There's no reason to link patents and copyright as all. Patents are a monstrosity that is destroying any sort of commercial innovation or scientific progress while killing poor children and destroying the economy with lawyer-bound monopolies supported by patent exchange agreements.
Copyright, on the other hand, seems to have socially redeeming value. It allows the creation of at least three types of expressive work that would be much more difficult in a world without copyright: High budget Movies, TV Shows, and Video Games.
Twenty years is still far too long. I think the right number is five years. That way the producers get 95+% of the potential revenues, and people get to remix (think "reuse" for code) it before it's completely culturally obsolete. In order to qualify for the 5 year protection, the author should be required to escrow the preferred modification form of the work with the copyright office for release when the copyright expires.
The only difference is that there is no expiration on the contracts, and therefore the information would never reach the public domain. Under copyright law, the information in theory will eventually reach the public domain and be usable by society as a whole without an arduous contract.
The difference between "life of the author + 70 years" and "forever" is basically irrelevant. Anyone who was alive when the work was released will be dead (at least retired in the "they were a baby" case) when is released to the public domain. In the contract case, it only takes one person ignoring the contract to break the chain, and then everyone who got it from them is unencumbered - much better.
...quite a social damper can be put on by a remark such as "yeah, I've heard/seen that tons of times... got anything new?"
For little kids, I think your argument has merit. They don't know if what they are seeing is "old", they just know if it is entertaining. Unfortunately, with socialization comes competition, and that means chasing the leading edge as they hit the age where brands, labels, and fashion matter. Just my opinion, but I think you'll find I'm not too far off the mark.
You are absolutely right, and that's why a 5 year copyright term would still be wildly profitable for the movie companies. But... it's damn hard to argue that people sharing movies on the internet are doing something wrong when no content released in their entire lives has been released into the public domain.
I'm not going to argue that there would be no demand for pirated new releases if we had reasonable copyright terms, but as it is the most recent content in the public domain is so archaic that it may have been produced before today's college student's parents were born. That's definately not even slightly culturally relevant anymore.
With a 10 year copyright term for video content (which I think is a bit long), we'd be talking about movies like Pulp Fiction, The Usual Suspects, and Toy Story. You've got to agree that those are a hell of a lot more culturally relevant than whatever movies were being made before World War II. Sure, there are a couple of slightly more recent exceptions where copyright didn't get renewed due to legal technicalities, but largely *nothing* for video content is in the public domain that anyone alive remembers seeing when it first came out.
If copyright expired after 5 or 10 years, they could download legal movies and TV shows made recently enough to be culturally relevent. In that world, asking them not to download the most recent episode of 24 or the chinese CAM release of Spiderman 3 is much more defensible.
Replacements exist for everything if you know where to look.
Right. And for the core OS components like libc, those "replacements" basically involve dropping in an entire other Unix-like OS like Solaris or *BSD.
Sure, getting everyone in the world to call the GNU/Linux OS by its proper name is pretty unlikely. That doesn't make it correct to call the GNU System "Linux". No, the fact that the GPL allows people to reuse GNU software without credit doesn't matter - except legally. And finally: No, the kernel is not a larger component of the operating system than all the non-kernel code combined.
In conclusion, there's no legal reason why Red Hat can't call their product "Red Hat Linux" and ignore the GNU Project. I even agree that people who correct others for saying "Linux" instead of "GNU/Linux" are being annoying. Still, the idea that Linux distros aren't basically the GNU system because they use a different kernel is absurd, and trying to correct people who say "GNU/Linux" to "Linux" is just being an asshole.
Let's try to run them on BSD. Wow, it works. So why isn't Stallman asking for people to call it "GNU/FreeBSD"? Read his answer to that in his FAQ. It's hilarious. He might as well have written "because Linux is the popular one, I don't care about BSD".
Or maybe because BSD Unix is a full OS in its own right, with its own libc, it's own shell, etc?
Not going to work without X.org, either. So why don't we call it "X11/Linux" instead?
Every major Unix-like OS includes X. We don't talk about GNU/Linux/X.org for the same reason that we don't talk about OpenBSD/X.org or Solaris/X.
Never been to embedded much, have you?
Note that I said "unix-like". It's not a very Unix-like system without a full libc, a shell, basic unix tools, etc - now is it?
There are replacements for anything GNU has ever published. From bash to GCC to Emacs and everything else. Using the GNU tools is just convenient. Again, not to minimize their importance, but creating a bunch of tools doesn't give you an operating system and it doesn't entitle you to append your name to other people's work.
