The more parts of the platform that are open, the harder it is to accomplish synergistic lock-in. Windows Client + Windows Server + Exchange + Microsoft Office sustains lock-in much better than Linux Client + Linux Server + Linux Backend + MS Office.
I've never paid more than $40 for a graphics card, and I've never found a game I couldn't play comfortably (high res, good frame rates).
Bullshit. Either that, or you only play games that are really old.
Take any game on a recent graphics card benchmark, and you'll discover that in order to get it running at a high resolution (high resolution today would be modes like 1920x1200) with decent framerates you really do need a card that costs more than $40. Even to run smoothly at a moderate resolution like 1400x1050 takes a $150 card.
In particular, it wouldn't take much to produce home-theater/video versions of cards: basically take a medium-range card with hardware MPEG-2/4 decompression, scaling, and deinterlace, and put some standard video outputs on it -- say HDMI, Component, and VGA (that way you could get to DVI easily using an HDMI adapter, and you can get to S-Video or Composite by combining the component signals).
That stuff all largely exists. It's just that gamer gear gets the marketing hype.
I have both Nvidia and ATI cards in Ubuntu desktops, and unless serious progress has been made in the past week the ATI drivers are *way* behind. They don't even provide 3D acceleration for a second logged in user.
I don't know how much experience you have developing real-world systems, but it can't be very much. Very, very many programs spend very, very little CPU time.
I'm not sure what your rant has to do with my post. Yes, there are large classes of system that aren't cpu-limited. I agree - there's no great need to re-write them to parallelize better since they're not CPU limited to begin with.
But... my point still holds. The vast majority of real word applications can take advantage of practical levels of parallelism - even if they're not "embarrassingly parallel". If it turns out that they are CPU limited in the future, most of them can be easily redesigned to take advantage of parallelism.
Nice parinoia, but the GNU GPL makes "closing the source" pretty hard.
Even without that, you'll find that simply being Free Software is pretty valuable. If it wasn't you'd be hearing a lot more about BSD/OS than about FreeBSD - but FreeBSD is orders of magnitude more popular because it's Free Software.
Solaris is an operating system because you can, say, run Abiword on it without installing the GNU System first. Linux, on the other hand, is not an operating system - you can't run Abiword on just Linux - you'd need a whole operating system like GNU/Linux to do that.
The problem here isn't that there is no Mac version of QuickBooks pro, it's that you've tied yourself down to one specific proprietary application. There are other options out there - it'll be far easier for you to switch to another option rather than waiting for a proprietary software developer to spend money for your benifit.
So... what you're saying is that they don't support *any* of the preferred open file formats except PDF. No OGG (Vorbis / Theora). No ODF. Their big claim to fame is supporting patent-encumbered pseudo standards like MP3 - and *maybe* that they support extremely widely accepted open standards like HTML and PNG - standards that even Microsoft largely supports.
Do you realize how rare shootings from people living in countries with strong gun control laws are? Your theory also sounds great, but the actual data we have strongly contradicts it.
Yea, you're right. They're stabbings instead.
A more interesting question is if the introduction of gun control laws decreases the frequency of violent crime overall, and there is evidence that it doesn't.
In the United States, gun control laws are unconstitutional. Given that, there would have to be some evidence that they're beneficial before I could consider supporting them.
Whatever is most compatible has a large advantage. Taking your IM example, none of AIM/MSN/YIM/GTalk has a massively overwhelming market share, so every even marginally technical user uses a multi-protocol client. If a network prevented multi-protocol clients from connecting (due to DRM or whatever), the technical users would stop using it and they'd tell their friends to switch to a more open network to talk to them.
As for "secure" movies and music, they're basically irrelevant with ThePirateBay offering better quality content for free.
"Principles", like "morals", are a sometimes-inaccurate heuristic for making judgment calls quickly. The "personal freedom" principle has developed for very good reason - we've seen again and again that when individuals are deprived of personal freedom there is a very high likelihood of unintended/unexpected negative consequences.
The founders of the United States thought that maintaining that principle was so important that they dedicated the first 10 amendments of our constitution to that single point. By your definition, they were zealots - and I'm damn happy they were.
Overall though, there would probably be a background of a shooting every couple of months when someone lost it momentarily and just happened to have a firearm handy.
Do you realize how rare shootings from people with CCPs are? Your theory sounds great, but the actual data we have strongly contradicts it.
People have a right to self defense. People have a right to use force in the defense of others. This is different from vigilantes, who use force not to defend someone (i.e. prevent harm) but for vengeance (i.e. to punish harm).
