPL, but we can't be sure that there isn't a little infringment lurking in SCO's past, can we? If indemnification is so important for open source vendors, how come it isn't for SCO? What's sauce for the goose is sauce for the gander.
If I understand the technique correctly, it relies on information specific to individual users. Unless there is a way for users to export their information, that means that the filtering can only be done after the email reaches its destination, not by the ISP or central mail server. So it may be helfpul to individual users, but unlike some proposed techniques, it won't cut down on total email traffic.
The scientists signing the letter do not represent the Union of Concerned Scientists. They are an independent group who are merely endorsing the UCS report. Furthermore, they include scientists who are not particularly left-wing, such as H-bomb designer Richard Garwin and physicist Norman Ramsey, both of whom served as advisers to Republican administrations. According to this news item, organizations opposing the Bush administration policy include: the National Academies of Science, the American Association for the Advancement of Science, the Federation of American Scientists, and the Association of American Medical Colleges. The opposition isn't coming from the left fringe; it is mainstream.
There's a good reason that the EFF isn't mounting a campaign against the RIAA: some of the RIAA's tactics may be scummy, and the penalties provided by the law excessive, but they're basically in the right. People who post and download copyrighted songs that they haven't already purchased in another format without the copyright holder's permission are in fact infringing valid copyrights. The purpose of the EFF is to defend civil liberties on-line, not the "right" to rip-off and distribute other people's property.
There are at least two ways the airport can pay for this. One is to include it in the fees charged to the airlines: landing fees, rental of hangar and counter space, various other services. Another is to include it in the rent paid by shops and restaurants.
The real story here is that companies and other organizations and institutions are setting machines up as servers and are too stupid to create an appropriate robots.txt file and/or keep their confidential information elsewhere. Google doesn't just drop in, even on networked machines. I have some sympathy for individuals who don't understand what they are doing when they make their machine a server, but surely any professional sysadmin, even one with limited training and experience, should know better than this. It's the same as leaving your briefcase on the front seat of an unlocked car.
What happened here is certainly appalling, but I'm not so sure that outsourcing is the main problem. Outsourcing arguably increases the risk of problems of this sort because an in-house programmer is more likely to know the rules of the game, but this seems to me to be a fine point. On the one hand, in-house IT staff are not necessarily going to be well-informed about privacy issues and the nature of the data they are working with. On the other hand, it is perfectly possible to make such constraints clear to contractors and to make them part of the contract.
It seems to me that there are several other issues here as well. For instance, why would any programmer be working with the whole, real database? I can see that if the job is convert an irregularly formatted text file into a usable database, but that is about the only situation in which the programmer needs the real data. Otherwise he or she just needs to know what the data looks like. If sample data is needed, it can be a small subset, and critical information can be camouflaged. Of course, the same applies to the programmer asking for help on RentACoder. There's no need for him to post his whole database.
It seems to me that the real problems here are:
I'd be interested in knowing how important Indian programmers consider the availability of software localized for Indian languages (a) for their own work and (b) for the consumers of their software. I'd also be interested in knowing how much progress they think has been made in this direction.
It seems to me that this piece conflates two issues:
- Should the net be controlled by large corporations?
- Should the content of the net be regulated?
and that it gets the priorities backwards. It only briefly addresses the problem of having a network controlled by large corporations and focusses on regulation. In my view, corporate control is dangerous, as is regulation.The primary problem with corporate control is that the corporations will act in their own business interests rather than in the interests of users and people in general. So far things haven't been too bad, but it is easy to see what could happen. We could get lockin to particular proprietary technologies, e.g. MS Windows and IE, including things like DRM and spyware. Furthermore, precisely because corporations are not governments, they are exempt from constraints on censorship such as the First Amendment in the United States. They could censor content in their own interests. So I would like to see control of the net taken away from the big corporations.
However, transferring control to governments is also a bad idea, precisely because that will facilitate regulation. The fact is, most countries in the world are not open and democratic. Many, probably most governments engage in censorship and would do what they could to censor the net. There is a long-standing movement in the United Nations for a "New International Communication Order". Some of the arguments for this reflect the legitiamte desire of less developed countries not to be dominated by rich, developed countries, but the actual proposals that have been made periodically in the UN, particularly by UNESCO, have clearly had censorship as their primary objective. The current political movement to transfer control of the net to governments is just the latest incarnation of this movement.