The only "replacements" for the GNU system are to drop in the non-kernel OS components from one of the other Free operating systems like Solaris or *BSD. Not that anyone has successfully done that on top of Linux AFAIK. If anything that proves the point that you don't have a Unix-like Linux based OS without GNU.
One definition of an "operating system" is "the software you need before you can run applications".
Let's take some applications: GNU Emacs, Abiword, Firefox, FreeCiv, Apache.
Let's try to run those on just Linux, the kernel. Not going to work. Clearly, Linux is not an operating system. Similarly, trying to run them on the GNU System would be hard given the lack of a kernel - so the GNU System is not a complete operating system. On the other hand, they all work fine on the GNU/Linux operating system (which includes the X Window System - just like any other Unix-like OS does).
Interestingly, people have built GNU-based operating systems without Linux but I don't know of any attempt to build a Linux-based unix-like OS without GNU.
The Herd is mostly interesting as a microkernel research project. Functional Free Software monolithic kernels are pretty common now (Linux, Solaris, the various BSDs), so developing another one wouldn't be any more interesting than trying to develop another [Main Menu/Window Manager/File Manager] Unix desktop environment. This is pretty obvious, which is why the Hurd development effort has been pretty low key.
The really interesting thing about the Hurd (and some other similar projects) is that when they do get it working and manage to take advantage of some of the security / stability properties of microkernels it will be a drop in replacement for the monolithic kernels that exist - there won't be a giant application incompatibility leap like there would be if we tried to switch to Coyotos or something.
The goal of the GNU project was to create an operating system. In order to accomplish this goal, they started by building the tools necessary to build an operating system - like gcc. By 1991, they had the whole system done - an entire OS, including a compiler, a C library, a shell, userspace tools - everything except one last piece, the kernel.
Now, I'm not going to say that Linus's kernel-space contribution is trivial (although it's not necessary to run the GNU System), but you can't seriously argue that because someone else developed the last few percent of the OS that the GNU Project doesn't deserve to have their name mentioned when talking about the OS they built.
It's perfectly reasonable to compare the price of an "Ubuntu Solution" to a "Microsoft Solution". Open Office is 100% supported by Canonical given a paid Ubuntu support agreement, the same is not true for Open Office on Windows.
It's "Linux", BTW. Most of us don't agree with Stallman's arguments about naming conventions.
How can you possibly be that big a jerk? The GNU Project spent years developing the majority of operating system you're using, and you can't even accept other people giving them credit? I mean, it's not like "Linux" would be much of an OS without GNU libc...
Copy a plugin into/Library/foo? Install a kernel extension? Delete all user documents?
The first two are equivalent to the last one from a security standpoint. Any code running on the machine at all can delete all user documents. There is room for better security, but it will require the introduction of a sandbox / capabilities model and some way to get developers to use it - the only system that I know of that does anything like what I'm talking about is the OLPC OS.
Computers don't exist simply to run Microsoft Office.
Take "creating textual documents". Sometimes a word processor like OOWriter is appropriate, but other times there is a better tool. Sometimes you want a desktop publishing program like Scribus, or a document processor like LyX. You may even really want an HTML editor like Bluefish.
Or image editing. Microsoft office really doesn't do that. Ubuntu comes with GIMP by default, but also provides tools OODraw and Inkscape for when a raster image editor is not appropriate.
Number/data crunching. Sometimes you want a spreadsheet like OOCalc. Sometimes you want a high end toolkit like Numeric Python or even a Fortran compiler.
Sure, you can get tools for most of these tasks on Windows. Sometimes those tools even have features that the Free Software programs that come with Ubuntu don't. Even ignoring software freedom for a moment (which is never a good idea), each one of those programs for Windows requires an expensive single seat software license. If one person wants to legally use Windows, Microsoft Office, Photoshop, Dreamweaver, Acrobat, and Illustrator (for example), that'll cost almost $3000 at full retail. Sure, maybe you can get discounts. On the other hand, maybe that person uses other software occasionally too - and that doesn't even consider later upgrade costs. With Ubuntu, you get very similar functionality to all of that built in at zero cost - for every user in your organization.
Maybe some of the Windows stuff is more feature packed - but with Ubuntu, every user gets all of it automatically. You never have to think "I only use photoshop every two weeks, is it really worth getting the license for *my* desktop?"