As for people having trouble identifying the mad gunman, that's unlikely to be relevant. It's possible, and that's a risk of carrying and using a gun, but it's more likely that more people with guns would simply have disrupted the mad gunman's ability to openly walk around shooting people uncontested.
If we can't have political discussions in the wake of politically significant events, I'm not sure what good democracy is.
It's extremely easy to predict that both sides of the gun control issue are going to get pushed here, and giving people crap for "pushing their agenda" is a waste of time. If gun control laws are intended to stop incedents like this, this is an excellent time to look at the question of how good a job they are doing.
If the only goal of gun control laws is to prevent this sort of shooting (I don't know if that's true), and if gun control laws actually prevented people from defending themselves with guns, then gun control laws should be weakened. If gun control laws are expected to reduce shootings of this kind (obviously they are), and if there is an obvious weakness in gun control laws that allowed this shooting to occur without some other benifit (It's possible), then that weakness should be addressed.
I'm saying stop pushing your agenda because there's no way to know one way or the other
Bullshit. It's not impossible to know the answer to simple questions like "would guns in the crowd have prevented deaths?".
Nor is this an inappropriate place to discuss the issue of gun laws. The primary political reaction from this incident is going to be a push for further restrictions on gun ownership - if that's not obvious to you, you're oblivious. Politcally, that is the issue that is relevant to this news story.
Thirty three deaths on a school campus is a horrible tragedy - it would be nice to implement policy that reduces the threat of such incidents in the future. Not only is is possible to figure out what policy choices are available (increasing or reducing gun restrictions are obvious choices), it's possible to make predictions about what the effects of those policies would be.
You go right ahead and load a DRM protected song from iTunes onto your Sandisk Sansa MP3 player using approved OS X / iTunes functionality. Once you've done that, you can make the claim that Apple doesn't screw their customers. Apple isn't bad overall, but their just as much a villain as Microsoft in this DRM thing.
And besides, some tasks are inherently sequential like a finite state machine for example.
A finite state machine is an algorithm not a task. You can achieve parallelism by using a different algorithm, or by using a splitting heuristic that divides the work among multiple threads running the non-parallel algorithm.
That'd be breaking the license just as much as hacking your license data to say "licensed for 4 processors".
I bet that for the price that you're paying for that software, you could have some undergrad rewrite the application slightly less efficiently and then buy another 8 Opteron boxes to run the new slower software on in a cluster.
knows that many apps simply do not lend themselves to multithreading and parallelism.
This is a myth, propagated by lazy developers and cheap end users.
There are some classes of computing problems that can't be parallelized, but very few of those problems are the applications that we want to run faster on modern computers.
The only application that shows up on benchmark sites that might not be easily parallelizable is file compression (i.e. "WinRAR"), and if that ever needs to be parallelized a small algorithmic change (compress blocks separately) will handle it.
As for GUI apps, games, media encoding, etc - that stuff parallelizes no problem.
There are some program types where it is (provably) impossible to parallelize them. They're much more rare in real world applications than people seem to think.
All of the examples you give can be parallelized - at least for a small number (say, up to 16) processors.
XML Parser:
Split the file into a number of equal sized pieces equal to the number of processors you have.
In each thread, read in your chunk of the file and build whatever portion of the tree you can make out.
Merge the trees together, including resolving tags that got cut in half by the split and stuff. (Since the threads share memory, this should involve parsing at most n-1 split tags and doing n*log(n) pointer operations to link up the subtrees)
Video Games:
The only reason that video games aren't massively parallelized is that context switches are an utter waste of CPU time on a single-CPU system - and gamers are twitchy about every last FPS. Once a good chunk of gamers have 4-core systems, it video game makers will use multithreaded physics and rendering engines because it will be a good trade to sacrifice 1-core performance for 4-core performance.
The Geek looks at structure and ignores culture. There is a tradition of winner-take-all, a dislike of multiple choice questions that begins in grade school and a profound distrust of coalitions that gives very small minorities inordinate power.
That's a cop-out, and not a very good one at that. There is absolutely cultural inertia that will resist any change, but I see no reason why Approval Voting is any less "winner take all" or any more "multiple choice" than the current voting method is. Further, I don't see how either method that I suggested encourages "coalitions that gives very small minorities inordinate power". You still need to get the plurality of voters to vote for you before you win.
The two party system in the United States definitely doesn't act to prevent coalitions that give small minorities inordinate power. All it does is makes sure that those coalitions are consistent, so that corporations have to make less campaign contributions to make sure the winner likes them.
The United States system prevents there being more than two serious parties. European countries tend to use proportional representation to solve the problem. In the United States, that doesn't work because our congresmen represent geographic areas - but the problem could be signficiantly reduced if we used a voting system like http://en.wikipedia.org/wiki/Approval_Voting or http://en.wikipedia.org/wiki/Condorcet_method that didn't severely punish third party votes strategically.