The argument for regulation made in the BBC piece is weak. It merely repeats tired old arguments that violent publications (whether on the net or on paper) foster violence and that there is too much porn. The evidence for this is incredibly weak. And in view of the very limited harm that certain kinds of content can be argued to do, as opposed to the very great harm that censorship would do, it seems clear to me that facilitating censorship is a bad idea.
According to this AP item in USA Today, the ABA has already been opposing increased oversight of antitrust settlements by the courts. The appointment of a Microsoft lawyer as Chair of the Antitrust Section may not be so much a matter of the fox guarding the chicken coop as the recruitment of an experienced and committed anti-anti-trust lawyer to help the ABA pursue its agenda. It would be interesting to know whether the ABA is actually soft on anti-trust enforcement or whether it perhaps regards judicial oversight as improper interference with the relationship between the two parties.
Taking the more specific point first, the fact that assembly language is readable doesn't mean that binary machine language is. You seem to be conflating the two. With compiled languages, there are generally three main levels of representation
There may of course be more, since there can be one or more levels of preprocessing of the source and one might distinguish between relocatable and absolute object code, and the machine code may be translated into microcode. People still sometimes write in assembly language and sometimes read it, because the program was originally written in assembly language, because they are testing a compiler, or because they have disassembled some machine code, but they almost never read machine code directly. The example you give is of reading assembly language, not machine code.
People do occasionally write machine language directly - I've done it myself (in binary, not hex, entered through front panel switches), but reading machine language is extremely difficult. Even in the days when much more programming and debugging was done in assembler and machine language, the few people who were any good at reading raw machine language were regarded as gurus. I don't know if anybody can do it today. So it isn't at all misleading to describe source code as what humans use and machine language as usable only by the computer.
Yes, it is possible to read assembler. But since the MIT crew's handout referred to binary, that isn't relevant. Assembly language is not binary.
For most software, the differences among Linux distributions are immaterial; if you port to one, it will run on all. In fact, in most cases, so long as the CPU is the same, the binaries will be compatible. For that matter, most properly written software will be portable, at the source level, among POSIX-compliant systems, meaning not only Linux but a wide range of other UNIX systems.
Except in the very unusual case in which different distributions use different versions of the kernel that differ in what system calls they support, and where your software makes use of these system calls, the differences among distributions are entirely a matter of what versions of what libraries they come with, and what other software. That means that software that compiles and runs on one distribution can always be compiled and run on another; the difference will be that in some cases the person doing the build will have to install a library or a program that did not come with the distribution. That is generally not a big deal. If your software requires something exotic, you can also provide statically-linked binaries that incorporate it for those who don't have the necessary library.
I've never encountered a problem due to differences among distributions. I have been using Mandrake on my own machines for the last several years. I have had no problem compiling software that I write on the Red Hat machines in our lab. In fact, I rarely encounter any problem compiling my software on our Suns. (When I do it is almost always because I am using GNU extensions that Sun libc does not support.)
The variety of distributions may seem confusing and chaotic to non-Linux people, and at the level of the desktop, I can see how inexperienced users would find the differences offputting. But it really isn't a problem for developing or porting software.
I don't think what they did is treason, but it certainly was criminal. What is even more offensive is that the Attorney General, John Ashcroft, lied about what happened. I heard him interviewed on TV and he said that Arar was detained and deported because he had come to the United States to do harm to Americans. It is UNDISPUTED that Arar was in transit and had no intention of leaving the airport. There was no need to deport him because his intention was to leave the US for Canada. What the US did was immoral and in violation of international law, and the Attorney General lied about it. He also said that they had received assurances from the Syrians that Arar would not be tortured. Either he's lying or he's too stupid and naive to be Attorney General. Both the affair and John Ashcroft are a disgrace to the United States.
I hate to tell you this, but those of us who predate TeX used *roff to write all sorts of things, including my dissertation. Writing elaborate man pages (e.g. with tables) is probably painful for anyone who doesn't already know *roff well, but most man pages require only a handful of macros and no understanding of how to write them or any of the other fine points.
As I underestand it, if MS patents their file formats, that will prevent anyone without a license from generating files in MS Office formats, but it will not prevent people from displaying them or converting the information into other formats. That's because such patents are for methods of "storing" information. I know this seems pedantic, but law is pedantic, and I'm thinking of the precedent of LZW compression. Without a license, you couldn't generate GIF images but you could display them and convert them. So, although I'm distrustful of Microsoft (and don't use their products), and opposed to software patents, perhaps these patents aren't as dangerous as they seem. Any lawyers know for sure?