2. Printing. Printing on Linux is a pain. It has been a pain since day one. But I know of more than one person that has had printing problems with Vista. I would call printing a tie.
Buy supported hardware. Well supported printers work great on Linux, with setup being completely painless. (i.e. less painful than inserting a driver disk even). This Linux works poorly with unsupported hardware stuff was old in 2002 - you don't complain about Windows not supporting PowerPC processors. You can replace every single piece of poorly supported hardware for less than the cost of office - and the supported hardware is usually nicer anyway.
Ubuntu has issues with detecting monitors.
This is really embarrassing. It doesn't always happen, and it's not hard to fix, but when it does happen it means that someone's going to have to edit a dreaded config file. This was supposed to get fixed in Ubuntu 7.04, but it got put off until 7.10 because the X.org guys are working on it.
Ubuntu is having some issues with Wifi. A lot of people are having problems even when their wifi card is in the kernal and worked under the last version of Ubuntu.
This is one specific card, not some widespread wifi issue. I agree that losing support for a piece of hardware is pretty lame. Still though, this is an issue with a specific piece of hardware that is now not supported in Ubuntu 7.04. There's no reason to mention it in a review / comparison at all, aside from perhaps "No OS supports all hardware. Make sure your hardware is supported before installing any OS."
Actually we are implementing pay for service very soon (mainly to cover the cost of re-wiring our older buildings as well as wiring newly purchased properties).. sooo... you're right about this one;)
If you actually implement a cost structure where you're charging based on the variable element of the user's usage that costs *you* money - i.e. upstream bandwidth charges, rewiring buildings to avoid internal bottlenecks, etc. If you set it up right you can completely eliminate the idea of treating heavy use as abuse.
Bandwidth isn't unlimited, there's only so much of it. Buying more capacity costs money.
Faced with more bandwidth demand than supply and no income from bandwidth charges, a University IT department (or an ISP selling "Unlimited Internet") will tend to look at their bandwidth usage and make calls like "60% of our bandwidth usage is coming from these 8 guys running servers. If we just ban servers we won't need to upgrade". So they implement that policy by blocking incoming TCP connections with a firewall and it buys them a couple months - at the horrific hidden cost of turning what was previously a general internet connection into a "consume only" connection. The first google servers were in Larry Paige's dorm room - who knows how college students might take advantage of having a general purpose internet connection...
Eventually, bandwidth usage will rise again and the IT department will be faced with the same situation: "60% of our bandwidth usage is coming from users doing P2P file sharing". But they can't fix this with a single firewall rule - so they're faced with the decision of buying active traffic shaping hardware or upgrading their bandwidth. Traffic shaping is cheaper than adding capacity, so they do that. They think they're only hurting "a couple students claiming to download Linux distros", but they're really killing ideas like blogtorrent. Basically, this is the same as the server block - it just costs more money in hardware.
If, instead, the school just charged for bandwidth - perhaps $0.25 for each gig after the first 5 gigs/month - this would play out completely differently. There would be an incentive for users to not outright waste bandwidth, and when it came time to chose between upgrading and degrading the school internet connection it would always be better to upgrade.
You have two problems, they can be handled separately.
First, you get RIAA letters. The appropriate response is a form letter saying that "Our school privacy policy prohibits us from releasing user information without a subpoena or court order" (obviously you'll want to verify that with a lawyer, but you shouldn't be sending out user information based on random letters). If you do get a legit subpoena or court order, send them the info if it's still available.
Second, you have excess bandwidth usage. This is really simple: Charge the students a reasonable fee for bandwidth overages - this will encourage users to conserve without unduly constraining people who actually are willing to pay for their bandwidth. It also has the advantage that as demand increases you automatically have the money to pay for upgrades.
So charge the students for their bandwidth usage at slightly above what it's costing. Use the extra money for upgrades.
Bandwidth is a resource that costs money - giving people "Unlimited" bandwidth and then degrading the internet connection to prevent "excess use" is absurd. Hell, if you give every student the first gig/month free most of them won't even notice the policy change.
After RHEL and Ubuntu, it's a toss up between supporting SLES next or just supporting Solaris next - the difference is about the same.
To be clearer, I should have said: If you test RHEL and Ubuntu, SLES probably works already. If you want a third tested "distro", I suggest Solaris - it's more likely to actually need a couple tweaks and then you get too support a whole new OS from a PR standpoint.