Another potential tactic would be to promote subparties. The final elections are solidly locked down to Democrat or Republican, but the actual Democrat/Republican primaries are much more open - an organized "branch" of a major party could probably get their candidate nominated with an effort that is possible to achieve.
It seems that way from my point of view also, but apparently business is thriving as IP law becomes more powerful.
Really? Is "business thriving" because of copyright and patent law, or in spite of copyright and patent law?
What do you mean by "business thriving"? Are you talking about more people being able to make a living as IP lawyers? Is that really an economic or social advantage?
Economics is somewhat complicated, but the profit of a single person or company - take alone - is not a good indicator of an overall benefit to an enonomy.
Slashdot Mirror
The more parts of the platform that are open, the harder it is to accomplish synergistic lock-in. Windows Client + Windows Server + Exchange + Microsoft Office sustains lock-in much better than Linux Client + Linux Server + Linux Backend + MS Office.
Bullshit. Either that, or you only play games that are really old.
Take any game on a recent graphics card benchmark, and you'll discover that in order to get it running at a high resolution (high resolution today would be modes like 1920x1200) with decent framerates you really do need a card that costs more than $40. Even to run smoothly at a moderate resolution like 1400x1050 takes a $150 card.
That stuff all largely exists. It's just that gamer gear gets the marketing hype.
I have both Nvidia and ATI cards in Ubuntu desktops, and unless serious progress has been made in the past week the ATI drivers are *way* behind. They don't even provide 3D acceleration for a second logged in user.
I'm not sure what your rant has to do with my post. Yes, there are large classes of system that aren't cpu-limited. I agree - there's no great need to re-write them to parallelize better since they're not CPU limited to begin with.
But... my point still holds. The vast majority of real word applications can take advantage of practical levels of parallelism - even if they're not "embarrassingly parallel". If it turns out that they are CPU limited in the future, most of them can be easily redesigned to take advantage of parallelism.
Nice parinoia, but the GNU GPL makes "closing the source" pretty hard.
Even without that, you'll find that simply being Free Software is pretty valuable. If it wasn't you'd be hearing a lot more about BSD/OS than about FreeBSD - but FreeBSD is orders of magnitude more popular because it's Free Software.
Solaris is an operating system because you can, say, run Abiword on it without installing the GNU System first. Linux, on the other hand, is not an operating system - you can't run Abiword on just Linux - you'd need a whole operating system like GNU/Linux to do that.
The problem here isn't that there is no Mac version of QuickBooks pro, it's that you've tied yourself down to one specific proprietary application. There are other options out there - it'll be far easier for you to switch to another option rather than waiting for a proprietary software developer to spend money for your benifit.
So... what you're saying is that they don't support *any* of the preferred open file formats except PDF. No OGG (Vorbis / Theora). No ODF. Their big claim to fame is supporting patent-encumbered pseudo standards like MP3 - and *maybe* that they support extremely widely accepted open standards like HTML and PNG - standards that even Microsoft largely supports.
Yea, you're right. They're stabbings instead.
A more interesting question is if the introduction of gun control laws decreases the frequency of violent crime overall, and there is evidence that it doesn't.
In the United States, gun control laws are unconstitutional. Given that, there would have to be some evidence that they're beneficial before I could consider supporting them.
Whatever is most compatible has a large advantage. Taking your IM example, none of AIM/MSN/YIM/GTalk has a massively overwhelming market share, so every even marginally technical user uses a multi-protocol client. If a network prevented multi-protocol clients from connecting (due to DRM or whatever), the technical users would stop using it and they'd tell their friends to switch to a more open network to talk to them.
As for "secure" movies and music, they're basically irrelevant with ThePirateBay offering better quality content for free.
This is possible, but it becomes less likely as the computer that you're working on gets more cores.
"Principles", like "morals", are a sometimes-inaccurate heuristic for making judgment calls quickly. The "personal freedom" principle has developed for very good reason - we've seen again and again that when individuals are deprived of personal freedom there is a very high likelihood of unintended/unexpected negative consequences.
The founders of the United States thought that maintaining that principle was so important that they dedicated the first 10 amendments of our constitution to that single point. By your definition, they were zealots - and I'm damn happy they were.
Do you realize how rare shootings from people with CCPs are? Your theory sounds great, but the actual data we have strongly contradicts it.
People have a right to self defense. People have a right to use force in the defense of others. This is different from vigilantes, who use force not to defend someone (i.e. prevent harm) but for vengeance (i.e. to punish harm).