It seems to me that software patents are worse for open source software because, in general, it is hard to tell whether a closed source program infringes or not. If the patent concerns the interface between the program and the outside world, then it is easy to tell, even if it is closed source. For example, if a certain kind of compression is patented, any program that produces output using that kind of compression infringes unless it is licensed. But if the patent concerns something purely internal, it will often be difficult or impossible to tell. For example, if a program uses a patented sort algorithm, how can anyone tell without reading the source? Performance may provide a hint, but in many cases it will be difficult to tell whether the performance is due to a better algorithm in one area or an improvement of another kind. And even if, say, the complexity function looks like that for a patented algorithm, who is to say that the program is not using another algorithm with the same complexity function?
The push for software patents may not be targetted at OSS, but it seems to me that they pose much more of a problem for OSS than for closed source software.
The US Army doesn't seem to have the same take as SCO on the threat to US security posed by FOSS. According to The Register, it has abandoned MSWindows and shifted to Linux for the Commander's Digital Assistant component of its Land Warrior program.
I've had plenty of experience with Immigration Canada and wouldn't characterize them as nasty at all. Like most other countries, Canada is careful about who comes to live and work there. Canada doesn't want criminals, and doesn't want people who won't be able to support themselves. But in my experience Immigration Canada was never nasty and, as bureaucracies go, pretty reasonable, even, in several cases, accomodating. And it isn't true that only people with graduate degrees in urgently needed fields get in. Canada admits substantial numbers of people as refugees and in the family unification category. You get a lot of points if you have a firm job offer, even if you're in a field that isn't urgently needed. The list of fields in which immigrants are desired is by no means limited to professionals. It includes a wide variety of occupations. Its true that if you're unskilled and have limited education you don't have much chance of getting in as an independent immigrant, but it certainly isn't critical to have a graduate degree.
In addition to the scientific value, that image makes terrific wallpaper, and it is scaled perfectly for my monitor.
Companies have been known to sue even when even just one portion of the name is similar. There was another case recently that was similar in that it involved a big American company going after a tiny Canadian outfit. In this case, Starbucks (no doubt the provider of cafeine to many /.-ers) sued Haidabucks Cafe, a small cafe owned by Haida Indians in Masset, British Columbia. The names are obviously quite different, both in writing and in speech.
Fortunately, they stood their ground and obtained
the services of a top notch law firm and a web site designer, with the result that Starbucks backed down. That's a good thing, and not just for them: boycotting Starbucks is tough!
The importance of this decision doesn't lie only in its liberal approach to fair use. It is also important because it acknowledges that even a device that can, or even is, used in an infringing way should be permitted if it also has non-infringing uses. This issue comes up over and over again, e.g. in the attempt by DirectTV to treat all purchasers of smartcards as thieves. Anything from a pry-bar to a debugger CAN be used to commit a crime or violate a copyright, so the doctrine that the possibility of infringing use doesn't justify prohibition or restriction is important for civil liberties in general.
Reading through the source isn't the only way in which people improve free software. Another way is by encountering bugs when they use it and fixing them, or by discovering that a feature would be useful and adding it.
This happens even with software that has been around for decades. To take a small example from my personal experience, some years ago I added a feature (the figname keyword) to GNU pic, the figure-drawing preprocessor for troff/groff and TeX. I found this feature very useful and eventually submitted a patch to the maintainers. They incorporated it, and it is now part of GNU pic.
The Semitic languages written in the Ethiopic alphabet, such as Amharic and Tigrigna, are written left to right.
Although unquestionably a forgery, the Protocols of the Elders of Zion is taken seriously by more than a few cranks. It is widely circulated in the Arab world where it is used to stir up hatred of Jews. According to the Egyptian newspaper al-Usbu', the new Alexandria Library, a showpiece with substantial international support, put an Arabic translation of the Protocols of the Elders of Zion on display in the same case as the Torah, claiming that it is a central work of Jewish thought. After much criticism, they have now removed it. This posting on Language Log contains details and links. Similarly, Palestinian Authority Educational Television has repeatedly broadcast "educational" programs about the Protocols in which it is treated as real. The most recent instance of this was on December 28th. Palestine Media Watch issued this report, which includes an English-language transcript.
Before someone starts a flame war, this is not a statement about the situation in Iraq, Israel, Islam, Al-Qaeda or any other issue.