Again, if you're capable of devoting the time to port your application to a new platform then dealing with the packaging issues on Linux is pretty trivial time-wise. A commercial proprietary program is slightly harder than RealPlayer, because you can't rely on the various distros packaging it for you, but even that isn't that big a deal - Unreal Tournament 2004 used the Loki installer which works fine.
I'll admit that if you look at the list of different distributions on a site like Distrowatch, it can look like there's a daunting packaging problem. But... worrying about most of that is sort of like worrying about your software not working on Windows because someone built their own installer - i.e. irrelevant. Test on RHEL and Ubuntu and then consider fixing installer bugs that effect other distributions *if* they come up. After RHEL and Ubuntu, it's a toss up between supporting SLES next or just supporting Solaris next - the difference is about the same.
Slashdot Mirror
There's no reason to link patents and copyright as all. Patents are a monstrosity that is destroying any sort of commercial innovation or scientific progress while killing poor children and destroying the economy with lawyer-bound monopolies supported by patent exchange agreements.
Copyright, on the other hand, seems to have socially redeeming value. It allows the creation of at least three types of expressive work that would be much more difficult in a world without copyright: High budget Movies, TV Shows, and Video Games.
Twenty years is still far too long. I think the right number is five years. That way the producers get 95+% of the potential revenues, and people get to remix (think "reuse" for code) it before it's completely culturally obsolete. In order to qualify for the 5 year protection, the author should be required to escrow the preferred modification form of the work with the copyright office for release when the copyright expires.
The difference between "life of the author + 70 years" and "forever" is basically irrelevant. Anyone who was alive when the work was released will be dead (at least retired in the "they were a baby" case) when is released to the public domain. In the contract case, it only takes one person ignoring the contract to break the chain, and then everyone who got it from them is unencumbered - much better.
You are absolutely right, and that's why a 5 year copyright term would still be wildly profitable for the movie companies. But... it's damn hard to argue that people sharing movies on the internet are doing something wrong when no content released in their entire lives has been released into the public domain.
I'm not going to argue that there would be no demand for pirated new releases if we had reasonable copyright terms, but as it is the most recent content in the public domain is so archaic that it may have been produced before today's college student's parents were born. That's definately not even slightly culturally relevant anymore.
With a 10 year copyright term for video content (which I think is a bit long), we'd be talking about movies like Pulp Fiction, The Usual Suspects, and Toy Story. You've got to agree that those are a hell of a lot more culturally relevant than whatever movies were being made before World War II. Sure, there are a couple of slightly more recent exceptions where copyright didn't get renewed due to legal technicalities, but largely *nothing* for video content is in the public domain that anyone alive remembers seeing when it first came out.
If copyright expired after 5 or 10 years, they could download legal movies and TV shows made recently enough to be culturally relevent. In that world, asking them not to download the most recent episode of 24 or the chinese CAM release of Spiderman 3 is much more defensible.
Why not just pick up a used laptop from a reliable seller on Ebay and drop Ubuntu on it?
Quick search came up with http://cgi.ebay.com/IBM-Thinkpad-T23-P3-1-26GHz-PI II-Laptop-Notebook-1-26_W0QQitemZ320108259669QQihZ 011QQcategoryZ140083QQcmdZViewItem, I'm sure there are tons of similar items, and a 1000 mhz+ PIII is definitely faster than the AMD Geode in the XO.
People can't even seem to accept that that applies to politics. Healthcare is even harder to get people to listen to policy suggestions about.
Right. And for the core OS components like libc, those "replacements" basically involve dropping in an entire other Unix-like OS like Solaris or *BSD.
Sure, getting everyone in the world to call the GNU/Linux OS by its proper name is pretty unlikely. That doesn't make it correct to call the GNU System "Linux". No, the fact that the GPL allows people to reuse GNU software without credit doesn't matter - except legally. And finally: No, the kernel is not a larger component of the operating system than all the non-kernel code combined.
In conclusion, there's no legal reason why Red Hat can't call their product "Red Hat Linux" and ignore the GNU Project. I even agree that people who correct others for saying "Linux" instead of "GNU/Linux" are being annoying. Still, the idea that Linux distros aren't basically the GNU system because they use a different kernel is absurd, and trying to correct people who say "GNU/Linux" to "Linux" is just being an asshole.
Or maybe because BSD Unix is a full OS in its own right, with its own libc, it's own shell, etc?
Every major Unix-like OS includes X. We don't talk about GNU/Linux/X.org for the same reason that we don't talk about OpenBSD/X.org or Solaris/X.