As for people having trouble identifying the mad gunman, that's unlikely to be relevant. It's possible, and that's a risk of carrying and using a gun, but it's more likely that more people with guns would simply have disrupted the mad gunman's ability to openly walk around shooting people uncontested.
If we can't have political discussions in the wake of politically significant events, I'm not sure what good democracy is.
It's extremely easy to predict that both sides of the gun control issue are going to get pushed here, and giving people crap for "pushing their agenda" is a waste of time. If gun control laws are intended to stop incedents like this, this is an excellent time to look at the question of how good a job they are doing.
If the only goal of gun control laws is to prevent this sort of shooting (I don't know if that's true), and if gun control laws actually prevented people from defending themselves with guns, then gun control laws should be weakened. If gun control laws are expected to reduce shootings of this kind (obviously they are), and if there is an obvious weakness in gun control laws that allowed this shooting to occur without some other benifit (It's possible), then that weakness should be addressed.
Bullshit. It's not impossible to know the answer to simple questions like "would guns in the crowd have prevented deaths?".
Nor is this an inappropriate place to discuss the issue of gun laws. The primary political reaction from this incident is going to be a push for further restrictions on gun ownership - if that's not obvious to you, you're oblivious. Politcally, that is the issue that is relevant to this news story.
Thirty three deaths on a school campus is a horrible tragedy - it would be nice to implement policy that reduces the threat of such incidents in the future. Not only is is possible to figure out what policy choices are available (increasing or reducing gun restrictions are obvious choices), it's possible to make predictions about what the effects of those policies would be.
You go right ahead and load a DRM protected song from iTunes onto your Sandisk Sansa MP3 player using approved OS X / iTunes functionality. Once you've done that, you can make the claim that Apple doesn't screw their customers. Apple isn't bad overall, but their just as much a villain as Microsoft in this DRM thing.
A finite state machine is an algorithm not a task. You can achieve parallelism by using a different algorithm, or by using a splitting heuristic that divides the work among multiple threads running the non-parallel algorithm.
That'd be breaking the license just as much as hacking your license data to say "licensed for 4 processors".
I bet that for the price that you're paying for that software, you could have some undergrad rewrite the application slightly less efficiently and then buy another 8 Opteron boxes to run the new slower software on in a cluster.
This is a myth, propagated by lazy developers and cheap end users.
There are some classes of computing problems that can't be parallelized, but very few of those problems are the applications that we want to run faster on modern computers.
The only application that shows up on benchmark sites that might not be easily parallelizable is file compression (i.e. "WinRAR"), and if that ever needs to be parallelized a small algorithmic change (compress blocks separately) will handle it.
As for GUI apps, games, media encoding, etc - that stuff parallelizes no problem.
There are some program types where it is (provably) impossible to parallelize them. They're much more rare in real world applications than people seem to think.
All of the examples you give can be parallelized - at least for a small number (say, up to 16) processors.
XML Parser:
Video Games:
The only reason that video games aren't massively parallelized is that context switches are an utter waste of CPU time on a single-CPU system - and gamers are twitchy about every last FPS. Once a good chunk of gamers have 4-core systems, it video game makers will use multithreaded physics and rendering engines because it will be a good trade to sacrifice 1-core performance for 4-core performance.
That's a cop-out, and not a very good one at that. There is absolutely cultural inertia that will resist any change, but I see no reason why Approval Voting is any less "winner take all" or any more "multiple choice" than the current voting method is. Further, I don't see how either method that I suggested encourages "coalitions that gives very small minorities inordinate power". You still need to get the plurality of voters to vote for you before you win.
The two party system in the United States definitely doesn't act to prevent coalitions that give small minorities inordinate power. All it does is makes sure that those coalitions are consistent, so that corporations have to make less campaign contributions to make sure the winner likes them.
The United States system prevents there being more than two serious parties. European countries tend to use proportional representation to solve the problem. In the United States, that doesn't work because our congresmen represent geographic areas - but the problem could be signficiantly reduced if we used a voting system like http://en.wikipedia.org/wiki/Approval_Voting or http://en.wikipedia.org/wiki/Condorcet_method that didn't severely punish third party votes strategically.
Another potential tactic would be to promote subparties. The final elections are solidly locked down to Democrat or Republican, but the actual Democrat/Republican primaries are much more open - an organized "branch" of a major party could probably get their candidate nominated with an effort that is possible to achieve.
Really? Is "business thriving" because of copyright and patent law, or in spite of copyright and patent law?
What do you mean by "business thriving"? Are you talking about more people being able to make a living as IP lawyers? Is that really an economic or social advantage?
Economics is somewhat complicated, but the profit of a single person or company - take alone - is not a good indicator of an overall benefit to an enonomy.