Note that I said "unix-like". It's not a very Unix-like system without a full libc, a shell, basic unix tools, etc - now is it?
The only "replacements" for the GNU system are to drop in the non-kernel OS components from one of the other Free operating systems like Solaris or *BSD. Not that anyone has successfully done that on top of Linux AFAIK. If anything that proves the point that you don't have a Unix-like Linux based OS without GNU.
One definition of an "operating system" is "the software you need before you can run applications".
Let's take some applications: GNU Emacs, Abiword, Firefox, FreeCiv, Apache.
Let's try to run those on just Linux, the kernel. Not going to work. Clearly, Linux is not an operating system. Similarly, trying to run them on the GNU System would be hard given the lack of a kernel - so the GNU System is not a complete operating system. On the other hand, they all work fine on the GNU/Linux operating system (which includes the X Window System - just like any other Unix-like OS does).
Interestingly, people have built GNU-based operating systems without Linux but I don't know of any attempt to build a Linux-based unix-like OS without GNU.
The Herd is mostly interesting as a microkernel research project. Functional Free Software monolithic kernels are pretty common now (Linux, Solaris, the various BSDs), so developing another one wouldn't be any more interesting than trying to develop another [Main Menu/Window Manager/File Manager] Unix desktop environment. This is pretty obvious, which is why the Hurd development effort has been pretty low key.
The really interesting thing about the Hurd (and some other similar projects) is that when they do get it working and manage to take advantage of some of the security / stability properties of microkernels it will be a drop in replacement for the monolithic kernels that exist - there won't be a giant application incompatibility leap like there would be if we tried to switch to Coyotos or something.
The goal of the GNU project was to create an operating system. In order to accomplish this goal, they started by building the tools necessary to build an operating system - like gcc. By 1991, they had the whole system done - an entire OS, including a compiler, a C library, a shell, userspace tools - everything except one last piece, the kernel.
Now, I'm not going to say that Linus's kernel-space contribution is trivial (although it's not necessary to run the GNU System), but you can't seriously argue that because someone else developed the last few percent of the OS that the GNU Project doesn't deserve to have their name mentioned when talking about the OS they built.
It's perfectly reasonable to compare the price of an "Ubuntu Solution" to a "Microsoft Solution". Open Office is 100% supported by Canonical given a paid Ubuntu support agreement, the same is not true for Open Office on Windows.
How can you possibly be that big a jerk? The GNU Project spent years developing the majority of operating system you're using, and you can't even accept other people giving them credit? I mean, it's not like "Linux" would be much of an OS without GNU libc...
The first two are equivalent to the last one from a security standpoint. Any code running on the machine at all can delete all user documents. There is room for better security, but it will require the introduction of a sandbox / capabilities model and some way to get developers to use it - the only system that I know of that does anything like what I'm talking about is the OLPC OS.
Computers don't exist simply to run Microsoft Office.
Take "creating textual documents". Sometimes a word processor like OOWriter is appropriate, but other times there is a better tool. Sometimes you want a desktop publishing program like Scribus, or a document processor like LyX. You may even really want an HTML editor like Bluefish.
Or image editing. Microsoft office really doesn't do that. Ubuntu comes with GIMP by default, but also provides tools OODraw and Inkscape for when a raster image editor is not appropriate.
Number/data crunching. Sometimes you want a spreadsheet like OOCalc. Sometimes you want a high end toolkit like Numeric Python or even a Fortran compiler.
Sure, you can get tools for most of these tasks on Windows. Sometimes those tools even have features that the Free Software programs that come with Ubuntu don't. Even ignoring software freedom for a moment (which is never a good idea), each one of those programs for Windows requires an expensive single seat software license. If one person wants to legally use Windows, Microsoft Office, Photoshop, Dreamweaver, Acrobat, and Illustrator (for example), that'll cost almost $3000 at full retail. Sure, maybe you can get discounts. On the other hand, maybe that person uses other software occasionally too - and that doesn't even consider later upgrade costs. With Ubuntu, you get very similar functionality to all of that built in at zero cost - for every user in your organization.
Maybe some of the Windows stuff is more feature packed - but with Ubuntu, every user gets all of it automatically. You never have to think "I only use photoshop every two weeks, is it really worth getting the license for *my* desktop?"
Buy supported hardware. Well supported printers work great on Linux, with setup being completely painless. (i.e. less painful than inserting a driver disk even). This Linux works poorly with unsupported hardware stuff was old in 2002 - you don't complain about Windows not supporting PowerPC processors. You can replace every single piece of poorly supported hardware for less than the cost of office - and the supported hardware is usually nicer anyway.
This is really embarrassing. It doesn't always happen, and it's not hard to fix, but when it does happen it means that someone's going to have to edit a dreaded config file. This was supposed to get fixed in Ubuntu 7.04, but it got put off until 7.10 because the X.org guys are working on it.
This is one specific card, not some widespread wifi issue. I agree that losing support for a piece of hardware is pretty lame. Still though, this is an issue with a specific piece of hardware that is now not supported in Ubuntu 7.04. There's no reason to mention it in a review / comparison at all, aside from perhaps "No OS supports all hardware. Make sure your hardware is supported before installing any OS."
If you actually implement a cost structure where you're charging based on the variable element of the user's usage that costs *you* money - i.e. upstream bandwidth charges, rewiring buildings to avoid internal bottlenecks, etc. If you set it up right you can completely eliminate the idea of treating heavy use as abuse.
Bandwidth isn't unlimited, there's only so much of it. Buying more capacity costs money.
Faced with more bandwidth demand than supply and no income from bandwidth charges, a University IT department (or an ISP selling "Unlimited Internet") will tend to look at their bandwidth usage and make calls like "60% of our bandwidth usage is coming from these 8 guys running servers. If we just ban servers we won't need to upgrade". So they implement that policy by blocking incoming TCP connections with a firewall and it buys them a couple months - at the horrific hidden cost of turning what was previously a general internet connection into a "consume only" connection. The first google servers were in Larry Paige's dorm room - who knows how college students might take advantage of having a general purpose internet connection...
Eventually, bandwidth usage will rise again and the IT department will be faced with the same situation: "60% of our bandwidth usage is coming from users doing P2P file sharing". But they can't fix this with a single firewall rule - so they're faced with the decision of buying active traffic shaping hardware or upgrading their bandwidth. Traffic shaping is cheaper than adding capacity, so they do that. They think they're only hurting "a couple students claiming to download Linux distros", but they're really killing ideas like blogtorrent. Basically, this is the same as the server block - it just costs more money in hardware.
If, instead, the school just charged for bandwidth - perhaps $0.25 for each gig after the first 5 gigs/month - this would play out completely differently. There would be an incentive for users to not outright waste bandwidth, and when it came time to chose between upgrading and degrading the school internet connection it would always be better to upgrade.
You have two problems, they can be handled separately.
First, you get RIAA letters. The appropriate response is a form letter saying that "Our school privacy policy prohibits us from releasing user information without a subpoena or court order" (obviously you'll want to verify that with a lawyer, but you shouldn't be sending out user information based on random letters). If you do get a legit subpoena or court order, send them the info if it's still available.
Second, you have excess bandwidth usage. This is really simple: Charge the students a reasonable fee for bandwidth overages - this will encourage users to conserve without unduly constraining people who actually are willing to pay for their bandwidth. It also has the advantage that as demand increases you automatically have the money to pay for upgrades.
So charge the students for their bandwidth usage at slightly above what it's costing. Use the extra money for upgrades.
Bandwidth is a resource that costs money - giving people "Unlimited" bandwidth and then degrading the internet connection to prevent "excess use" is absurd. Hell, if you give every student the first gig/month free most of them won't even notice the policy change.
Yea, and putting a second trailer on a big rig is a waste too because I have no trouble fitting my groceries in my sedan.
To be clearer, I should have said: If you test RHEL and Ubuntu, SLES probably works already. If you want a third tested "distro", I suggest Solaris - it's more likely to actually need a couple tweaks and then you get too support a whole new OS from a PR standpoint.
Again, if you're capable of devoting the time to port your application to a new platform then dealing with the packaging issues on Linux is pretty trivial time-wise. A commercial proprietary program is slightly harder than RealPlayer, because you can't rely on the various distros packaging it for you, but even that isn't that big a deal - Unreal Tournament 2004 used the Loki installer which works fine.
I'll admit that if you look at the list of different distributions on a site like Distrowatch, it can look like there's a daunting packaging problem. But... worrying about most of that is sort of like worrying about your software not working on Windows because someone built their own installer - i.e. irrelevant. Test on RHEL and Ubuntu and then consider fixing installer bugs that effect other distributions *if* they come up. After RHEL and Ubuntu, it's a toss up between supporting SLES next or just supporting Solaris next - the difference is about the same.